Soccer predictions: actual data vs predicted data

Slate 20261108 · 10 match(es)

How to read this report. Each match shows the model's headline prediction, the stage-by-stage lambda (goal-rate) build, both teams' standing/form inputs, and every starter's projected line next to their real season baseline. A moderate / large deviation badge flags a projection that strays far from the player's season form — a quick way to spot when the inputs look right but a projection does not. When the real result is known an actual-vs-predicted banner appears.

All matches overview

Match	Score	Winner	H / D / A	O2.5	BTTS	Conf	λ H/A	Actual	Result
Man Utd vs Aston Villa Premier League	1-0	Man Utd	51.7% / 24.2% / 24.1%	52.6%	54.1%	0.68	1.70 / 1.08	-	pending
Atlético Madrid vs Barcelona LaLiga	1-2	Barcelona	27.6% / 21.0% / 51.4%	70.4%	68.8%	0.69	1.52 / 2.12	-	pending
Elche vs Real Betis LaLiga	2-1	Elche	41.4% / 24.5% / 34.2%	57.3%	60.0%	0.65	1.57 / 1.41	-	pending
Ipswich vs Bournemouth Premier League	1-1	Draw	36.3% / 25.6% / 38.1%	52.6%	56.5%	0.64	1.37 / 1.41	-	pending
Brighton vs Brentford Premier League	2-1	Brighton	41.9% / 23.8% / 34.3%	60.3%	62.3%	0.66	1.65 / 1.47	-	pending
Sunderland vs Chelsea Premier League	1-2	Chelsea	26.4% / 23.8% / 49.8%	56.2%	57.6%	0.68	1.20 / 1.73	-	pending
Villarreal vs Getafe LaLiga	2-0	Villarreal	67.3% / 19.3% / 13.5%	57.3%	50.2%	0.73	2.14 / 0.84	-	pending
Celta vs Levante LaLiga	2-1	Celta	50.1% / 23.5% / 26.4%	57.3%	58.5%	0.68	1.76 / 1.22	-	pending
Crystal Palace vs Liverpool Premier League	1-1	Draw	35.8% / 25.6% / 38.6%	52.6%	56.5%	0.65	1.36 / 1.42	-	pending
Real Sociedad vs Rayo Vallecano LaLiga	2-1	Real Sociedad	56.9% / 22.2% / 20.8%	57.3%	56.1%	0.7	1.91 / 1.07	-	pending

Match detail

Man Utd vs Aston Villa

Premier League · 20261108 · 4-2-3-1 vs 4-2-3-1

Predicted 1-0Winner: Man UtdWinning H 1-0 (10.4%)Winning A 0-1 (6.6%)Confidence 0.68λ home/away 1.70 / 1.08Most-likely P 10.4%

Actual vs predicted: result not available yet (upcoming fixture).

XGBoost Reasoning & FBref Dataset Tracing

Antagonist + XGBoost match analysis for Man Utd vs Aston Villa. Player statistics regressed using position-specific XGBoost models (G, D, M, F) with 2026-2027 FBref priors. Micro-duels resolved across 4-2-3-1 vs 4-2-3-1 (duel lambdas: Home 1.573, Away 1.085). Final calibrated Poisson lambdas: Home 1.698, Away 1.083 (most likely score 1-0).

Home Starters 11/11 XGBoost (9 FBref) Away Starters 11/11 XGBoost (9 FBref) Bench Modeled H: 7 / A: 7 Duel λ (H/A) 1.57 / 1.08 Final λ (H/A) 1.70 / 1.08

Modeling Pipeline Stages (5 layers)

1. XGBoost player regressions for volume & event rates with market-value quality scaling
2. Multi-source prior blending (FBref 2026-2027 tables + Sofascore season history)
3. Antagonist formation duel matrix resolving 1v1 and zonal matchups
4. Team-level prior calibration (FBref possession and attack/defense efficiencies)
5. Bivariate Poisson matrix with Dixon-Coles low-score correction

Standing & form context

Metric	Man Utd	Aston Villa
League position	17.00	9.00
Points per game	1.01	1.34
Attack index	0.96	0.88
Defense (concede) index	0.98	1.15
Form multiplier	0.98	1.02
Standing λ multiplier	0.97	1.01

Lambda decomposition (how the goal rates were built)

Stage	Man Utd λ	Aston Villa λ
Pre-competitive-floor	1.779	1.022
Pre-calibration (Micro-duels)	1.573	1.085
Final λ	1.698	1.083

competition calibration scale 0.993competition observed total goals 2.781quality gap applied Truemismatch score pressure applied Falsecompetitive underdog floor applied Falsestage discount 1.0

Scoreline distribution & markets

Top 5 scorelines	P
1-1	11.5%
1-0	10.4%
2-1	9.7%
2-0	8.9%
0-1	6.6%

Home win	51.7%
Draw	24.2%
Away win	24.1%
Over 2.5	52.6%
BTTS yes	54.1%

Pregame fair line — favorite Man Utd (51.7%), fair total 2.78, fair spread 0.61

Projected players vs season baseline

Proj = projected for this match · Season = this season's per-game average from match data · deviation badge compares projected xG to the player's season xG/game.

Man Utd — projected line vs season baseline (4-2-3-1)

Player Pos Min Proj G Proj xG Season xG/g Proj Sh Season Sh/g Proj Rtg Season Rtg Rec Model / FBref Prior

Senne Lammens G 90 0 0.00 - n/a 0.0 6.9 60.0 33 FBref Prior G_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost G_service raw output over 3 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0086, quality 0.770)	`market_value_eur=38000000` `quality_factor=0.7699` `boost_factor=1.0086` `reference_quality=0.505` `history_matches=41` `history_shrink=0.1087`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0094)	`blend_weight=0.0094` `history_matches=41` `season_overlap_discount=1.0` `per90_to_match_factor=1.0` `prior_reliability=0.1333` `competition=Premier League`	-
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` `duel_context={'opponent_faced': 'Nicolas Jackson, Johan Manzambi', 'matchup_zone': 'Goalkeeper shot-stopping context', 'opponent_lambda': 1.085}`	-
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 1 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=1`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.70 (squad xG sum 1.2053, xG scale 1.4087, xA scale 1.4087, team xA 1.16 = 0.68 x lambda)	`scoreline_reset=true` `team_lambda=1.698` `squad_xg_sum=1.2053` `squad_xa_sum=0.8253` `team_expected_assists=1.1626` `team_xa_to_xg_ratio=0.6847`	-
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 7 by projected rating, fifa_overall 82.0)	`match_squad_rank=7` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 1 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=1` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.0` `scorer_weight=0.0` `squad_weight_total=1.7229`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.0` `provider_weight=0.0` `squad_weight_total=1.1277`	-

Matthijs de Ligt D 90 0 0.19 - n/a 1 0.0 6.9 - 8 XGBoost D_service

Step-by-step (9 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost D_service raw output over 7 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0151, quality 0.686)	`market_value_eur=26000000` `quality_factor=0.6858` `boost_factor=1.0151` `reference_quality=0.505` `history_matches=13` `history_shrink=0.2778`	-
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` `duel_context={'opponent_faced': 'Nicolas Jackson', 'matchup_zone': 'Striker vs Two Centre Backs — 1v2 Aerial/Physical Battle'}`	-
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 1 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=1`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.70 (squad xG sum 1.2053, xG scale 1.4087, xA scale 1.4087, team xA 1.16 = 0.68 x lambda)	`scoreline_reset=true` `team_lambda=1.698` `squad_xg_sum=1.2053` `squad_xa_sum=0.8253` `team_expected_assists=1.1626` `team_xa_to_xg_ratio=0.6847`	expected_goals: 0.1327 -> 0.1869 expected_assists: 0.0644 -> 0.0907
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 9 by projected rating, fifa_overall 82.0)	`match_squad_rank=9` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 1 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=1` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.1705` `goal_rate=0.1252` `scorer_weight=0.1792`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.0867` `assist_rate=0.0` `provider_weight=0.0726`	-

Harry Maguire D 90 0 0.05 0.04 ok 1 0.6 6.9 55.4 30 FBref Prior D_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost D_service raw output over 7 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0000, quality 0.506)	`market_value_eur=8600000` `quality_factor=0.5059` `boost_factor=1.0` `reference_quality=0.505` `history_matches=32` `history_shrink=0.1351`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0117)	`blend_weight=0.0117` `history_matches=32` `season_overlap_discount=1.0` `per90_to_match_factor=0.9944` `prior_reliability=0.1333` `competition=Premier League`	totalShots: 0.5777 -> 0.5884 wonTackle: 0.7832 -> 0.774 totalCross: 0.3087 -> 0.3167
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` `duel_context={'opponent_faced': 'Nicolas Jackson', 'matchup_zone': 'Striker vs Two Centre Backs — 1v2 Aerial/Physical Battle'}`	-
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 1 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=1`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.70 (squad xG sum 1.2053, xG scale 1.4087, xA scale 1.4087, team xA 1.16 = 0.68 x lambda)	`scoreline_reset=true` `team_lambda=1.698` `squad_xg_sum=1.2053` `squad_xa_sum=0.8253` `team_expected_assists=1.1626` `team_xa_to_xg_ratio=0.6847`	expected_goals: 0.0339 -> 0.0478 expected_assists: 0.0385 -> 0.0542
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 4 by projected rating, fifa_overall 82.0)	`match_squad_rank=4` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 1 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=1` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.0467` `goal_rate=0.0027` `scorer_weight=0.0382`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.0528` `assist_rate=0.02` `provider_weight=0.0434`	-

Luke Shaw D 85 0 0.02 0.02 ok 1 0.4 6.8 56.0 29 FBref Prior D_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost D_service raw output over 7 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 0.9996, quality 0.493)	`market_value_eur=7500000` `quality_factor=0.4927` `boost_factor=0.9996` `reference_quality=0.505` `history_matches=43` `history_shrink=0.1042`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0081)	`blend_weight=0.0081` `history_matches=43` `season_overlap_discount=1.0` `per90_to_match_factor=0.9459` `prior_reliability=0.12` `competition=Premier League`	totalCross: 1.5215 -> 1.539 fouls: 0.9998 -> 0.9917
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` `duel_context={'opponent_faced': 'Lamare Bogarde', 'matchup_zone': 'Right Wing vs Left Back — Flank Battle'}`	-
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 1 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=1`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.70 (squad xG sum 1.2053, xG scale 1.4087, xA scale 1.4087, team xA 1.16 = 0.68 x lambda)	`scoreline_reset=true` `team_lambda=1.698` `squad_xg_sum=1.2053` `squad_xa_sum=0.8253` `team_expected_assists=1.1626` `team_xa_to_xg_ratio=0.6847`	expected_goals: 0.0175 -> 0.0247 expected_assists: 0.0837 -> 0.1179
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 11 by projected rating, fifa_overall 79.0)	`match_squad_rank=11` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 1 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=1` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.0244` `goal_rate=0.059` `scorer_weight=0.0309`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.1112` `assist_rate=0.1082` `provider_weight=0.1395`	-

Lisandro Martínez D 80 0 0.03 0.01 133% 1 0.5 6.9 53.3 25 FBref Prior D_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost D_service raw output over 7 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0151, quality 0.817)	`market_value_eur=48000000` `quality_factor=0.8168` `boost_factor=1.0151` `reference_quality=0.505` `history_matches=26` `history_shrink=0.1613`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0070)	`blend_weight=0.007` `history_matches=26` `season_overlap_discount=1.0` `per90_to_match_factor=0.8864` `prior_reliability=0.0667` `competition=Premier League`	fouls: 1.0243 -> 1.0171 interceptionWon: 0.9449 -> 0.9383 wonTackle: 0.7598 -> 0.7545
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` `duel_context={'opponent_faced': 'Johan Manzambi', 'matchup_zone': 'Left Wing vs Right Back — Flank Battle'}`	-
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 1 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=1`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.70 (squad xG sum 1.2053, xG scale 1.4087, xA scale 1.4087, team xA 1.16 = 0.68 x lambda)	`scoreline_reset=true` `team_lambda=1.698` `squad_xg_sum=1.2053` `squad_xa_sum=0.8253` `team_expected_assists=1.1626` `team_xa_to_xg_ratio=0.6847`	expected_goals: 0.0248 -> 0.0349 expected_assists: 0.0271 -> 0.0382
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 8 by projected rating, fifa_overall 82.0)	`match_squad_rank=8` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 1 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=1` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.0343` `goal_rate=0.0` `scorer_weight=0.0279`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.0375` `assist_rate=0.0` `provider_weight=0.0306`	-

Bruno Fernandes M 90 0 0.36 0.21 68% 3 2.0 7.2 60.8 32 FBref Prior M_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost M_service raw output over 8 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0071, quality 0.732)	`market_value_eur=32000000` `quality_factor=0.732` `boost_factor=1.0071` `reference_quality=0.505` `history_matches=43` `history_shrink=0.1042`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0090)	`blend_weight=0.009` `history_matches=43` `season_overlap_discount=1.0` `per90_to_match_factor=0.9959` `prior_reliability=0.1333` `competition=Premier League`	totalShots: 2.0306 -> 2.0617 onTargetScoringAttempt: 0.6521 -> 0.6642 goals: 0.2543 -> 0.2655 wasFouled: 0.8104 -> 0.8031 wonTackle: 1.1007 -> 1.0953
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` duel_context={'opponent_faced': 'Victor Lindelöf', 'matchup_zone': 'Left Wing vs Right Back — Flank Battle', 'duel_win_rate': 0.559, 'pressing_faced': 'LOW', 'numerical_situation': '1v1', 'xg_multiplier': 1.1164, 'skill_edge': 0.1112, 'height_edge': -0.032, 'attacker_skill': 0.732, 'defender_skill': 0.4849, 'attacker_role': 'LW', 'attacker_mv_price_index': 1.0, 'defender_adjusted_parts': [{'player': 'Victor Lindelöf', 'market_value': 4800000, 'role': 'RB', 'price_index': 0.7, 'adjusted_quality': 0.4849}], 'xg_override': 0.2522}	aerialWon: 0.208 -> 1.0 totalShots: 2.0617 -> 2.3017 shotOffTarget: 1.3784 -> 1.5388 onTargetScoringAttempt: 0.6642 -> 0.7415 bigChanceMissed: 0.2721 -> 0.3038 goals: 0.2655 -> 0.2964 expectedGoals: 0.2259 -> 0.2522 expectedGoalsOnTarget: 0.1889 -> 0.2109
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 1 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=1`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.70 (squad xG sum 1.2053, xG scale 1.4087, xA scale 1.4087, team xA 1.16 = 0.68 x lambda)	`scoreline_reset=true` `team_lambda=1.698` `squad_xg_sum=1.2053` `squad_xa_sum=0.8253` `team_expected_assists=1.1626` `team_xa_to_xg_ratio=0.6847`	expected_goals: 0.2522 -> 0.3553 expected_assists: 0.1977 -> 0.2786
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 1 by projected rating, fifa_overall 89.0)	`match_squad_rank=1` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 1 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=1` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.299` `goal_rate=0.2964` `scorer_weight=0.4073`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=1` `assist_probability=0.2432` `assist_rate=0.2146` `provider_weight=0.2976`	-

Youri Tielemans M 76 0 0.08 0.04 90% 1 0.7 7.1 55.7 26 FBref Prior M_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost M_service raw output over 8 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0115, quality 0.796)	`market_value_eur=43000000` `quality_factor=0.7957` `boost_factor=1.0115` `reference_quality=0.505` `history_matches=33` `history_shrink=0.1316`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0097)	`blend_weight=0.0097` `history_matches=33` `season_overlap_discount=1.0` `per90_to_match_factor=0.8427` `prior_reliability=0.1133` `competition=Premier League`	wasFouled: 1.9305 -> 1.9406 totalShots: 0.9137 -> 0.9237 totalCross: 1.4576 -> 1.4483 onTargetScoringAttempt: 0.2856 -> 0.2922 wonTackle: 1.1708 -> 1.1643
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` duel_context={'opponent_faced': 'Ian Maatsen', 'matchup_zone': 'Right Wing vs Left Back — Flank Battle', 'duel_win_rate': 0.468, 'pressing_faced': 'HIGH', 'numerical_situation': '1v1', 'xg_multiplier': 0.9823, 'skill_edge': -0.008, 'height_edge': -0.004, 'attacker_skill': 0.7957, 'defender_skill': 0.8135, 'attacker_role': 'RW', 'attacker_mv_price_index': 1.0, 'defender_adjusted_parts': [{'player': 'Ian Maatsen', 'market_value': 33000000, 'role': 'LB', 'price_index': 0.7, 'adjusted_quality': 0.8135}], 'xg_override': 0.0592}	totalShots: 0.9237 -> 0.9074 aerialWon: 1.0115 -> 1.0 shotOffTarget: 0.6281 -> 0.617 onTargetScoringAttempt: 0.2922 -> 0.287
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 1 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=1`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.70 (squad xG sum 1.2053, xG scale 1.4087, xA scale 1.4087, team xA 1.16 = 0.68 x lambda)	`scoreline_reset=true` `team_lambda=1.698` `squad_xg_sum=1.2053` `squad_xa_sum=0.8253` `team_expected_assists=1.1626` `team_xa_to_xg_ratio=0.6847`	expected_goals: 0.0592 -> 0.0834 expected_assists: 0.1124 -> 0.1583
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 3 by projected rating, fifa_overall 85.0)	`match_squad_rank=3` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 1 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=1` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.08` `goal_rate=0.0586` `scorer_weight=0.0829`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.1464` `assist_rate=0.0831` `provider_weight=0.1297`	-

Carlos Baleba M 72 0 0.08 0.03 222% 1 0.6 6.9 46.5 26 XGBoost M_service

Step-by-step (9 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost M_service raw output over 8 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0133, quality 0.859)	`market_value_eur=63000000` `quality_factor=0.8589` `boost_factor=1.0133` `reference_quality=0.505` `history_matches=35` `history_shrink=0.125`	-
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` duel_context={'opponent_faced': 'Boubacar Kamara, João Gomes', 'matchup_zone': 'Attacking Mid vs Double Pivot — Tight Space Battle', 'duel_win_rate': 0.256, 'pressing_faced': 'HIGH', 'numerical_situation': '1v2', 'xg_multiplier': 1.024, 'skill_edge': 0.0203, 'height_edge': -0.004, 'attacker_skill': 0.865, 'defender_skill': 0.8198, 'attacker_role': 'CAM', 'attacker_mv_price_index': 0.95, 'defender_adjusted_parts': [{'player': 'Boubacar Kamara', 'market_value': 49000000, 'role': 'CDM', 'price_index': 0.85, 'adjusted_quality': 0.847}, {'player': 'João Gomes', 'market_value': 36000000, 'role': 'CDM', 'price_index': 0.85, 'adjusted_quality': 0.7926}], 'xg_override': 0.0591}	aerialWon: 1.2996 -> 1.0 totalShots: 0.9085 -> 0.9303 shotOffTarget: 0.6997 -> 0.7165
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 1 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=1`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.70 (squad xG sum 1.2053, xG scale 1.4087, xA scale 1.4087, team xA 1.16 = 0.68 x lambda)	`scoreline_reset=true` `team_lambda=1.698` `squad_xg_sum=1.2053` `squad_xa_sum=0.8253` `team_expected_assists=1.1626` `team_xa_to_xg_ratio=0.6847`	expected_goals: 0.0591 -> 0.0833 expected_assists: 0.0362 -> 0.051
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 10 by projected rating, fifa_overall 80.0)	`match_squad_rank=10` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 1 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=1` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.0799` `goal_rate=0.0622` `scorer_weight=0.0857`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.0497` `assist_rate=0.0571` `provider_weight=0.0634`	-

Kobbie Mainoo M 87 0 0.07 0.02 202% 1 0.6 7.1 58.9 29 FBref Prior M_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost M_service raw output over 8 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0140, quality 0.879)	`market_value_eur=77000000` `quality_factor=0.8792` `boost_factor=1.014` `reference_quality=0.505` `history_matches=35` `history_shrink=0.125`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0060)	`blend_weight=0.006` `history_matches=35` `season_overlap_discount=1.0` `per90_to_match_factor=0.966` `prior_reliability=0.0733` `competition=Premier League`	totalShots: 0.7147 -> 0.7306 fouls: 0.9276 -> 0.9221
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` duel_context={'opponent_faced': 'Boubacar Kamara', 'matchup_zone': 'CDM vs CDM', 'duel_win_rate': 0.439, 'pressing_faced': 'ULTRA_HIGH', 'numerical_situation': '1v1', 'xg_multiplier': 1.028, 'skill_edge': 0.0191, 'height_edge': 0.0, 'attacker_skill': 0.8893, 'defender_skill': 0.847, 'attacker_role': 'CDM', 'attacker_mv_price_index': 0.85, 'defender_adjusted_parts': [{'player': 'Boubacar Kamara', 'market_value': 49000000, 'role': 'CDM', 'price_index': 0.85, 'adjusted_quality': 0.847}], 'xg_override': 0.0525}	aerialWon: 0.3665 -> 1.1 totalShots: 0.7306 -> 0.7511 shotOffTarget: 0.535 -> 0.55 onTargetScoringAttempt: 0.1786 -> 0.1836
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 1 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=1`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.70 (squad xG sum 1.2053, xG scale 1.4087, xA scale 1.4087, team xA 1.16 = 0.68 x lambda)	`scoreline_reset=true` `team_lambda=1.698` `squad_xg_sum=1.2053` `squad_xa_sum=0.8253` `team_expected_assists=1.1626` `team_xa_to_xg_ratio=0.6847`	expected_goals: 0.0525 -> 0.074 expected_assists: 0.0748 -> 0.1054
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 2 by projected rating, fifa_overall 81.0)	`match_squad_rank=2` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 1 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=1` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.0713` `goal_rate=0.0512` `scorer_weight=0.0731`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.1` `assist_rate=0.0675` `provider_weight=0.0992`	-

Andrey Santos M 78 0 0.08 0.04 92% 1 0.2 6.9 41.8 33 FBref Prior M_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost M_service raw output over 8 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0117, quality 0.800)	`market_value_eur=44000000` `quality_factor=0.8002` `boost_factor=1.0117` `reference_quality=0.505` `history_matches=33` `history_shrink=0.1316`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0051)	`blend_weight=0.0051` `history_matches=33` `season_overlap_discount=1.0` `per90_to_match_factor=0.8611` `prior_reliability=0.06` `competition=Premier League`	interceptionWon: 1.2386 -> 1.2322
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` duel_context={'opponent_faced': 'Boubacar Kamara', 'matchup_zone': 'CDM vs CDM', 'duel_win_rate': 0.396, 'pressing_faced': 'ULTRA_HIGH', 'numerical_situation': '1v1', 'xg_multiplier': 0.9653, 'skill_edge': -0.0076, 'height_edge': -0.016, 'attacker_skill': 0.8301, 'defender_skill': 0.847, 'attacker_role': 'CDM', 'attacker_mv_price_index': 0.85, 'defender_adjusted_parts': [{'player': 'Boubacar Kamara', 'market_value': 49000000, 'role': 'CDM', 'price_index': 0.85, 'adjusted_quality': 0.847}], 'xg_override': 0.0545}	aerialWon: 0.9972 -> 1.1 totalShots: 0.5928 -> 0.5722 shotOffTarget: 0.4315 -> 0.4165 onTargetScoringAttempt: 0.1613 -> 0.1557
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 1 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=1`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.70 (squad xG sum 1.2053, xG scale 1.4087, xA scale 1.4087, team xA 1.16 = 0.68 x lambda)	`scoreline_reset=true` `team_lambda=1.698` `squad_xg_sum=1.2053` `squad_xa_sum=0.8253` `team_expected_assists=1.1626` `team_xa_to_xg_ratio=0.6847`	expected_goals: 0.0545 -> 0.0768 expected_assists: 0.037 -> 0.0521
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 6 by projected rating, fifa_overall 80.0)	`match_squad_rank=6` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 1 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=1` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.0739` `goal_rate=0.0515` `scorer_weight=0.0746`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.0508` `assist_rate=0.0539` `provider_weight=0.0622`	-

Bryan Mbeumo F 87 1 0.55 0.39 41% 3 2.1 6.9 58.4 36 FBref Prior F_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost F_service raw output over 7 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0118, quality 0.875)	`market_value_eur=73000000` `quality_factor=0.8747` `boost_factor=1.0118` `reference_quality=0.505` `history_matches=42` `history_shrink=0.1064`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0092)	`blend_weight=0.0092` `history_matches=42` `season_overlap_discount=1.0` `per90_to_match_factor=0.9671` `prior_reliability=0.1333` `competition=Premier League`	totalCross: 2.1067 -> 2.1541 totalShots: 2.3987 -> 2.4167 wonTackle: 0.3357 -> 0.3415 wasFouled: 1.0741 -> 1.0687 interceptionWon: 0.5556 -> 0.5505
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` duel_context={'opponent_faced': 'Matty Cash, Tyrone Mings', 'matchup_zone': 'Striker vs Two Centre Backs — 1v2 Aerial/Physical Battle', 'duel_win_rate': 0.277, 'pressing_faced': 'LOW', 'numerical_situation': '1v2', 'xg_multiplier': 1.1126, 'skill_edge': 0.1366, 'height_edge': -0.06, 'attacker_skill': 0.8747, 'defender_skill': 0.5712, 'attacker_role': 'ST', 'attacker_mv_price_index': 1.0, 'defender_adjusted_parts': [{'player': 'Matty Cash', 'market_value': 19800000, 'role': 'CB', 'price_index': 0.75, 'adjusted_quality': 0.6892}, {'player': 'Tyrone Mings', 'market_value': 3200000, 'role': 'CB', 'price_index': 0.75, 'adjusted_quality': 0.4531}], 'xg_override': 0.3881}	totalShots: 2.4167 -> 2.6888 shotOffTarget: 1.4418 -> 1.6041 onTargetScoringAttempt: 0.9615 -> 1.0698 bigChanceMissed: 0.5462 -> 0.6077 expectedGoalsOnTarget: 0.422 -> 0.4695 expectedGoals: 0.3488 -> 0.3881 goals: 0.3367 -> 0.3746
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 1 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=1`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.70 (squad xG sum 1.2053, xG scale 1.4087, xA scale 1.4087, team xA 1.16 = 0.68 x lambda)	`scoreline_reset=true` `team_lambda=1.698` `squad_xg_sum=1.2053` `squad_xa_sum=0.8253` `team_expected_assists=1.1626` `team_xa_to_xg_ratio=0.6847`	expected_goals: 0.3881 -> 0.5466 expected_assists: 0.1128 -> 0.1589
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 5 by projected rating, fifa_overall 84.0)	`match_squad_rank=5` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 1 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=1` `allocation_applied=true` `goals_allocated=1` `goal_probability=0.4211` `goal_rate=0.3746` `scorer_weight=0.5298`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.1469` `assist_rate=0.0157` `provider_weight=0.1271`	-

Team totals 1 1.70 16 7.0

Bench Candidates (7 modeled substitutes with empirical playing time)

Player	Pos	Exp Min	App Prob	Substitute Modeling Details
Dermot Mee	G	4.4m	10.9%	Market-value substitute prior
Leny Yoro	D	16.1m	63.1%	Market-value substitute prior
Ayden Heaven	D	14.7m	24.1%	Market-value substitute prior
Patrick Dorgu	M	16.5m	52.4%	Market-value substitute prior
Mason Mount	M	22.9m	53.9%	Market-value substitute prior
Matheus Cunha	F	27.1m	37.0%	Market-value substitute prior

Aston Villa — projected line vs season baseline (4-2-3-1)

Player Pos Min Proj G Proj xG Season xG/g Proj Sh Season Sh/g Proj Rtg Season Rtg Rec Model / FBref Prior

Zion Suzuki G 90 0 0.00 - n/a 0.0 7.3 42.8 14 FBref Prior G_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost G_service raw output over 3 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0254, quality 0.725)	`market_value_eur=31000000` `quality_factor=0.7249` `boost_factor=1.0254` `reference_quality=0.505` `history_matches=8` `history_shrink=0.3846`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0167)	`blend_weight=0.0167` `history_matches=8` `season_overlap_discount=1.0` `per90_to_match_factor=1.0` `prior_reliability=0.0667` `competition=Premier League`	saves: 3.162 -> 3.1259 wasFouled: 0.4561 -> 0.4485
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` `duel_context={'opponent_faced': 'Bryan Mbeumo, Bruno Fernandes', 'matchup_zone': 'Goalkeeper shot-stopping context', 'opponent_lambda': 1.573}`	-
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 0 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=0`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.08 (squad xG sum 1.1472, xG scale 0.9440, xA scale 1.2035, team xA 0.70 = 0.65 x lambda)	`scoreline_reset=true` `team_lambda=1.083` `squad_xg_sum=1.1472` `squad_xa_sum=0.5849` `team_expected_assists=0.7039` `team_xa_to_xg_ratio=0.65`	shots_faced: 14.9 -> 16.1 expected_saves: 3.64 -> 3.94 high_pressure_saves: 1.26 -> 1.36
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 1 by projected rating, fifa_overall 76.0)	`match_squad_rank=1` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation skipped: 0 projected goals	`team_goals=0` `allocation_applied=false` `goals_allocated=0` `goal_probability=0.0` `scorer_weight=0.0` `squad_weight_total=1.1046`	-
`integer_assist_projection`	Team scoreline allocation skipped: 0 projected assists	`allocation_applied=false` `team_assists=0` `assists_allocated=0` `assist_probability=0.0` `provider_weight=0.0` `squad_weight_total=0.722`	-

Victor Lindelöf D 88 0 0.03 0.02 49% 0 0.1 6.9 38.0 32 FBref Prior D_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost D_service raw output over 7 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 0.9978, quality 0.460)	`market_value_eur=4800000` `quality_factor=0.4597` `boost_factor=0.9978` `reference_quality=0.505` `history_matches=26` `history_shrink=0.1613`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0140)	`blend_weight=0.014` `history_matches=26` `season_overlap_discount=1.0` `per90_to_match_factor=0.9756` `prior_reliability=0.1333` `competition=Premier League`	interceptionWon: 0.7641 -> 0.7534 wonTackle: 0.575 -> 0.567
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` `duel_context={'opponent_faced': 'Bruno Fernandes', 'matchup_zone': 'Left Wing vs Right Back — Flank Battle'}`	-
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 0 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=0`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.08 (squad xG sum 1.1472, xG scale 0.9440, xA scale 1.2035, team xA 0.70 = 0.65 x lambda)	`scoreline_reset=true` `team_lambda=1.083` `squad_xg_sum=1.1472` `squad_xa_sum=0.5849` `team_expected_assists=0.7039` `team_xa_to_xg_ratio=0.65`	-
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 7 by projected rating, fifa_overall 77.0)	`match_squad_rank=7` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation skipped: 0 projected goals	`team_goals=0` `allocation_applied=false` `goals_allocated=0` `goal_probability=0.0328` `goal_rate=0.0` `scorer_weight=0.0266`	-
`integer_assist_projection`	Team scoreline allocation skipped: 0 projected assists	`allocation_applied=false` `team_assists=0` `assists_allocated=0` `assist_probability=0.0158` `assist_rate=0.007` `provider_weight=0.0143`	-

Ian Maatsen D 85 0 0.05 0.04 37% 1 0.7 6.7 50.1 30 FBref Prior D_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost D_service raw output over 7 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0086, quality 0.739)	`market_value_eur=33000000` `quality_factor=0.7389` `boost_factor=1.0086` `reference_quality=0.505` `history_matches=36` `history_shrink=0.122`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0100)	`blend_weight=0.01` `history_matches=36` `season_overlap_discount=1.0` `per90_to_match_factor=0.9476` `prior_reliability=0.1267` `competition=Premier League`	fouls: 0.9756 -> 0.9908 totalCross: 1.7818 -> 1.7689 wonTackle: 1.4636 -> 1.4539 interceptionWon: 1.1209 -> 1.1297 totalShots: 0.889 -> 0.8951
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` `duel_context={'opponent_faced': 'Youri Tielemans', 'matchup_zone': 'Right Wing vs Left Back — Flank Battle'}`	-
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 0 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=0`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.08 (squad xG sum 1.1472, xG scale 0.9440, xA scale 1.2035, team xA 0.70 = 0.65 x lambda)	`scoreline_reset=true` `team_lambda=1.083` `squad_xg_sum=1.1472` `squad_xa_sum=0.5849` `team_expected_assists=0.7039` `team_xa_to_xg_ratio=0.65`	expected_assists: 0.0923 -> 0.1111
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 14 by projected rating, fifa_overall 79.0)	`match_squad_rank=14` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation skipped: 0 projected goals	`team_goals=0` `allocation_applied=false` `goals_allocated=0` `goal_probability=0.0516` `goal_rate=0.0` `scorer_weight=0.0424`	-
`integer_assist_projection`	Team scoreline allocation skipped: 0 projected assists	`allocation_applied=false` `team_assists=0` `assists_allocated=0` `assist_probability=0.1052` `assist_rate=0.0355` `provider_weight=0.0889`	-

Matty Cash D 88 0 0.02 0.02 ok 1 0.6 6.7 56.2 30 FBref Prior D_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost D_service raw output over 7 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0039, quality 0.629)	`market_value_eur=19800000` `quality_factor=0.6291` `boost_factor=1.0039` `reference_quality=0.505` `history_matches=43` `history_shrink=0.1042`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0086)	`blend_weight=0.0086` `history_matches=43` `season_overlap_discount=1.0` `per90_to_match_factor=0.9749` `prior_reliability=0.1267` `competition=Premier League`	totalCross: 2.2294 -> 2.2103 wonTackle: 1.4452 -> 1.4548 wasFouled: 1.1825 -> 1.19
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` `duel_context={'opponent_faced': 'Bryan Mbeumo', 'matchup_zone': 'Striker vs Two Centre Backs — 1v2 Aerial/Physical Battle'}`	-
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 0 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=0`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.08 (squad xG sum 1.1472, xG scale 0.9440, xA scale 1.2035, team xA 0.70 = 0.65 x lambda)	`scoreline_reset=true` `team_lambda=1.083` `squad_xg_sum=1.1472` `squad_xa_sum=0.5849` `team_expected_assists=0.7039` `team_xa_to_xg_ratio=0.65`	expected_assists: 0.052 -> 0.0626
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 11 by projected rating, fifa_overall 81.0)	`match_squad_rank=11` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation skipped: 0 projected goals	`team_goals=0` `allocation_applied=false` `goals_allocated=0` `goal_probability=0.0222` `goal_rate=0.0038` `scorer_weight=0.0179`	-
`integer_assist_projection`	Team scoreline allocation skipped: 0 projected assists	`allocation_applied=false` `team_assists=0` `assists_allocated=0` `assist_probability=0.0607` `assist_rate=0.0407` `provider_weight=0.0557`	-

Tyrone Mings D 89 0 0.05 0.04 ok 1 0.3 6.8 36.6 29 FBref Prior D_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost D_service raw output over 7 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 0.9965, quality 0.440)	`market_value_eur=3200000` `quality_factor=0.4399` `boost_factor=0.9965` `reference_quality=0.505` `history_matches=23` `history_shrink=0.1786`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0015)	`blend_weight=0.0015` `history_matches=23` `season_overlap_discount=1.0` `per90_to_match_factor=0.9889` `prior_reliability=0.0133` `competition=Premier League`	wonTackle: 0.5261 -> 0.5329
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` `duel_context={'opponent_faced': 'Bryan Mbeumo', 'matchup_zone': 'Striker vs Two Centre Backs — 1v2 Aerial/Physical Battle'}`	-
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 0 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=0`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.08 (squad xG sum 1.1472, xG scale 0.9440, xA scale 1.2035, team xA 0.70 = 0.65 x lambda)	`scoreline_reset=true` `team_lambda=1.083` `squad_xg_sum=1.1472` `squad_xa_sum=0.5849` `team_expected_assists=0.7039` `team_xa_to_xg_ratio=0.65`	-
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 10 by projected rating, fifa_overall 77.0)	`match_squad_rank=10` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation skipped: 0 projected goals	`team_goals=0` `allocation_applied=false` `goals_allocated=0` `goal_probability=0.0519` `goal_rate=0.0` `scorer_weight=0.0426`	-
`integer_assist_projection`	Team scoreline allocation skipped: 0 projected assists	`allocation_applied=false` `team_assists=0` `assists_allocated=0` `assist_probability=0.0201` `assist_rate=0.0` `provider_weight=0.0162`	-

Boubacar Kamara M 87 0 0.04 0.01 173% 0 0.2 6.9 48.4 16 FBref Prior M_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost M_service raw output over 8 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0148, quality 0.821)	`market_value_eur=49000000` `quality_factor=0.8206` `boost_factor=1.0148` `reference_quality=0.505` `history_matches=27` `history_shrink=0.1562`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0135)	`blend_weight=0.0135` `history_matches=27` `season_overlap_discount=1.0` `per90_to_match_factor=0.963` `prior_reliability=0.1333` `competition=Premier League`	wasFouled: 2.4387 -> 2.4122 wonTackle: 1.7789 -> 1.7679 totalShots: 0.5149 -> 0.5079 fouls: 1.9986 -> 2.0041 interceptionWon: 0.8701 -> 0.8648 totalCross: 0.0988 -> 0.104
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` duel_context={'opponent_faced': 'Kobbie Mainoo', 'matchup_zone': 'CDM vs CDM', 'duel_win_rate': 0.401, 'pressing_faced': 'ULTRA_HIGH', 'numerical_situation': '1v1', 'xg_multiplier': 0.972, 'skill_edge': -0.0191, 'height_edge': 0.0, 'attacker_skill': 0.847, 'defender_skill': 0.8893, 'attacker_role': 'CDM', 'attacker_mv_price_index': 0.85, 'defender_adjusted_parts': [{'player': 'Kobbie Mainoo', 'market_value': 77000000, 'role': 'CDM', 'price_index': 0.85, 'adjusted_quality': 0.8893}], 'xg_override': 0.0417}	aerialWon: 1.2149 -> 1.1 totalShots: 0.5129 -> 0.4986 shotOffTarget: 0.3693 -> 0.359
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 0 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=0`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.08 (squad xG sum 1.1472, xG scale 0.9440, xA scale 1.2035, team xA 0.70 = 0.65 x lambda)	`scoreline_reset=true` `team_lambda=1.083` `squad_xg_sum=1.1472` `squad_xa_sum=0.5849` `team_expected_assists=0.7039` `team_xa_to_xg_ratio=0.65`	expected_assists: 0.0333 -> 0.0401
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 6 by projected rating, fifa_overall 84.0)	`match_squad_rank=6` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation skipped: 0 projected goals	`team_goals=0` `allocation_applied=false` `goals_allocated=0` `goal_probability=0.0386` `goal_rate=0.0502` `scorer_weight=0.0431`	-
`integer_assist_projection`	Team scoreline allocation skipped: 0 projected assists	`allocation_applied=false` `team_assists=0` `assists_allocated=0` `assist_probability=0.0393` `assist_rate=0.0573` `provider_weight=0.0501`	-

Amadou Onana M 85 0 0.08 0.08 ok 1 0.8 6.8 45.3 17 XGBoost M_service

Step-by-step (9 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost M_service raw output over 8 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0130, quality 0.764)	`market_value_eur=37000000` `quality_factor=0.7641` `boost_factor=1.013` `reference_quality=0.505` `history_matches=25` `history_shrink=0.1667`	-
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` duel_context={'opponent_faced': 'Kobbie Mainoo, Andrey Santos', 'matchup_zone': 'Attacking Mid vs Double Pivot — Tight Space Battle', 'duel_win_rate': 0.254, 'pressing_faced': 'ULTRA_HIGH', 'numerical_situation': '1v2', 'xg_multiplier': 1.0205, 'skill_edge': -0.038, 'height_edge': 0.052, 'attacker_skill': 0.7752, 'defender_skill': 0.8597, 'attacker_role': 'CAM', 'attacker_mv_price_index': 0.95, 'defender_adjusted_parts': [{'player': 'Kobbie Mainoo', 'market_value': 77000000, 'role': 'CDM', 'price_index': 0.85, 'adjusted_quality': 0.8893}, {'player': 'Andrey Santos', 'market_value': 44000000, 'role': 'CDM', 'price_index': 0.85, 'adjusted_quality': 0.8301}], 'xg_override': 0.0844}	aerialWon: 1.9008 -> 1.4 totalShots: 1.0254 -> 1.0464 shotOffTarget: 0.7464 -> 0.7617 bigChanceMissed: 0.4305 -> 0.4393 onTargetScoringAttempt: 0.279 -> 0.2847
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 0 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=0`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.08 (squad xG sum 1.1472, xG scale 0.9440, xA scale 1.2035, team xA 0.70 = 0.65 x lambda)	`scoreline_reset=true` `team_lambda=1.083` `squad_xg_sum=1.1472` `squad_xa_sum=0.5849` `team_expected_assists=0.7039` `team_xa_to_xg_ratio=0.65`	expected_assists: 0.0372 -> 0.0448
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 8 by projected rating, fifa_overall 81.0)	`match_squad_rank=8` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation skipped: 0 projected goals	`team_goals=0` `allocation_applied=false` `goals_allocated=0` `goal_probability=0.0766` `goal_rate=0.0651` `scorer_weight=0.0683`	-
`integer_assist_projection`	Team scoreline allocation skipped: 0 projected assists	`allocation_applied=false` `team_assists=0` `assists_allocated=0` `assist_probability=0.0438` `assist_rate=0.0542` `provider_weight=0.0535`	-

João Gomes M 82 0 0.03 0.02 ok 0 0.4 6.8 53.9 29 FBref Prior M_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost M_service raw output over 8 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0083, quality 0.758)	`market_value_eur=36000000` `quality_factor=0.7581` `boost_factor=1.0083` `reference_quality=0.505` `history_matches=41` `history_shrink=0.1087`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0019)	`blend_weight=0.0019` `history_matches=41` `season_overlap_discount=1.0` `per90_to_match_factor=0.9076` `prior_reliability=0.0267` `competition=Premier League`	wonTackle: 1.7996 -> 1.8048
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` duel_context={'opponent_faced': 'Kobbie Mainoo', 'matchup_zone': 'CDM vs CDM', 'duel_win_rate': 0.344, 'pressing_faced': 'ULTRA_HIGH', 'numerical_situation': '1v1', 'xg_multiplier': 0.889, 'skill_edge': -0.0435, 'height_edge': -0.032, 'attacker_skill': 0.7926, 'defender_skill': 0.8893, 'attacker_role': 'CDM', 'attacker_mv_price_index': 0.85, 'defender_adjusted_parts': [{'player': 'Kobbie Mainoo', 'market_value': 77000000, 'role': 'CDM', 'price_index': 0.85, 'adjusted_quality': 0.8893}], 'xg_override': 0.0274}	aerialWon: 0.5536 -> 0.9 totalShots: 0.5093 -> 0.4528 shotOffTarget: 0.3613 -> 0.3212 onTargetScoringAttempt: 0.148 -> 0.1316 goals: 0.0465 -> 0.0413
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 0 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=0`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.08 (squad xG sum 1.1472, xG scale 0.9440, xA scale 1.2035, team xA 0.70 = 0.65 x lambda)	`scoreline_reset=true` `team_lambda=1.083` `squad_xg_sum=1.1472` `squad_xa_sum=0.5849` `team_expected_assists=0.7039` `team_xa_to_xg_ratio=0.65`	expected_assists: 0.05 -> 0.0602
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 9 by projected rating, fifa_overall 78.0)	`match_squad_rank=9` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation skipped: 0 projected goals	`team_goals=0` `allocation_applied=false` `goals_allocated=0` `goal_probability=0.0256` `goal_rate=0.0413` `scorer_weight=0.0306`	-
`integer_assist_projection`	Team scoreline allocation skipped: 0 projected assists	`allocation_applied=false` `team_assists=0` `assists_allocated=0` `assist_probability=0.0584` `assist_rate=0.0576` `provider_weight=0.0648`	-

Johan Manzambi M 76 0 0.24 0.15 65% 3 1.5 7.0 39.3 11 XGBoost M_service

Step-by-step (9 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost M_service raw output over 8 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0501, quality 0.872)	`market_value_eur=71000000` `quality_factor=0.8721` `boost_factor=1.0501` `reference_quality=0.505` `history_matches=6` `history_shrink=0.4545`	-
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` duel_context={'opponent_faced': 'Lisandro Martínez', 'matchup_zone': 'Left Wing vs Right Back — Flank Battle', 'duel_win_rate': 0.51, 'pressing_faced': 'ULTRA_HIGH', 'numerical_situation': '1v1', 'xg_multiplier': 1.0435, 'skill_edge': 0.0016, 'height_edge': 0.028, 'attacker_skill': 0.8721, 'defender_skill': 0.8686, 'attacker_role': 'LW', 'attacker_mv_price_index': 1.0, 'defender_adjusted_parts': [{'player': 'Lisandro Martínez', 'market_value': 48000000, 'role': 'RB', 'price_index': 0.7, 'adjusted_quality': 0.8686}], 'xg_override': 0.2546}	aerialWon: 0.0 -> 1.1 totalShots: 2.6802 -> 2.7968 shotOffTarget: 1.7668 -> 1.8437 onTargetScoringAttempt: 0.9133 -> 0.953 expectedGoalsOnTarget: 0.2916 -> 0.3043 goals: 0.2472 -> 0.258 expectedGoals: 0.244 -> 0.2546
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 0 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=0`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.08 (squad xG sum 1.1472, xG scale 0.9440, xA scale 1.2035, team xA 0.70 = 0.65 x lambda)	`scoreline_reset=true` `team_lambda=1.083` `squad_xg_sum=1.1472` `squad_xa_sum=0.5849` `team_expected_assists=0.7039` `team_xa_to_xg_ratio=0.65`	expected_goals: 0.2546 -> 0.2403 expected_assists: 0.1173 -> 0.1411
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 5 by projected rating, fifa_overall 80.0)	`match_squad_rank=5` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation skipped: 0 projected goals	`team_goals=0` `allocation_applied=false` `goals_allocated=0` `goal_probability=0.2136` `goal_rate=0.258` `scorer_weight=0.243`	-
`integer_assist_projection`	Team scoreline allocation skipped: 0 projected assists	`allocation_applied=false` `team_assists=0` `assists_allocated=0` `assist_probability=0.1316` `assist_rate=0.1634` `provider_weight=0.1764`	-

Lamare Bogarde M 80 0 0.04 - n/a 1 0.0 6.7 30.7 32 FBref Prior M_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost M_service raw output over 8 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0044, quality 0.606)	`market_value_eur=17500000` `quality_factor=0.6058` `boost_factor=1.0044` `reference_quality=0.505` `history_matches=29` `history_shrink=0.1471`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0013)	`blend_weight=0.0013` `history_matches=29` `season_overlap_discount=1.0` `per90_to_match_factor=0.8942` `prior_reliability=0.0133` `competition=Premier League`	-
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` duel_context={'opponent_faced': 'Luke Shaw', 'matchup_zone': 'Right Wing vs Left Back — Flank Battle', 'duel_win_rate': 0.534, 'pressing_faced': 'LOW', 'numerical_situation': '1v1', 'xg_multiplier': 1.079, 'skill_edge': 0.0337, 'height_edge': 0.02, 'attacker_skill': 0.6058, 'defender_skill': 0.5308, 'attacker_role': 'RW', 'attacker_mv_price_index': 1.0, 'defender_adjusted_parts': [{'player': 'Luke Shaw', 'market_value': 7500000, 'role': 'LB', 'price_index': 0.7, 'adjusted_quality': 0.5308}], 'xg_override': 0.0387}	aerialWon: 0.4989 -> 1.1 totalShots: 0.6079 -> 0.6559 shotOffTarget: 0.4272 -> 0.4609 onTargetScoringAttempt: 0.1807 -> 0.195
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 0 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=0`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.08 (squad xG sum 1.1472, xG scale 0.9440, xA scale 1.2035, team xA 0.70 = 0.65 x lambda)	`scoreline_reset=true` `team_lambda=1.083` `squad_xg_sum=1.1472` `squad_xa_sum=0.5849` `team_expected_assists=0.7039` `team_xa_to_xg_ratio=0.65`	expected_assists: 0.0338 -> 0.0407
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 12 by projected rating, fifa_overall 74.0)	`match_squad_rank=12` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation skipped: 0 projected goals	`team_goals=0` `allocation_applied=false` `goals_allocated=0` `goal_probability=0.0358` `goal_rate=0.0636` `scorer_weight=0.0456`	-
`integer_assist_projection`	Team scoreline allocation skipped: 0 projected assists	`allocation_applied=false` `team_assists=0` `assists_allocated=0` `assist_probability=0.0399` `assist_rate=0.0455` `provider_weight=0.0464`	-

Nicolas Jackson F 82 0 0.37 0.17 124% 3 1.6 7.2 46.6 13 FBref Prior F_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost F_service raw output over 7 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0341, quality 0.800)	`market_value_eur=44000000` `quality_factor=0.8002` `boost_factor=1.0341` `reference_quality=0.505` `history_matches=8` `history_shrink=0.3846`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0133)	`blend_weight=0.0133` `history_matches=8` `season_overlap_discount=1.0` `per90_to_match_factor=0.9122` `prior_reliability=0.0533` `competition=Premier League`	totalShots: 2.466 -> 2.4331 wasFouled: 1.2927 -> 1.2755 onTargetScoringAttempt: 0.9066 -> 0.8945 totalCross: 0.2601 -> 0.2718 totalOffside: 0.6975 -> 0.6882 goals: 0.4099 -> 0.4044 fouls: 0.3872 -> 0.382
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` duel_context={'opponent_faced': 'Matthijs de Ligt, Harry Maguire', 'matchup_zone': 'Striker vs Two Centre Backs — 1v2 Aerial/Physical Battle', 'duel_win_rate': 0.25, 'pressing_faced': 'LOW', 'numerical_situation': '1v2', 'xg_multiplier': 1.0735, 'skill_edge': 0.07, 'height_edge': -0.02, 'attacker_skill': 0.8002, 'defender_skill': 0.6447, 'attacker_role': 'ST', 'attacker_mv_price_index': 1.0, 'defender_adjusted_parts': [{'player': 'Matthijs de Ligt', 'market_value': 26000000, 'role': 'CB', 'price_index': 0.75, 'adjusted_quality': 0.7498}, {'player': 'Harry Maguire', 'market_value': 8600000, 'role': 'CB', 'price_index': 0.75, 'adjusted_quality': 0.5395}], 'xg_override': 0.3933}	totalShots: 2.4539 -> 2.6343 shotOffTarget: 1.5594 -> 1.674 onTargetScoringAttempt: 0.8945 -> 0.9602 goals: 0.4044 -> 0.4341 expectedGoalsOnTarget: 0.3669 -> 0.3939 expectedGoals: 0.3664 -> 0.3933 bigChanceMissed: 0.2654 -> 0.2849
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 0 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=0`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.08 (squad xG sum 1.1472, xG scale 0.9440, xA scale 1.2035, team xA 0.70 = 0.65 x lambda)	`scoreline_reset=true` `team_lambda=1.083` `squad_xg_sum=1.1472` `squad_xa_sum=0.5849` `team_expected_assists=0.7039` `team_xa_to_xg_ratio=0.65`	expected_goals: 0.3933 -> 0.3713 expected_assists: 0.0658 -> 0.0792
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 3 by projected rating, fifa_overall 79.0)	`match_squad_rank=3` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation skipped: 0 projected goals	`team_goals=0` `allocation_applied=false` `goals_allocated=0` `goal_probability=0.3102` `goal_rate=0.4341` `scorer_weight=0.4055`	-
`integer_assist_projection`	Team scoreline allocation skipped: 0 projected assists	`allocation_applied=false` `team_assists=0` `assists_allocated=0` `assist_probability=0.0761` `assist_rate=0.0` `provider_weight=0.0634`	-

Team totals 0 1.08 14 6.9

Bench Candidates (7 modeled substitutes with empirical playing time)

Player	Pos	Exp Min	App Prob	Substitute Modeling Details
Marco Bizot	G	4.4m	13.1%	Market-value substitute prior
Matteo Ruggeri	D	11.5m	62.3%	Market-value substitute prior
Taylor Harwood-Bellis	D	14.6m	34.5%	Market-value substitute prior
Leon Goretzka	M	22.3m	51.2%	Market-value substitute prior
Ross Barkley	M	22.0m	30.8%	Market-value substitute prior
Ibrahim Mbaye	F	26.4m	60.8%	Market-value substitute prior

Antagonist / tactical summary

Midfield is evenly contested — Kobbie Mainoo vs Johan Manzambi will be the pivotal duel. Total antagonist xG — Home: 0.87, Away: 0.84. Home danger zone: Striker vs Two Centre Backs — 1v2 Aerial/Physical Battle. Away danger zone: Striker vs Two Centre Backs — 1v2 Aerial/Physical Battle.

Midfield — possession 50.1% / 49.9%, dominance ratio 0.52, key duel: Kobbie Mainoo vs Johan Manzambi

Bruno Fernandes xG 0.25Youri Tielemans xG 0.06Carlos Baleba xG 0.06Johan Manzambi xG 0.25Lamare Bogarde xG 0.04Amadou Onana xG 0.08

Summary

Antagonist matchup analysis for Man Utd vs Aston Villa. Most likely scoreline: 1-0. Lambdas derived from 6 home and 6 away individual player duels. Head-to-head: Man Utd 6W-2D-2L vs Aston Villa over 10 meetings.

Tactical parity between 4-2-3-1 and 4-2-3-1. Match outcome driven primarily by individual player quality. Midfield is evenly contested — Kobbie Mainoo vs Johan Manzambi will be the pivotal duel. Total antagonist xG — Home: 0.87, Away: 0.84. Home danger zone: Striker vs Two Centre Backs — 1v2 Aerial/Physical Battle. Away danger zone: Striker vs Two Centre Backs — 1v2 Aerial/Physical Battle.

Atlético Madrid vs Barcelona

LaLiga · 20261108 · 4-4-2 vs 4-3-3

Predicted 1-2Winner: BarcelonaWinning H 2-1 (6.4%)Winning A 1-2 (9.0%)Confidence 0.69λ home/away 1.52 / 2.12Most-likely P 9.0%

Actual vs predicted: result not available yet (upcoming fixture).

XGBoost Reasoning & FBref Dataset Tracing

Antagonist + XGBoost match analysis for Atlético Madrid vs Barcelona. Player statistics regressed using position-specific XGBoost models (G, D, M, F) with 2026-2027 FBref priors. Micro-duels resolved across 4-4-2 vs 4-3-3 (duel lambdas: Home 1.585, Away 2.675). Final calibrated Poisson lambdas: Home 1.521, Away 2.119 (most likely score 1-2).

Home Starters 11/11 XGBoost (7 FBref) Away Starters 11/11 XGBoost (11 FBref) Bench Modeled H: 7 / A: 7 Duel λ (H/A) 1.58 / 2.67 Final λ (H/A) 1.52 / 2.12

FBref Team Priors: Atlético Madrid (Gls/90: 2.3333333333333335, GA/90: 2.33 → E[λ]: 3.5) vs FC Barcelona (Gls/90: 3.6666666666666665, GA/90: 4.0 → E[λ]: 3.5) | Blend Weight: 0.228

Modeling Pipeline Stages (5 layers)

1. XGBoost player regressions for volume & event rates with market-value quality scaling
2. Multi-source prior blending (FBref 2026-2027 tables + Sofascore season history)
3. Antagonist formation duel matrix resolving 1v1 and zonal matchups
4. Team-level prior calibration (FBref possession and attack/defense efficiencies)
5. Bivariate Poisson matrix with Dixon-Coles low-score correction

Standing & form context

Metric	Atlético Madrid	Barcelona
League position	2.00	1.00
Points per game	2.13	2.25
Attack index	1.21	1.77
Defense (concede) index	0.83	0.79
Form multiplier	0.99	1.01
Standing λ multiplier	1.00	1.15

Lambda decomposition (how the goal rates were built)

Stage	Atlético Madrid λ	Barcelona λ
FBref Squad Prior E[λ] (weight 0.228)	3.500	3.500
Pre-competitive-floor	1.551	2.732
Pre-calibration (Micro-duels)	1.585	2.675
Final λ	1.521	2.119

competition calibration scale 0.85competition observed total goals 2.984quality gap applied Truemismatch score pressure applied Falsecompetitive underdog floor applied Falsestage discount 1.0

Scoreline distribution & markets

Top 5 scorelines	P
1-2	9.0%
1-1	8.5%
2-2	6.8%
2-1	6.4%
1-3	6.3%

Home win	27.6%
Draw	21.0%
Away win	51.4%
Over 2.5	70.4%
BTTS yes	68.8%

Pregame fair line — favorite Barcelona (51.4%), fair total 3.64, fair spread -0.60

Projected players vs season baseline

Proj = projected for this match · Season = this season's per-game average from match data · deviation badge compares projected xG to the player's season xG/game.

Atlético Madrid — projected line vs season baseline (4-4-2)

Player Pos Min Proj G Proj xG Season xG/g Proj Sh Season Sh/g Proj Rtg Season Rtg Rec Model / FBref Prior

Jan Oblak G 90 0 0.00 - n/a 0.0 7.0 59.2 29 FBref Prior G_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost G_service raw output over 3 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0028, quality 0.590)	`market_value_eur=16000000` `quality_factor=0.59` `boost_factor=1.0028` `reference_quality=0.505` `history_matches=41` `history_shrink=0.1087`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0141)	`blend_weight=0.0141` `history_matches=41` `season_overlap_discount=1.0` `per90_to_match_factor=1.0` `prior_reliability=0.2` `competition=La Liga`	-
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` `duel_context={'opponent_faced': 'Raphinha, Lamine Yamal', 'matchup_zone': 'Goalkeeper shot-stopping context', 'opponent_lambda': 2.6747}`	-
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 1 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=1`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.52 (squad xG sum 1.0663, xG scale 1.4264, xA scale 1.1809, team xA 1.22 = 0.80 x lambda)	`scoreline_reset=true` `team_lambda=1.521` `squad_xg_sum=1.0663` `squad_xa_sum=1.0304` `team_expected_assists=1.2168` `team_xa_to_xg_ratio=0.8`	shots_faced: 25.4 -> 20.1 expected_saves: 6.22 -> 4.92 high_pressure_saves: 2.14 -> 1.7
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 8 by projected rating, fifa_overall 88.0)	`match_squad_rank=8` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 1 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=1` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.0` `scorer_weight=0.0` `squad_weight_total=1.6848`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.0` `provider_weight=0.0` `squad_weight_total=1.1637`	-

Cristian Romero D 79 0 0.17 0.08 114% 1 0.8 7.1 57.6 19 FBref Prior D_service

Step-by-step (9 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost D_service raw output over 7 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0114, quality 0.764)	`market_value_eur=37000000` `quality_factor=0.7641` `boost_factor=1.0114` `reference_quality=0.505` `history_matches=29` `history_shrink=0.1471`	-
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` `duel_context={'opponent_faced': 'Raphinha', 'matchup_zone': 'Right Wing vs Left Back — Flank Battle'}`	-
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 1 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=1`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.52 (squad xG sum 1.0663, xG scale 1.4264, xA scale 1.1809, team xA 1.22 = 0.80 x lambda)	`scoreline_reset=true` `team_lambda=1.521` `squad_xg_sum=1.0663` `squad_xa_sum=1.0304` `team_expected_assists=1.2168` `team_xa_to_xg_ratio=0.8`	expected_goals: 0.1169 -> 0.1667 expected_assists: 0.0609 -> 0.0719
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 2 by projected rating, fifa_overall 82.0)	`match_squad_rank=2` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 1 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=1` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.1535` `goal_rate=0.2199` `scorer_weight=0.2084`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.0694` `assist_rate=0.0036` `provider_weight=0.0575`	-

Marc Pubill D 83 0 0.06 0.04 76% 0 0.4 7.0 51.8 30 FBref Prior D_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost D_service raw output over 7 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0142, quality 0.817)	`market_value_eur=48000000` `quality_factor=0.8168` `boost_factor=1.0142` `reference_quality=0.505` `history_matches=28` `history_shrink=0.1515`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0112)	`blend_weight=0.0112` `history_matches=28` `season_overlap_discount=1.0` `per90_to_match_factor=0.9213` `prior_reliability=0.1133` `competition=La Liga`	interceptionWon: 0.8736 -> 0.8639 totalShots: 0.4825 -> 0.489 totalCross: 0.5212 -> 0.5275 goals: 0.0924 -> 0.0974
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` `duel_context={'opponent_faced': 'Anthony Gordon', 'matchup_zone': 'Striker vs Two Centre Backs — 1v2 Aerial/Physical Battle'}`	-
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 1 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=1`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.52 (squad xG sum 1.0663, xG scale 1.4264, xA scale 1.1809, team xA 1.22 = 0.80 x lambda)	`scoreline_reset=true` `team_lambda=1.521` `squad_xg_sum=1.0663` `squad_xa_sum=1.0304` `team_expected_assists=1.2168` `team_xa_to_xg_ratio=0.8`	expected_goals: 0.0455 -> 0.0649
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 9 by projected rating, fifa_overall 87.0)	`match_squad_rank=9` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv_filled` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 1 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=1` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.0628` `goal_rate=0.0974` `scorer_weight=0.0811`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.0214` `assist_rate=0.006` `provider_weight=0.0177`	-

Marcos Llorente D 85 0 0.06 0.04 65% 1 0.6 7.1 62.7 33 FBref Prior D_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost D_service raw output over 7 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0045, quality 0.641)	`market_value_eur=21000000` `quality_factor=0.6408` `boost_factor=1.0045` `reference_quality=0.505` `history_matches=40` `history_shrink=0.1111`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0072)	`blend_weight=0.0072` `history_matches=40` `season_overlap_discount=1.0` `per90_to_match_factor=0.9429` `prior_reliability=0.1` `competition=La Liga`	fouls: 0.9675 -> 0.9878 totalCross: 1.5559 -> 1.5447 interceptionWon: 1.0244 -> 1.0352 wasFouled: 0.1724 -> 0.1802
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` `duel_context={'opponent_faced': 'Lamine Yamal', 'matchup_zone': 'Left Wing vs Right Back — Flank Battle'}`	-
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 1 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=1`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.52 (squad xG sum 1.0663, xG scale 1.4264, xA scale 1.1809, team xA 1.22 = 0.80 x lambda)	`scoreline_reset=true` `team_lambda=1.521` `squad_xg_sum=1.0663` `squad_xa_sum=1.0304` `team_expected_assists=1.2168` `team_xa_to_xg_ratio=0.8`	expected_goals: 0.0438 -> 0.0625 expected_assists: 0.0746 -> 0.0881
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 5 by projected rating, fifa_overall 85.0)	`match_squad_rank=5` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 1 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=1` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.0606` `goal_rate=0.0` `scorer_weight=0.05`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.0843` `assist_rate=0.079` `provider_weight=0.0912`	-

Dávid Hancko D 88 0 0.07 0.05 57% 1 0.7 7.0 56.0 29 FBref Prior D_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost D_service raw output over 7 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0083, quality 0.758)	`market_value_eur=36000000` `quality_factor=0.7581` `boost_factor=1.0083` `reference_quality=0.505` `history_matches=41` `history_shrink=0.1087`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0141)	`blend_weight=0.0141` `history_matches=41` `season_overlap_discount=1.0` `per90_to_match_factor=0.9832` `prior_reliability=0.2` `competition=La Liga`	interceptionWon: 0.9477 -> 0.9621 wasFouled: 0.1548 -> 0.1665 wonTackle: 0.7045 -> 0.7131
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` `duel_context={'opponent_faced': 'Anthony Gordon', 'matchup_zone': 'Striker vs Two Centre Backs — 1v2 Aerial/Physical Battle'}`	-
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 1 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=1`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.52 (squad xG sum 1.0663, xG scale 1.4264, xA scale 1.1809, team xA 1.22 = 0.80 x lambda)	`scoreline_reset=true` `team_lambda=1.521` `squad_xg_sum=1.0663` `squad_xa_sum=1.0304` `team_expected_assists=1.2168` `team_xa_to_xg_ratio=0.8`	expected_goals: 0.0521 -> 0.0743 expected_assists: 0.0397 -> 0.0469
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 7 by projected rating, fifa_overall 83.0)	`match_squad_rank=7` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 1 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=1` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.0716` `goal_rate=0.0` `scorer_weight=0.0594`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.0458` `assist_rate=0.1271` `provider_weight=0.0586`	-

Pablo Barrios M 71 0 0.09 0.04 157% 1 0.8 7.2 61.9 16 FBref Prior M_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost M_service raw output over 8 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0156, quality 0.879)	`market_value_eur=77000000` `quality_factor=0.8792` `boost_factor=1.0156` `reference_quality=0.505` `history_matches=31` `history_shrink=0.1389`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0138)	`blend_weight=0.0138` `history_matches=31` `season_overlap_discount=1.0` `per90_to_match_factor=0.7857` `prior_reliability=0.1533` `competition=La Liga`	wasFouled: 0.7257 -> 0.7393 totalCross: 0.9926 -> 0.9836 totalShots: 0.919 -> 0.9111
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` duel_context={'opponent_faced': 'Rodri, Pedri', 'matchup_zone': 'CM vs CDM + CM', 'duel_win_rate': 0.297, 'pressing_faced': 'ULTRA_HIGH', 'numerical_situation': '1v2', 'xg_multiplier': 0.9954, 'skill_edge': 0.0029, 'height_edge': -0.006, 'attacker_skill': 0.8863, 'defender_skill': 0.8799, 'attacker_role': 'CM', 'attacker_mv_price_index': 0.9, 'defender_adjusted_parts': [{'player': 'Rodri', 'market_value': 54000000, 'role': 'CDM', 'price_index': 0.85, 'adjusted_quality': 0.8599}, {'player': 'Pedri', 'market_value': 164000000, 'role': 'CM', 'price_index': 0.9, 'adjusted_quality': 0.8999}], 'xg_override': 0.0634}	aerialWon: 0.6727 -> 1.1
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 1 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=1`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.52 (squad xG sum 1.0663, xG scale 1.4264, xA scale 1.1809, team xA 1.22 = 0.80 x lambda)	`scoreline_reset=true` `team_lambda=1.521` `squad_xg_sum=1.0663` `squad_xa_sum=1.0304` `team_expected_assists=1.2168` `team_xa_to_xg_ratio=0.8`	expected_goals: 0.0634 -> 0.0904 expected_assists: 0.1154 -> 0.1363
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 1 by projected rating, fifa_overall 83.0)	`match_squad_rank=1` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 1 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=1` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.0864` `goal_rate=0.062` `scorer_weight=0.0893`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.1274` `assist_rate=0.0942` `provider_weight=0.1188`	-

Alejandro Grimaldo M 89 0 0.19 0.18 ok 1 1.2 7.1 54.9 21 XGBoost M_service

Step-by-step (9 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost M_service raw output over 8 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0120, quality 0.625)	`market_value_eur=19400000` `quality_factor=0.6251` `boost_factor=1.012` `reference_quality=0.505` `history_matches=10` `history_shrink=0.3333`	-
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` duel_context={'opponent_faced': 'Eric García', 'matchup_zone': 'Left Mid vs Right Back — Wide Channel', 'duel_win_rate': 0.32, 'pressing_faced': 'ULTRA_HIGH', 'numerical_situation': '1v1', 'xg_multiplier': 0.7947, 'skill_edge': -0.0956, 'height_edge': -0.044, 'attacker_skill': 0.6461, 'defender_skill': 0.8586, 'attacker_role': 'LM', 'attacker_mv_price_index': 0.9, 'defender_adjusted_parts': [{'player': 'Eric García', 'market_value': 44000000, 'role': 'RB', 'price_index': 0.7, 'adjusted_quality': 0.8586}], 'xg_override': 0.1302}	aerialWon: 0.2793 -> 0.8 totalShots: 1.6114 -> 1.2806 shotOffTarget: 1.0006 -> 0.7952 onTargetScoringAttempt: 0.6108 -> 0.4854 expectedGoalsOnTarget: 0.3581 -> 0.2846 goals: 0.1664 -> 0.1322 expectedGoals: 0.1638 -> 0.1302 bigChanceMissed: 0.1314 -> 0.1044
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 1 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=1`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.52 (squad xG sum 1.0663, xG scale 1.4264, xA scale 1.1809, team xA 1.22 = 0.80 x lambda)	`scoreline_reset=true` `team_lambda=1.521` `squad_xg_sum=1.0663` `squad_xa_sum=1.0304` `team_expected_assists=1.2168` `team_xa_to_xg_ratio=0.8`	expected_goals: 0.1302 -> 0.1857 expected_assists: 0.1628 -> 0.1922
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 3 by projected rating, fifa_overall 89.0)	`match_squad_rank=3` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv_filled` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 1 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=1` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.1695` `goal_rate=0.1322` `scorer_weight=0.1878`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=1` `assist_probability=0.1749` `assist_rate=0.1625` `provider_weight=0.1919`	-

Obed Vargas M 82 0 0.06 0.01 741% 0 0.2 6.8 34.5 24 XGBoost M_service

Step-by-step (9 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost M_service raw output over 8 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0008, quality 0.514)	`market_value_eur=9300000` `quality_factor=0.5142` `boost_factor=1.0008` `reference_quality=0.505` `history_matches=12` `history_shrink=0.2941`	-
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` duel_context={'opponent_faced': 'Xavi Espart', 'matchup_zone': 'Right Mid vs Left Back — Wide Channel', 'duel_win_rate': 0.353, 'pressing_faced': 'MEDIUM', 'numerical_situation': '1v1', 'xg_multiplier': 0.8422, 'skill_edge': -0.1074, 'height_edge': 0.0, 'attacker_skill': 0.5264, 'defender_skill': 0.765, 'attacker_role': 'RM', 'attacker_mv_price_index': 0.9, 'defender_adjusted_parts': [{'player': 'Xavi Espart', 'market_value': 26000000, 'role': 'LB', 'price_index': 0.7, 'adjusted_quality': 0.765}], 'xg_override': 0.0444}	aerialWon: 0.2407 -> 0.9 totalShots: 0.5864 -> 0.4939 shotOffTarget: 0.4147 -> 0.3493 onTargetScoringAttempt: 0.1717 -> 0.1446 goals: 0.0599 -> 0.0504 expectedGoals: 0.0527 -> 0.0444
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 1 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=1`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.52 (squad xG sum 1.0663, xG scale 1.4264, xA scale 1.1809, team xA 1.22 = 0.80 x lambda)	`scoreline_reset=true` `team_lambda=1.521` `squad_xg_sum=1.0663` `squad_xa_sum=1.0304` `team_expected_assists=1.2168` `team_xa_to_xg_ratio=0.8`	expected_goals: 0.0444 -> 0.0633 expected_assists: 0.0459 -> 0.0542
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 13 by projected rating, fifa_overall 72.0)	`match_squad_rank=13` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 1 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=1` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.0613` `goal_rate=0.0504` `scorer_weight=0.0673`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.0528` `assist_rate=0.0643` `provider_weight=0.0646`	-

Giuliano Simeone M 80 0 0.19 0.11 74% 2 1.0 7.1 61.3 32 FBref Prior M_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost M_service raw output over 8 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0094, quality 0.800)	`market_value_eur=44000000` `quality_factor=0.8002` `boost_factor=1.0094` `reference_quality=0.505` `history_matches=42` `history_shrink=0.1064`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0111)	`blend_weight=0.0111` `history_matches=42` `season_overlap_discount=1.0` `per90_to_match_factor=0.8879` `prior_reliability=0.16` `competition=La Liga`	interceptionWon: 0.624 -> 0.6335 wonTackle: 1.3249 -> 1.3307
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` duel_context={'opponent_faced': 'Rodri, Pedri', 'matchup_zone': 'CM vs CDM + CM', 'duel_win_rate': 0.235, 'pressing_faced': 'ULTRA_HIGH', 'numerical_situation': '1v2', 'xg_multiplier': 0.9046, 'skill_edge': -0.0269, 'height_edge': -0.038, 'attacker_skill': 0.8202, 'defender_skill': 0.8799, 'attacker_role': 'CM', 'attacker_mv_price_index': 0.9, 'defender_adjusted_parts': [{'player': 'Rodri', 'market_value': 54000000, 'role': 'CDM', 'price_index': 0.85, 'adjusted_quality': 0.8599}, {'player': 'Pedri', 'market_value': 164000000, 'role': 'CM', 'price_index': 0.9, 'adjusted_quality': 0.8999}], 'xg_override': 0.1365}	aerialWon: 1.7213 -> 0.9 totalShots: 1.5374 -> 1.3907 shotOffTarget: 0.9537 -> 0.8627 onTargetScoringAttempt: 0.5819 -> 0.5264 goals: 0.1738 -> 0.1572 expectedGoals: 0.1509 -> 0.1365 expectedGoalsOnTarget: 0.1238 -> 0.112 bigChanceMissed: 0.0705 -> 0.0638
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 1 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=1`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.52 (squad xG sum 1.0663, xG scale 1.4264, xA scale 1.1809, team xA 1.22 = 0.80 x lambda)	`scoreline_reset=true` `team_lambda=1.521` `squad_xg_sum=1.0663` `squad_xa_sum=1.0304` `team_expected_assists=1.2168` `team_xa_to_xg_ratio=0.8`	expected_goals: 0.1365 -> 0.1947 expected_assists: 0.1244 -> 0.1469
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 6 by projected rating, fifa_overall 81.0)	`match_squad_rank=6` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv_filled` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 1 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=1` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.1769` `goal_rate=0.1572` `scorer_weight=0.2163`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.1366` `assist_rate=0.1045` `provider_weight=0.1301`	-

Kang-in Lee F 72 0 0.16 0.15 ok 2 2.1 7.1 56.7 17 XGBoost F_service

Step-by-step (9 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost F_service raw output over 7 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0243, quality 0.764)	`market_value_eur=37000000` `quality_factor=0.7641` `boost_factor=1.0243` `reference_quality=0.505` `history_matches=11` `history_shrink=0.3125`	-
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` duel_context={'opponent_faced': 'Pau Cubarsí, Gerard Martín', 'matchup_zone': 'Striker vs Two Centre Backs — 1v2 Aerial/Physical Battle', 'duel_win_rate': 0.18, 'pressing_faced': 'ULTRA_HIGH', 'numerical_situation': '1v2', 'xg_multiplier': 0.8677, 'skill_edge': -0.042, 'height_edge': -0.048, 'attacker_skill': 0.7641, 'defender_skill': 0.8575, 'attacker_role': 'ST', 'attacker_mv_price_index': 1.0, 'defender_adjusted_parts': [{'player': 'Pau Cubarsí', 'market_value': 94000000, 'role': 'CB', 'price_index': 0.75, 'adjusted_quality': 0.8981}, {'player': 'Gerard Martín', 'market_value': 36000000, 'role': 'CB', 'price_index': 0.75, 'adjusted_quality': 0.8168}], 'xg_override': 0.1111}	totalShots: 2.2502 -> 1.9525 shotOffTarget: 1.5687 -> 1.3612 onTargetScoringAttempt: 0.6815 -> 0.5913 goals: 0.1824 -> 0.1583 expectedGoalsOnTarget: 0.1657 -> 0.1438 expectedGoals: 0.128 -> 0.1111 bigChanceMissed: 0.1127 -> 0.0978
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 1 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=1`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.52 (squad xG sum 1.0663, xG scale 1.4264, xA scale 1.1809, team xA 1.22 = 0.80 x lambda)	`scoreline_reset=true` `team_lambda=1.521` `squad_xg_sum=1.0663` `squad_xa_sum=1.0304` `team_expected_assists=1.2168` `team_xa_to_xg_ratio=0.8`	expected_goals: 0.1111 -> 0.1585 expected_assists: 0.185 -> 0.2185
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 4 by projected rating, fifa_overall 80.0)	`match_squad_rank=4` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 1 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=1` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.1466` `goal_rate=0.1583` `scorer_weight=0.1981`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.1963` `assist_rate=0.0641` `provider_weight=0.1748`	-

Ademola Lookman F 61 1 0.22 0.21 ok 2 1.4 6.8 59.7 28 FBref Prior F_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost F_service raw output over 7 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0158, quality 0.800)	`market_value_eur=44000000` `quality_factor=0.8002` `boost_factor=1.0158` `reference_quality=0.505` `history_matches=23` `history_shrink=0.1786`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0201)	`blend_weight=0.0201` `history_matches=23` `season_overlap_discount=1.0` `per90_to_match_factor=0.678` `prior_reliability=0.1733` `competition=La Liga`	wasFouled: 0.6804 -> 0.6929 totalCross: 0.9711 -> 0.9621 interceptionWon: 0.012 -> 0.017
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` duel_context={'opponent_faced': 'Pau Cubarsí, Gerard Martín', 'matchup_zone': 'Striker vs Two Centre Backs — 1v2 Aerial/Physical Battle', 'duel_win_rate': 0.18, 'pressing_faced': 'ULTRA_HIGH', 'numerical_situation': '1v2', 'xg_multiplier': 0.8975, 'skill_edge': -0.0257, 'height_edge': -0.044, 'attacker_skill': 0.8002, 'defender_skill': 0.8575, 'attacker_role': 'ST', 'attacker_mv_price_index': 1.0, 'defender_adjusted_parts': [{'player': 'Pau Cubarsí', 'market_value': 94000000, 'role': 'CB', 'price_index': 0.75, 'adjusted_quality': 0.8981}, {'player': 'Gerard Martín', 'market_value': 36000000, 'role': 'CB', 'price_index': 0.75, 'adjusted_quality': 0.8168}], 'xg_override': 0.1574}	totalShots: 1.6412 -> 1.473 shotOffTarget: 1.0032 -> 0.9004 onTargetScoringAttempt: 0.638 -> 0.5726 bigChanceMissed: 0.3449 -> 0.3095 goals: 0.2191 -> 0.1966 expectedGoals: 0.1754 -> 0.1574 expectedGoalsOnTarget: 0.1115 -> 0.1001
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 1 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=1`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.52 (squad xG sum 1.0663, xG scale 1.4264, xA scale 1.1809, team xA 1.22 = 0.80 x lambda)	`scoreline_reset=true` `team_lambda=1.521` `squad_xg_sum=1.0663` `squad_xa_sum=1.0304` `team_expected_assists=1.2168` `team_xa_to_xg_ratio=0.8`	expected_goals: 0.1574 -> 0.2246 expected_assists: 0.0766 -> 0.0905
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 15 by projected rating, fifa_overall 83.0)	`match_squad_rank=15` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 1 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=1` `allocation_applied=true` `goals_allocated=1` `goal_probability=0.2012` `goal_rate=0.1966` `scorer_weight=0.267`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.0865` `assist_rate=0.0` `provider_weight=0.0724`	-

Team totals 1 1.52 14 7.0

Bench Candidates (7 modeled substitutes with empirical playing time)

Player	Pos	Exp Min	App Prob	Substitute Modeling Details
Juan Musso	G	4.4m	10.9%	Market-value substitute prior
Robin Le Normand	D	14.2m	23.0%	Market-value substitute prior
Jorge Domínguez	D	13.8m	22.2%	Market-value substitute prior
Morten Hjulmand	M	17.7m	46.6%	Market-value substitute prior
Rodrigo Mendoza	M	13.6m	45.2%	Market-value substitute prior
Julián Alvarez	F	25.9m	75.6%	Market-value substitute prior

Barcelona — projected line vs season baseline (4-3-3)

Player Pos Min Proj G Proj xG Season xG/g Proj Sh Season Sh/g Proj Rtg Season Rtg Rec Model / FBref Prior

Joan García G 87 0 0.00 - n/a 0.0 6.9 63.1 32 FBref Prior G_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost G_service raw output over 3 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0101, quality 0.809)	`market_value_eur=46000000` `quality_factor=0.8089` `boost_factor=1.0101` `reference_quality=0.505` `history_matches=40` `history_shrink=0.1111`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0144)	`blend_weight=0.0144` `history_matches=40` `season_overlap_discount=1.0` `per90_to_match_factor=0.9627` `prior_reliability=0.2` `competition=La Liga`	saves: 2.714 -> 2.698
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` `duel_context={'opponent_faced': 'Ademola Lookman, Giuliano Simeone', 'matchup_zone': 'Goalkeeper shot-stopping context', 'opponent_lambda': 1.5854}`	-
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 2 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=2`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 2.12 (squad xG sum 2.2093, xG scale 0.9591, xA scale 0.9591, team xA 1.45 = 0.68 x lambda)	`scoreline_reset=true` `team_lambda=2.119` `squad_xg_sum=2.2093` `squad_xa_sum=1.5111` `team_expected_assists=1.4494` `team_xa_to_xg_ratio=0.684`	shots_faced: 15.1 -> 14.4 expected_saves: 3.7 -> 3.52 high_pressure_saves: 1.27 -> 1.22
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 13 by projected rating, fifa_overall 86.0)	`match_squad_rank=13` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 2 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=2` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.0` `scorer_weight=0.0` `squad_weight_total=2.4716`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.0` `provider_weight=0.0` `squad_weight_total=1.6516`	-

Pau Cubarsí D 87 0 0.03 0.02 ok 0 0.2 7.2 55.7 33 FBref Prior D_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost D_service raw output over 7 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0132, quality 0.891)	`market_value_eur=94000000` `quality_factor=0.891` `boost_factor=1.0132` `reference_quality=0.505` `history_matches=39` `history_shrink=0.1136`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0098)	`blend_weight=0.0098` `history_matches=39` `season_overlap_discount=1.0` `per90_to_match_factor=0.9674` `prior_reliability=0.1333` `competition=La Liga`	interceptionWon: 1.1901 -> 1.1831 totalShots: 0.3374 -> 0.3436 wonTackle: 0.9014 -> 0.9068
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` `duel_context={'opponent_faced': 'Kang-in Lee', 'matchup_zone': 'Striker vs Two Centre Backs — 1v2 Aerial/Physical Battle'}`	-
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 2 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=2`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 2.12 (squad xG sum 2.2093, xG scale 0.9591, xA scale 0.9591, team xA 1.45 = 0.68 x lambda)	`scoreline_reset=true` `team_lambda=2.119` `squad_xg_sum=2.2093` `squad_xa_sum=1.5111` `team_expected_assists=1.4494` `team_xa_to_xg_ratio=0.684`	-
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 6 by projected rating, fifa_overall 86.0)	`match_squad_rank=6` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 2 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=2` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.0247` `goal_rate=0.0069` `scorer_weight=0.02`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.0778` `assist_rate=0.2178` `provider_weight=0.1013`	-

Xavi Espart D 79 0 0.04 0.01 176% 1 0.4 6.9 31.6 24 FBref Prior D_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost D_service raw output over 7 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0160, quality 0.686)	`market_value_eur=26000000` `quality_factor=0.6858` `boost_factor=1.016` `reference_quality=0.505` `history_matches=12` `history_shrink=0.2941`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0382)	`blend_weight=0.0382` `history_matches=12` `season_overlap_discount=1.0` `per90_to_match_factor=0.8827` `prior_reliability=0.2` `competition=La Liga`	fouls: 1.6331 -> 1.6044 totalCross: 0.4689 -> 0.496 wasFouled: 1.004 -> 0.9769 goalAssist: 0.1516 -> 0.1683 interceptionWon: 1.0828 -> 1.0976 totalShots: 0.8839 -> 0.8727
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` `duel_context={'opponent_faced': 'Obed Vargas', 'matchup_zone': 'Right Mid vs Left Back — Wide Channel'}`	-
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 2 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=2`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 2.12 (squad xG sum 2.2093, xG scale 0.9591, xA scale 0.9591, team xA 1.45 = 0.68 x lambda)	`scoreline_reset=true` `team_lambda=2.119` `squad_xg_sum=2.2093` `squad_xa_sum=1.5111` `team_expected_assists=1.4494` `team_xa_to_xg_ratio=0.684`	-
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 14 by projected rating, fifa_overall 67.0)	`match_squad_rank=14` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 2 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=2` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.0361` `goal_rate=0.1171` `scorer_weight=0.046`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.0645` `assist_rate=0.1683` `provider_weight=0.0834`	-

Eric García D 82 0 0.03 0.08 65% 1 0.6 7.1 58.3 30 FBref Prior D_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost D_service raw output over 7 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0094, quality 0.800)	`market_value_eur=44000000` `quality_factor=0.8002` `boost_factor=1.0094` `reference_quality=0.505` `history_matches=42` `history_shrink=0.1064`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0088)	`blend_weight=0.0088` `history_matches=42` `season_overlap_discount=1.0` `per90_to_match_factor=0.9074` `prior_reliability=0.1267` `competition=La Liga`	wonTackle: 0.8947 -> 0.9078
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` `duel_context={'opponent_faced': 'Alejandro Grimaldo', 'matchup_zone': 'Left Mid vs Right Back — Wide Channel'}`	-
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 2 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=2`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 2.12 (squad xG sum 2.2093, xG scale 0.9591, xA scale 0.9591, team xA 1.45 = 0.68 x lambda)	`scoreline_reset=true` `team_lambda=2.119` `squad_xg_sum=2.2093` `squad_xa_sum=1.5111` `team_expected_assists=1.4494` `team_xa_to_xg_ratio=0.684`	-
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 11 by projected rating, fifa_overall 85.0)	`match_squad_rank=11` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 2 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=2` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.0265` `goal_rate=0.0007` `scorer_weight=0.0215`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.0684` `assist_rate=0.0127` `provider_weight=0.0567`	-

Gerard Martín D 80 0 0.02 0.01 88% 0 0.2 7.1 57.8 32 FBref Prior D_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost D_service raw output over 7 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0081, quality 0.758)	`market_value_eur=36000000` `quality_factor=0.7581` `boost_factor=1.0081` `reference_quality=0.505` `history_matches=42` `history_shrink=0.1064`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0092)	`blend_weight=0.0092` `history_matches=42` `season_overlap_discount=1.0` `per90_to_match_factor=0.888` `prior_reliability=0.1333` `competition=La Liga`	-
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` `duel_context={'opponent_faced': 'Kang-in Lee', 'matchup_zone': 'Striker vs Two Centre Backs — 1v2 Aerial/Physical Battle'}`	-
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 2 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=2`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 2.12 (squad xG sum 2.2093, xG scale 0.9591, xA scale 0.9591, team xA 1.45 = 0.68 x lambda)	`scoreline_reset=true` `team_lambda=2.119` `squad_xg_sum=2.2093` `squad_xa_sum=1.5111` `team_expected_assists=1.4494` `team_xa_to_xg_ratio=0.684`	-
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 8 by projected rating, fifa_overall 79.0)	`match_squad_rank=8` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 2 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=2` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.0191` `goal_rate=0.0` `scorer_weight=0.0154`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.0431` `assist_rate=0.0` `provider_weight=0.0353`	-

Rodri M 84 0 0.07 0.07 ok 1 0.7 7.4 70.4 27 FBref Prior M_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost M_service raw output over 8 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0138, quality 0.837)	`market_value_eur=54000000` `quality_factor=0.837` `boost_factor=1.0138` `reference_quality=0.505` `history_matches=31` `history_shrink=0.1389`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0036)	`blend_weight=0.0036` `history_matches=31` `season_overlap_discount=1.0` `per90_to_match_factor=0.9367` `prior_reliability=0.04` `competition=La Liga`	-
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` duel_context={'opponent_faced': 'Pablo Barrios', 'matchup_zone': 'Holding Mid vs Central Mid — Midfield Anchor Duel', 'duel_win_rate': 0.428, 'pressing_faced': 'ULTRA_HIGH', 'numerical_situation': '1v1', 'xg_multiplier': 1.0414, 'skill_edge': -0.0119, 'height_edge': 0.04, 'attacker_skill': 0.8599, 'defender_skill': 0.8863, 'attacker_role': 'CDM', 'attacker_mv_price_index': 0.85, 'defender_adjusted_parts': [{'player': 'Pablo Barrios', 'market_value': 77000000, 'role': 'CM', 'price_index': 0.9, 'adjusted_quality': 0.8863}], 'xg_override': 0.0726}	aerialWon: 1.4841 -> 1.3 totalShots: 0.7147 -> 0.7443 shotOffTarget: 0.4803 -> 0.5002 onTargetScoringAttempt: 0.2344 -> 0.2441
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 2 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=2`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 2.12 (squad xG sum 2.2093, xG scale 0.9591, xA scale 0.9591, team xA 1.45 = 0.68 x lambda)	`scoreline_reset=true` `team_lambda=2.119` `squad_xg_sum=2.2093` `squad_xa_sum=1.5111` `team_expected_assists=1.4494` `team_xa_to_xg_ratio=0.684`	-
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 3 by projected rating, fifa_overall 90.0)	`match_squad_rank=3` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 2 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=2` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.0672` `goal_rate=0.088` `scorer_weight=0.0783`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.11` `assist_rate=0.247` `provider_weight=0.1456`	-

Pedri M 77 0 0.06 0.04 56% 1 0.5 7.3 65.6 29 FBref Prior M_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost M_service raw output over 8 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0132, quality 0.900)	`market_value_eur=164000000` `quality_factor=0.8997` `boost_factor=1.0132` `reference_quality=0.505` `history_matches=40` `history_shrink=0.1111`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0111)	`blend_weight=0.0111` `history_matches=40` `season_overlap_discount=1.0` `per90_to_match_factor=0.8581` `prior_reliability=0.1533` `competition=La Liga`	wonTackle: 0.7808 -> 0.7887 interceptionWon: 0.5149 -> 0.5092
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` duel_context={'opponent_faced': 'Pablo Barrios', 'matchup_zone': 'Central Mid vs Central Mid — Box-to-Box Midfield Contest', 'duel_win_rate': 0.398, 'pressing_faced': 'ULTRA_HIGH', 'numerical_situation': '1v1', 'xg_multiplier': 0.9678, 'skill_edge': 0.0061, 'height_edge': -0.028, 'attacker_skill': 0.8999, 'defender_skill': 0.8863, 'attacker_role': 'CM', 'attacker_mv_price_index': 0.9, 'defender_adjusted_parts': [{'player': 'Pablo Barrios', 'market_value': 77000000, 'role': 'CM', 'price_index': 0.9, 'adjusted_quality': 0.8863}], 'xg_override': 0.0675}	aerialWon: 0.1882 -> 0.9 totalShots: 0.7383 -> 0.7145 shotOffTarget: 0.5399 -> 0.5225 onTargetScoringAttempt: 0.1961 -> 0.1898
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 2 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=2`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 2.12 (squad xG sum 2.2093, xG scale 0.9591, xA scale 0.9591, team xA 1.45 = 0.68 x lambda)	`scoreline_reset=true` `team_lambda=2.119` `squad_xg_sum=2.2093` `squad_xa_sum=1.5111` `team_expected_assists=1.4494` `team_xa_to_xg_ratio=0.684`	expected_assists: 0.1897 -> 0.1819
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 5 by projected rating, fifa_overall 90.0)	`match_squad_rank=5` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 2 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=2` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.0627` `goal_rate=0.0846` `scorer_weight=0.0741`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.1663` `assist_rate=0.1621` `provider_weight=0.161`	-

Fermín López M 79 0 0.18 0.25 ok 2 1.9 7.1 65.2 31 FBref Prior M_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost M_service raw output over 8 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0130, quality 0.896)	`market_value_eur=113000000` `quality_factor=0.8965` `boost_factor=1.013` `reference_quality=0.505` `history_matches=40` `history_shrink=0.1111`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0111)	`blend_weight=0.0111` `history_matches=40` `season_overlap_discount=1.0` `per90_to_match_factor=0.8771` `prior_reliability=0.1533` `competition=La Liga`	onTargetScoringAttempt: 0.7664 -> 0.779 goals: 0.2235 -> 0.2337 wasFouled: 0.7252 -> 0.7341 wonTackle: 0.5721 -> 0.5784
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` duel_context={'opponent_faced': 'Pablo Barrios', 'matchup_zone': 'Central Mid vs Central Mid — Box-to-Box Midfield Contest', 'duel_win_rate': 0.405, 'pressing_faced': 'ULTRA_HIGH', 'numerical_situation': '1v1', 'xg_multiplier': 0.9784, 'skill_edge': 0.0053, 'height_edge': -0.02, 'attacker_skill': 0.8981, 'defender_skill': 0.8863, 'attacker_role': 'CM', 'attacker_mv_price_index': 0.9, 'defender_adjusted_parts': [{'player': 'Pablo Barrios', 'market_value': 77000000, 'role': 'CM', 'price_index': 0.9, 'adjusted_quality': 0.8863}], 'xg_override': 0.1871}	aerialWon: 0.4589 -> 0.9 totalShots: 2.1743 -> 2.1274 shotOffTarget: 1.3953 -> 1.3652 onTargetScoringAttempt: 0.779 -> 0.7622 expectedGoalsOnTarget: 0.4777 -> 0.4674 goals: 0.2337 -> 0.2287
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 2 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=2`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 2.12 (squad xG sum 2.2093, xG scale 0.9591, xA scale 0.9591, team xA 1.45 = 0.68 x lambda)	`scoreline_reset=true` `team_lambda=2.119` `squad_xg_sum=2.2093` `squad_xa_sum=1.5111` `team_expected_assists=1.4494` `team_xa_to_xg_ratio=0.684`	expected_goals: 0.1871 -> 0.1795 expected_assists: 0.1922 -> 0.1843
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 10 by projected rating, fifa_overall 90.0)	`match_squad_rank=10` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv_filled` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 2 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=2` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.1643` `goal_rate=0.2287` `scorer_weight=0.211`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.1683` `assist_rate=0.2038` `provider_weight=0.1931`	-

Lamine Yamal F 82 1 0.57 0.50 ok 6 3.8 8.4 69.3 32 FBref Prior F_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost F_service raw output over 7 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0135, quality 0.900)	`market_value_eur=237000000` `quality_factor=0.9` `boost_factor=1.0135` `reference_quality=0.505` `history_matches=39` `history_shrink=0.1136`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0133)	`blend_weight=0.0133` `history_matches=39` `season_overlap_discount=1.0` `per90_to_match_factor=0.9064` `prior_reliability=0.18` `competition=La Liga`	totalOffside: 0.6054 -> 0.6241 totalCross: 2.4648 -> 2.4811 goalAssist: 0.7137 -> 0.7042 fouls: 1.2605 -> 1.2527
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` duel_context={'opponent_faced': 'Marcos Llorente', 'matchup_zone': 'Left Wing vs Right Back — Flank Battle', 'duel_win_rate': 0.558, 'pressing_faced': 'MEDIUM', 'numerical_situation': '1v1', 'xg_multiplier': 1.1148, 'skill_edge': 0.0821, 'height_edge': -0.004, 'attacker_skill': 0.9, 'defender_skill': 0.7176, 'attacker_role': 'LW', 'attacker_mv_price_index': 1.0, 'defender_adjusted_parts': [{'player': 'Marcos Llorente', 'market_value': 21000000, 'role': 'RB', 'price_index': 0.7, 'adjusted_quality': 0.7176}], 'xg_override': 0.5924}	totalShots: 5.0286 -> 5.6059 shotOffTarget: 3.5176 -> 3.9214 onTargetScoringAttempt: 1.511 -> 1.6845 goals: 0.8143 -> 0.9078 expectedGoalsOnTarget: 0.68 -> 0.7581 expectedGoals: 0.5314 -> 0.5924 bigChanceMissed: 0.5155 -> 0.5747
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 2 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=2`	goals: 1.0 -> 0 assists: 1.0 -> 0
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 2.12 (squad xG sum 2.2093, xG scale 0.9591, xA scale 0.9591, team xA 1.45 = 0.68 x lambda)	`scoreline_reset=true` `team_lambda=2.119` `squad_xg_sum=2.2093` `squad_xa_sum=1.5111` `team_expected_assists=1.4494` `team_xa_to_xg_ratio=0.684`	expected_goals: 0.5924 -> 0.5682 expected_assists: 0.37 -> 0.355
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 1 by projected rating, fifa_overall 90.0)	`match_squad_rank=1` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 2 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=2` `allocation_applied=true` `goals_allocated=1` `goal_probability=0.4335` `goal_rate=0.9078` `scorer_weight=0.7103`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=1` `assist_probability=0.2988` `assist_rate=0.7042` `provider_weight=0.4437`	-

Raphinha F 76 1 0.62 0.59 ok 4 2.4 7.5 61.8 27 FBref Prior F_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost F_service raw output over 7 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0143, quality 0.859)	`market_value_eur=63000000` `quality_factor=0.8589` `boost_factor=1.0143` `reference_quality=0.505` `history_matches=32` `history_shrink=0.1351`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0146)	`blend_weight=0.0146` `history_matches=32` `season_overlap_discount=1.0` `per90_to_match_factor=0.8487` `prior_reliability=0.1667` `competition=La Liga`	goals: 0.8049 -> 0.818 onTargetScoringAttempt: 1.6487 -> 1.6592 totalCross: 2.6884 -> 2.6987 fouls: 0.3963 -> 0.4054 totalOffside: 0.4905 -> 0.4982
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` duel_context={'opponent_faced': 'Cristian Romero', 'matchup_zone': 'Right Wing vs Left Back — Flank Battle', 'duel_win_rate': 0.455, 'pressing_faced': 'HIGH', 'numerical_situation': '1v1', 'xg_multiplier': 0.9638, 'skill_edge': 0.0114, 'height_edge': -0.036, 'attacker_skill': 0.8589, 'defender_skill': 0.8336, 'attacker_role': 'RW', 'attacker_mv_price_index': 1.0, 'defender_adjusted_parts': [{'player': 'Cristian Romero', 'market_value': 37000000, 'role': 'LB', 'price_index': 0.7, 'adjusted_quality': 0.8336}], 'xg_override': 0.6492}	totalShots: 3.5302 -> 3.4024 shotOffTarget: 1.871 -> 1.8033 onTargetScoringAttempt: 1.6592 -> 1.5991 expectedGoalsOnTarget: 1.0861 -> 1.0468 goals: 0.818 -> 0.7884 bigChanceMissed: 0.6852 -> 0.6604 expectedGoals: 0.6736 -> 0.6492
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 2 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=2`	goals: 1.0 -> 0
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 2.12 (squad xG sum 2.2093, xG scale 0.9591, xA scale 0.9591, team xA 1.45 = 0.68 x lambda)	`scoreline_reset=true` `team_lambda=2.119` `squad_xg_sum=2.2093` `squad_xa_sum=1.5111` `team_expected_assists=1.4494` `team_xa_to_xg_ratio=0.684`	expected_goals: 0.6492 -> 0.6227 expected_assists: 0.1267 -> 0.1215
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 2 by projected rating, fifa_overall 88.0)	`match_squad_rank=2` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 2 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=2` `allocation_applied=true` `goals_allocated=1` `goal_probability=0.4635` `goal_rate=0.7884` `scorer_weight=0.7467`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.1144` `assist_rate=0.216` `provider_weight=0.1519`	-

Anthony Gordon F 78 0 0.30 0.26 ok 3 1.7 7.3 54.9 30 FBref Prior F_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost F_service raw output over 7 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0139, quality 0.884)	`market_value_eur=82000000` `quality_factor=0.8837` `boost_factor=1.0139` `reference_quality=0.505` `history_matches=36` `history_shrink=0.122`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0106)	`blend_weight=0.0106` `history_matches=36` `season_overlap_discount=1.0` `per90_to_match_factor=0.8626` `prior_reliability=0.1333` `competition=La Liga`	totalCross: 1.7488 -> 1.7349 goalAssist: 0.1713 -> 0.1832 totalOffside: 0.3281 -> 0.3337
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` duel_context={'opponent_faced': 'Marc Pubill, Dávid Hancko', 'matchup_zone': 'Striker vs Two Centre Backs — 1v2 Aerial/Physical Battle', 'duel_win_rate': 0.196, 'pressing_faced': 'HIGH', 'numerical_situation': '1v2', 'xg_multiplier': 0.9944, 'skill_edge': 0.0202, 'height_edge': -0.024, 'attacker_skill': 0.8837, 'defender_skill': 0.8388, 'attacker_role': 'ST', 'attacker_mv_price_index': 1.0, 'defender_adjusted_parts': [{'player': 'Marc Pubill', 'market_value': 48000000, 'role': 'CB', 'price_index': 0.75, 'adjusted_quality': 0.8608}, {'player': 'Dávid Hancko', 'market_value': 36000000, 'role': 'CB', 'price_index': 0.75, 'adjusted_quality': 0.8168}], 'xg_override': 0.3118}	totalShots: 2.6382 -> 2.6234 shotOffTarget: 1.4009 -> 1.3931 onTargetScoringAttempt: 1.2339 -> 1.227
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 2 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=2`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 2.12 (squad xG sum 2.2093, xG scale 0.9591, xA scale 0.9591, team xA 1.45 = 0.68 x lambda)	`scoreline_reset=true` `team_lambda=2.119` `squad_xg_sum=2.2093` `squad_xa_sum=1.5111` `team_expected_assists=1.4494` `team_xa_to_xg_ratio=0.684`	expected_goals: 0.3118 -> 0.2991 expected_assists: 0.1326 -> 0.1272
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 4 by projected rating, fifa_overall 82.0)	`match_squad_rank=4` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 2 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=2` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.2585` `goal_rate=0.3354` `scorer_weight=0.3186`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.1194` `assist_rate=0.1832` `provider_weight=0.159`	-

Team totals 2 2.12 22 7.3

Bench Candidates (7 modeled substitutes with empirical playing time)

Player	Pos	Exp Min	App Prob	Substitute Modeling Details
Dominik Livaković	G	4.5m	11.1%	Market-value substitute prior
Jules Koundé	D	12.9m	50.7%	Market-value substitute prior
Alejandro Balde	D	15.0m	36.4%	Market-value substitute prior
Dani Olmo	M	25.5m	59.7%	Market-value substitute prior
Marc Bernal	M	23.3m	33.4%	Market-value substitute prior
Karim Adeyemi	F	19.1m	61.9%	Market-value substitute prior

Antagonist / tactical summary

Midfield is evenly contested — Pablo Barrios vs Pedri will be the pivotal duel. Total antagonist xG — Home: 0.64, Away: 1.88. Home danger zone: Striker vs Two Centre Backs — 1v2 Aerial/Physical Battle. Away danger zone: Right Wing vs Left Back — Flank Battle.

Midfield — possession 48.8% / 51.2%, dominance ratio 0.45, key duel: Pablo Barrios vs Pedri

Kang-in Lee xG 0.11Ademola Lookman xG 0.16Alejandro Grimaldo xG 0.13Lamine Yamal xG 0.59Raphinha xG 0.65Anthony Gordon xG 0.31

Summary

Antagonist matchup analysis for Atlético Madrid vs Barcelona. Most likely scoreline: 1-2. Lambdas derived from 6 home and 6 away individual player duels. Head-to-head: Atlético Madrid 3W-1D-6L vs Barcelona over 10 meetings.

Tactical parity between 4-4-2 and 4-3-3. Match outcome driven primarily by individual player quality. Midfield is evenly contested — Pablo Barrios vs Pedri will be the pivotal duel. Total antagonist xG — Home: 0.64, Away: 1.88. Home danger zone: Striker vs Two Centre Backs — 1v2 Aerial/Physical Battle. Away danger zone: Right Wing vs Left Back — Flank Battle.

Elche vs Real Betis

LaLiga · 20261108 · 4-3-3 vs 4-2-3-1

Predicted 2-1Winner: ElcheWinning H 2-1 (8.8%)Winning A 1-2 (7.9%)Confidence 0.65λ home/away 1.57 / 1.41Most-likely P 8.8%

Actual vs predicted: result not available yet (upcoming fixture).

XGBoost Reasoning & FBref Dataset Tracing

Antagonist + XGBoost match analysis for Elche vs Real Betis. Player statistics regressed using position-specific XGBoost models (G, D, M, F) with 2026-2027 FBref priors. Micro-duels resolved across 4-3-3 vs 4-2-3-1 (duel lambdas: Home 1.171, Away 1.294). Final calibrated Poisson lambdas: Home 1.574, Away 1.410 (most likely score 2-1).

Home Starters 11/11 XGBoost (9 FBref) Away Starters 11/11 XGBoost (9 FBref) Bench Modeled H: 7 / A: 7 Duel λ (H/A) 1.17 / 1.29 Final λ (H/A) 1.57 / 1.41

FBref Team Priors: Elche (Gls/90: 1.0, GA/90: 1.0 → E[λ]: 0.985) vs Real Betis (Gls/90: 1.3333333333333333, GA/90: 1.33 → E[λ]: 0.988) | Blend Weight: 0.228

Modeling Pipeline Stages (5 layers)

1. XGBoost player regressions for volume & event rates with market-value quality scaling
2. Multi-source prior blending (FBref 2026-2027 tables + Sofascore season history)
3. Antagonist formation duel matrix resolving 1v1 and zonal matchups
4. Team-level prior calibration (FBref possession and attack/defense efficiencies)
5. Bivariate Poisson matrix with Dixon-Coles low-score correction

Standing & form context

Metric	Elche	Real Betis
League position	16.00	4.00
Points per game	1.14	1.84
Attack index	0.97	1.00
Defense (concede) index	0.91	1.06
Form multiplier	0.96	1.04
Standing λ multiplier	0.94	1.03

Lambda decomposition (how the goal rates were built)

Stage	Elche λ	Real Betis λ
FBref Squad Prior E[λ] (weight 0.228)	0.985	0.988
Pre-competitive-floor	1.251	1.291
Pre-calibration (Micro-duels)	1.171	1.294
Final λ	1.574	1.410

competition calibration scale 1.174competition observed total goals 2.984quality gap applied Truemismatch score pressure applied Falsecompetitive underdog floor applied Falsestage discount 1.0

Scoreline distribution & markets

Top 5 scorelines	P
1-1	11.3%
2-1	8.8%
1-2	7.9%
1-0	7.9%
0-1	7.0%

Home win	41.4%
Draw	24.5%
Away win	34.2%
Over 2.5	57.3%
BTTS yes	60.0%

Pregame fair line — favorite Elche (41.4%), fair total 2.98, fair spread 0.16

Projected players vs season baseline

Proj = projected for this match · Season = this season's per-game average from match data · deviation badge compares projected xG to the player's season xG/game.

Elche — projected line vs season baseline (4-3-3)

Player Pos Min Proj G Proj xG Season xG/g Proj Sh Season Sh/g Proj Rtg Season Rtg Rec Model / FBref Prior

Matías Dituro G 90 0 0.00 - n/a 0.0 7.3 54.9 34 FBref Prior G_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost G_service raw output over 3 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 0.9973, quality 0.437)	`market_value_eur=105000` `quality_factor=0.4374` `boost_factor=0.9973` `reference_quality=0.505` `history_matches=32` `history_shrink=0.1351`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0176)	`blend_weight=0.0176` `history_matches=32` `season_overlap_discount=1.0` `per90_to_match_factor=1.0` `prior_reliability=0.2` `competition=La Liga`	saves: 2.9148 -> 2.9046
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` `duel_context={'opponent_faced': 'Abdessamad Ezzalzouli, Pablo Fornals', 'matchup_zone': 'Goalkeeper shot-stopping context', 'opponent_lambda': 1.2935}`	-
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 2 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=2`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.57 (squad xG sum 0.9357, xG scale 1.6822, xA scale 1.6822, team xA 1.04 = 0.66 x lambda)	`scoreline_reset=true` `team_lambda=1.574` `squad_xg_sum=0.9357` `squad_xa_sum=0.6189` `team_expected_assists=1.0411` `team_xa_to_xg_ratio=0.6615`	shots_faced: 12.3 -> 13.4 expected_saves: 3.01 -> 3.28 high_pressure_saves: 1.03 -> 1.13
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 1 by projected rating, fifa_overall 76.0)	`match_squad_rank=1` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 2 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=2` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.0` `scorer_weight=0.0` `squad_weight_total=1.7598`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.0` `provider_weight=0.0` `squad_weight_total=1.0833`	-

Buba Sangaré D 77 0 0.01 0.01 ok 0 0.1 6.7 51.0 27 FBref Prior D_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost D_service raw output over 7 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 0.9975, quality 0.461)	`market_value_eur=4900000` `quality_factor=0.4609` `boost_factor=0.9975` `reference_quality=0.505` `history_matches=21` `history_shrink=0.1923`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0242)	`blend_weight=0.0242` `history_matches=21` `season_overlap_discount=1.0` `per90_to_match_factor=0.8517` `prior_reliability=0.1933` `competition=La Liga`	wonTackle: 1.0755 -> 1.0921 interceptionWon: 0.712 -> 0.7019 totalCross: 0.9708 -> 0.9615 goals: 0.0602 -> 0.0658 onTargetScoringAttempt: 0.0602 -> 0.0657 fouls: 1.2586 -> 1.2637
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` `duel_context={'opponent_faced': 'Abdessamad Ezzalzouli', 'matchup_zone': 'Striker vs Two Centre Backs — 1v2 Aerial/Physical Battle'}`	-
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 2 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=2`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.57 (squad xG sum 0.9357, xG scale 1.6822, xA scale 1.6822, team xA 1.04 = 0.66 x lambda)	`scoreline_reset=true` `team_lambda=1.574` `squad_xg_sum=0.9357` `squad_xa_sum=0.6189` `team_expected_assists=1.0411` `team_xa_to_xg_ratio=0.6615`	expected_assists: 0.014 -> 0.0236
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 17 by projected rating, fifa_overall 69.0)	`match_squad_rank=17` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 2 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=2` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.0062` `goal_rate=0.0657` `scorer_weight=0.0077`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.0233` `assist_rate=0.0` `provider_weight=0.0189`	-

Víctor Chust D 90 0 0.05 0.02 162% 1 0.3 6.9 62.1 32 FBref Prior D_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost D_service raw output over 7 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 0.9978, quality 0.437)	`market_value_eur=2700000` `quality_factor=0.4374` `boost_factor=0.9978` `reference_quality=0.505` `history_matches=41` `history_shrink=0.1087`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0141)	`blend_weight=0.0141` `history_matches=41` `season_overlap_discount=1.0` `per90_to_match_factor=0.9958` `prior_reliability=0.2` `competition=La Liga`	totalCross: 0.9215 -> 0.9366
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` `duel_context={'opponent_faced': 'Pablo Fornals', 'matchup_zone': 'Right Wing vs Left Back — Flank Battle'}`	-
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 2 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=2`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.57 (squad xG sum 0.9357, xG scale 1.6822, xA scale 1.6822, team xA 1.04 = 0.66 x lambda)	`scoreline_reset=true` `team_lambda=1.574` `squad_xg_sum=0.9357` `squad_xa_sum=0.6189` `team_expected_assists=1.0411` `team_xa_to_xg_ratio=0.6615`	expected_goals: 0.0326 -> 0.0548 expected_assists: 0.0306 -> 0.0515
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 5 by projected rating, fifa_overall 74.0)	`match_squad_rank=5` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 2 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=2` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.0533` `goal_rate=0.033` `scorer_weight=0.0551`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.0502` `assist_rate=0.056` `provider_weight=0.0644`	-

Roy Revivo D 89 0 0.03 0.02 57% 1 0.3 6.8 48.3 10 XGBoost D_service

Step-by-step (9 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost D_service raw output over 7 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 0.9940, quality 0.457)	`market_value_eur=4600000` `quality_factor=0.4572` `boost_factor=0.994` `reference_quality=0.505` `history_matches=7` `history_shrink=0.4167`	-
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` `duel_context={'opponent_faced': 'Abdessamad Ezzalzouli', 'matchup_zone': 'Striker vs Two Centre Backs — 1v2 Aerial/Physical Battle'}`	-
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 2 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=2`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.57 (squad xG sum 0.9357, xG scale 1.6822, xA scale 1.6822, team xA 1.04 = 0.66 x lambda)	`scoreline_reset=true` `team_lambda=1.574` `squad_xg_sum=0.9357` `squad_xa_sum=0.6189` `team_expected_assists=1.0411` `team_xa_to_xg_ratio=0.6615`	expected_goals: 0.0199 -> 0.0335 expected_assists: 0.1244 -> 0.2092
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 11 by projected rating, fifa_overall 80.0)	`match_squad_rank=11` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv_filled` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 2 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=2` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.0329` `goal_rate=0.0` `scorer_weight=0.0268`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=1` `assist_probability=0.1888` `assist_rate=0.294` `provider_weight=0.2615`	-

Pedro Bigas D 81 0 0.03 0.02 48% 0 0.3 6.8 44.6 30 FBref Prior D_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost D_service raw output over 7 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 0.9973, quality 0.437)	`market_value_eur=310000` `quality_factor=0.4374` `boost_factor=0.9973` `reference_quality=0.505` `history_matches=33` `history_shrink=0.1316`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0046)	`blend_weight=0.0046` `history_matches=33` `season_overlap_discount=1.0` `per90_to_match_factor=0.9043` `prior_reliability=0.0533` `competition=La Liga`	-
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` `duel_context={'opponent_faced': 'Isco', 'matchup_zone': 'Left Wing vs Right Back — Flank Battle'}`	-
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 2 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=2`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.57 (squad xG sum 0.9357, xG scale 1.6822, xA scale 1.6822, team xA 1.04 = 0.66 x lambda)	`scoreline_reset=true` `team_lambda=1.574` `squad_xg_sum=0.9357` `squad_xa_sum=0.6189` `team_expected_assists=1.0411` `team_xa_to_xg_ratio=0.6615`	expected_goals: 0.0163 -> 0.0274 expected_assists: 0.0373 -> 0.0627
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 10 by projected rating, fifa_overall 80.0)	`match_squad_rank=10` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv_filled` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 2 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=2` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.027` `goal_rate=0.0587` `scorer_weight=0.0343`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.0608` `assist_rate=0.1275` `provider_weight=0.0784`	-

Marc Aguado M 69 0 0.06 0.01 487% 0 0.2 6.8 56.6 30 FBref Prior M_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost M_service raw output over 8 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 0.9981, quality 0.446)	`market_value_eur=3700000` `quality_factor=0.4461` `boost_factor=0.9981` `reference_quality=0.505` `history_matches=41` `history_shrink=0.1087`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0104)	`blend_weight=0.0104` `history_matches=41` `season_overlap_discount=1.0` `per90_to_match_factor=0.7677` `prior_reliability=0.1467` `competition=La Liga`	interceptionWon: 1.0397 -> 1.0289
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` duel_context={'opponent_faced': 'Marc Roca', 'matchup_zone': 'Central Mid vs Defensive Mid — Pressing/Creation Duel', 'duel_win_rate': 0.421, 'pressing_faced': 'LOW', 'numerical_situation': '1v1', 'xg_multiplier': 0.9579, 'skill_edge': -0.0046, 'height_edge': -0.024, 'attacker_skill': 0.4512, 'defender_skill': 0.4615, 'attacker_role': 'CM', 'attacker_mv_price_index': 0.9, 'defender_adjusted_parts': [{'player': 'Marc Roca', 'market_value': 4200000, 'role': 'CDM', 'price_index': 0.85, 'adjusted_quality': 0.4615}], 'xg_override': 0.0331}	aerialWon: 0.1803 -> 1.0 totalShots: 0.5061 -> 0.4848 shotOffTarget: 0.3835 -> 0.3674 onTargetScoringAttempt: 0.1226 -> 0.1174
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 2 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=2`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.57 (squad xG sum 0.9357, xG scale 1.6822, xA scale 1.6822, team xA 1.04 = 0.66 x lambda)	`scoreline_reset=true` `team_lambda=1.574` `squad_xg_sum=0.9357` `squad_xa_sum=0.6189` `team_expected_assists=1.0411` `team_xa_to_xg_ratio=0.6615`	expected_goals: 0.0331 -> 0.0557 expected_assists: 0.027 -> 0.0454
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 7 by projected rating, fifa_overall 74.0)	`match_squad_rank=7` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 2 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=2` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.0542` `goal_rate=0.0477` `scorer_weight=0.0655`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.0444` `assist_rate=0.0489` `provider_weight=0.0568`	-

Gonzalo Villar M 76 0 0.12 0.03 390% 1 0.4 6.9 52.3 29 FBref Prior M_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost M_service raw output over 8 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 0.9962, quality 0.437)	`market_value_eur=1700000` `quality_factor=0.4374` `boost_factor=0.9962` `reference_quality=0.505` `history_matches=22` `history_shrink=0.1852`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0241)	`blend_weight=0.0241` `history_matches=22` `season_overlap_discount=1.0` `per90_to_match_factor=0.8427` `prior_reliability=0.2` `competition=La Liga`	wasFouled: 1.93 -> 1.9782 totalCross: 1.0878 -> 1.1225 fouls: 1.3978 -> 1.4182 onTargetScoringAttempt: 0.2613 -> 0.2753 totalShots: 0.8491 -> 0.8556
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` duel_context={'opponent_faced': 'Marc Roca', 'matchup_zone': 'Central Mid vs Defensive Mid — Pressing/Creation Duel', 'duel_win_rate': 0.427, 'pressing_faced': 'LOW', 'numerical_situation': '1v1', 'xg_multiplier': 0.9665, 'skill_edge': -0.0108, 'height_edge': -0.012, 'attacker_skill': 0.4374, 'defender_skill': 0.4615, 'attacker_role': 'CM', 'attacker_mv_price_index': 0.9, 'defender_adjusted_parts': [{'player': 'Marc Roca', 'market_value': 4200000, 'role': 'CDM', 'price_index': 0.85, 'adjusted_quality': 0.4615}], 'xg_override': 0.0737}	aerialWon: 0.6516 -> 1.1 totalShots: 0.8631 -> 0.8342 shotOffTarget: 0.5878 -> 0.5681 onTargetScoringAttempt: 0.2753 -> 0.2661
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 2 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=2`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.57 (squad xG sum 0.9357, xG scale 1.6822, xA scale 1.6822, team xA 1.04 = 0.66 x lambda)	`scoreline_reset=true` `team_lambda=1.574` `squad_xg_sum=0.9357` `squad_xa_sum=0.6189` `team_expected_assists=1.0411` `team_xa_to_xg_ratio=0.6615`	expected_goals: 0.0737 -> 0.124 expected_assists: 0.0909 -> 0.1529
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 3 by projected rating, fifa_overall 72.0)	`match_squad_rank=3` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 2 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=2` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.1166` `goal_rate=0.067` `scorer_weight=0.1173`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.1418` `assist_rate=0.102` `provider_weight=0.1649`	-

Javi Morcillo M 87 0 0.06 0.02 281% 1 0.2 6.7 41.8 21 FBref Prior M_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost M_service raw output over 8 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 0.9944, quality 0.437)	`market_value_eur=2100000` `quality_factor=0.4374` `boost_factor=0.9944` `reference_quality=0.505` `history_matches=13` `history_shrink=0.2778`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0048)	`blend_weight=0.0048` `history_matches=13` `season_overlap_discount=1.0` `per90_to_match_factor=0.9707` `prior_reliability=0.0267` `competition=La Liga`	fouls: 0.725 -> 0.7332 wonTackle: 1.1344 -> 1.1406 interceptionWon: 1.0719 -> 1.0667
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` `duel_context={'opponent_faced': 'Marc Roca', 'matchup_zone': 'CDM vs CDM', 'duel_win_rate': 0.405, 'pressing_faced': 'LOW', 'numerical_situation': '1v1', 'xg_multiplier': 0.9782, 'skill_edge': -0.0108, 'height_edge': -0.004, 'attacker_skill': 0.4374, 'defender_skill': 0.4615, 'attacker_role': 'CDM', 'attacker_mv_price_index': 0.85, 'defender_adjusted_parts': [{'player': 'Marc Roca', 'market_value': 4200000, 'role': 'CDM', 'price_index': 0.85, 'adjusted_quality': 0.4615}], 'xg_override': 0.0385}`	aerialWon: 0.6603 -> 1.1 totalShots: 0.5542 -> 0.5422 shotOffTarget: 0.424 -> 0.4148
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 2 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=2`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.57 (squad xG sum 0.9357, xG scale 1.6822, xA scale 1.6822, team xA 1.04 = 0.66 x lambda)	`scoreline_reset=true` `team_lambda=1.574` `squad_xg_sum=0.9357` `squad_xa_sum=0.6189` `team_expected_assists=1.0411` `team_xa_to_xg_ratio=0.6615`	expected_goals: 0.0385 -> 0.0648 expected_assists: 0.0258 -> 0.0434
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 14 by projected rating, fifa_overall 79.0)	`match_squad_rank=14` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv_filled` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 2 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=2` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.0627` `goal_rate=0.0519` `scorer_weight=0.0746`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.0425` `assist_rate=0.044` `provider_weight=0.0538`	-

Germán Valera F 86 0 0.22 0.05 393% 1 0.5 7.0 58.5 42 XGBoost F_service

Step-by-step (9 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost F_service raw output over 7 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 0.9992, quality 0.477)	`market_value_eur=6200000` `quality_factor=0.4769` `boost_factor=0.9992` `reference_quality=0.505` `history_matches=46` `history_shrink=0.098`	-
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` duel_context={'opponent_faced': 'Junior Firpo', 'matchup_zone': 'Left Wing vs Right Back — Flank Battle', 'duel_win_rate': 0.431, 'pressing_faced': 'LOW', 'numerical_situation': '1v1', 'xg_multiplier': 0.9286, 'skill_edge': 0.0074, 'height_edge': -0.056, 'attacker_skill': 0.4769, 'defender_skill': 0.4604, 'attacker_role': 'LW', 'attacker_mv_price_index': 1.0, 'defender_adjusted_parts': [{'player': 'Junior Firpo', 'market_value': 3400000, 'role': 'RB', 'price_index': 0.7, 'adjusted_quality': 0.4604}], 'xg_override': 0.1334}	totalShots: 1.1989 -> 1.1133 shotOffTarget: 0.7187 -> 0.6674 onTargetScoringAttempt: 0.4801 -> 0.4458 goals: 0.2152 -> 0.1998 expectedGoals: 0.1437 -> 0.1334 expectedGoalsOnTarget: 0.0807 -> 0.0749
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 2 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=2`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.57 (squad xG sum 0.9357, xG scale 1.6822, xA scale 1.6822, team xA 1.04 = 0.66 x lambda)	`scoreline_reset=true` `team_lambda=1.574` `squad_xg_sum=0.9357` `squad_xa_sum=0.6189` `team_expected_assists=1.0411` `team_xa_to_xg_ratio=0.6615`	expected_goals: 0.1334 -> 0.2244 expected_assists: 0.1162 -> 0.1955
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 2 by projected rating, fifa_overall 82.0)	`match_squad_rank=2` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv_filled` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 2 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=2` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.201` `goal_rate=0.1998` `scorer_weight=0.2805`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.1776` `assist_rate=0.0349` `provider_weight=0.1564`	-

Fernando Niño F 73 1 0.57 0.32 81% 2 1.2 6.8 54.1 13 FBref Prior F_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost F_service raw output over 7 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 0.9940, quality 0.445)	`market_value_eur=3600000` `quality_factor=0.4449` `boost_factor=0.994` `reference_quality=0.505` `history_matches=10` `history_shrink=0.3333`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0361)	`blend_weight=0.0361` `history_matches=10` `season_overlap_discount=1.0` `per90_to_match_factor=0.8061` `prior_reliability=0.1667` `competition=La Liga`	totalShots: 2.1609 -> 2.1291 fouls: 1.0612 -> 1.0345 totalOffside: 0.3119 -> 0.3006 wasFouled: 1.354 -> 1.3633 wonTackle: 0.8152 -> 0.8207 interceptionWon: 0.1843 -> 0.1893
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` duel_context={'opponent_faced': 'Natan, Diego Llorente', 'matchup_zone': 'Striker vs Two Centre Backs — 1v2 Aerial/Physical Battle', 'duel_win_rate': 0.18, 'pressing_faced': 'LOW', 'numerical_situation': '1v2', 'xg_multiplier': 0.9056, 'skill_edge': -0.0782, 'height_edge': 0.014, 'attacker_skill': 0.4449, 'defender_skill': 0.6187, 'attacker_role': 'ST', 'attacker_mv_price_index': 1.0, 'defender_adjusted_parts': [{'player': 'Natan', 'market_value': 32000000, 'role': 'CB', 'price_index': 0.75, 'adjusted_quality': 0.7941}, {'player': 'Diego Llorente', 'market_value': 2600000, 'role': 'CB', 'price_index': 0.75, 'adjusted_quality': 0.4432}], 'xg_override': 0.3386}	totalShots: 2.1576 -> 1.9539 shotOffTarget: 1.1103 -> 1.0055 onTargetScoringAttempt: 1.0473 -> 0.9484 expectedGoalsOnTarget: 0.6831 -> 0.6186 goals: 0.4092 -> 0.3706 expectedGoals: 0.3739 -> 0.3386 bigChanceMissed: 0.1567 -> 0.1419
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 2 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=2`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.57 (squad xG sum 0.9357, xG scale 1.6822, xA scale 1.6822, team xA 1.04 = 0.66 x lambda)	`scoreline_reset=true` `team_lambda=1.574` `squad_xg_sum=0.9357` `squad_xa_sum=0.6189` `team_expected_assists=1.0411` `team_xa_to_xg_ratio=0.6615`	expected_goals: 0.3386 -> 0.5695 expected_assists: 0.0385 -> 0.0648
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 9 by projected rating, fifa_overall 81.0)	`match_squad_rank=9` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv_filled` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 2 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=2` `allocation_applied=true` `goals_allocated=1` `goal_probability=0.4342` `goal_rate=0.3706` `scorer_weight=0.6164`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.0627` `assist_rate=0.0` `provider_weight=0.0518`	-

Ali Houary F 67 1 0.23 0.14 67% 2 0.8 6.9 30.3 12 FBref Prior F_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost F_service raw output over 7 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 0.9899, quality 0.437)	`market_value_eur=620000` `quality_factor=0.4374` `boost_factor=0.9899` `reference_quality=0.505` `history_matches=5` `history_shrink=0.5`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0455)	`blend_weight=0.0455` `history_matches=5` `season_overlap_discount=1.0` `per90_to_match_factor=0.7434` `prior_reliability=0.14` `competition=La Liga`	fouls: 1.5921 -> 1.5519 totalShots: 1.4119 -> 1.4285 totalCross: 0.4197 -> 0.4328 wasFouled: 0.5947 -> 0.5837 goals: 0.2361 -> 0.2254 wonTackle: 0.4961 -> 0.4896 onTargetScoringAttempt: 0.5839 -> 0.5898 totalOffside: 0.2269 -> 0.2327
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` duel_context={'opponent_faced': 'Fran García', 'matchup_zone': 'Right Wing vs Left Back — Flank Battle', 'duel_win_rate': 0.474, 'pressing_faced': 'LOW', 'numerical_situation': '1v1', 'xg_multiplier': 0.9919, 'skill_edge': -0.0375, 'height_edge': 0.032, 'attacker_skill': 0.4374, 'defender_skill': 0.5208, 'attacker_role': 'RW', 'attacker_mv_price_index': 1.0, 'defender_adjusted_parts': [{'player': 'Fran García', 'market_value': 6900000, 'role': 'LB', 'price_index': 0.7, 'adjusted_quality': 0.5208}], 'xg_override': 0.1363}	totalShots: 1.4285 -> 1.4169 shotOffTarget: 0.828 -> 0.8213
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 2 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=2`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.57 (squad xG sum 0.9357, xG scale 1.6822, xA scale 1.6822, team xA 1.04 = 0.66 x lambda)	`scoreline_reset=true` `team_lambda=1.574` `squad_xg_sum=0.9357` `squad_xa_sum=0.6189` `team_expected_assists=1.0411` `team_xa_to_xg_ratio=0.6615`	expected_goals: 0.1363 -> 0.2293 expected_assists: 0.0701 -> 0.1179
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 6 by projected rating, fifa_overall 65.0)	`match_squad_rank=6` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 2 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=2` `allocation_applied=true` `goals_allocated=1` `goal_probability=0.2049` `goal_rate=0.2236` `scorer_weight=0.2866`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.1112` `assist_rate=0.0` `provider_weight=0.0943`	-

Team totals 2 1.57 11 6.9

Bench Candidates (7 modeled substitutes with empirical playing time)

Player	Pos	Exp Min	App Prob	Substitute Modeling Details
Alejandro Iturbe	G	4.4m	10.9%	Market-value substitute prior
Kevin Lomónaco	D	13.7m	30.4%	Market-value substitute prior
Rubén Sánchez	D	13.6m	29.9%	Market-value substitute prior
Martim Neto	M	24.8m	50.0%	Market-value substitute prior
Federico Redondo	M	22.0m	30.7%	Market-value substitute prior
Facundo Buonanotte	F	26.4m	53.4%	Market-value substitute prior

Real Betis — projected line vs season baseline (4-2-3-1)

Player Pos Min Proj G Proj xG Season xG/g Proj Sh Season Sh/g Proj Rtg Season Rtg Rec Model / FBref Prior

Álvaro Valles G 90 0 0.00 - n/a 0.0 7.2 63.1 34 FBref Prior G_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost G_service raw output over 3 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0003, quality 0.514)	`market_value_eur=9300000` `quality_factor=0.5142` `boost_factor=1.0003` `reference_quality=0.505` `history_matches=39` `history_shrink=0.1136`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0148)	`blend_weight=0.0148` `history_matches=39` `season_overlap_discount=1.0` `per90_to_match_factor=1.0` `prior_reliability=0.2` `competition=La Liga`	-
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` `duel_context={'opponent_faced': 'Fernando Niño, Ali Houary', 'matchup_zone': 'Goalkeeper shot-stopping context', 'opponent_lambda': 1.1709}`	-
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 1 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=1`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.41 (squad xG sum 1.0814, xG scale 1.3039, xA scale 1.1830, team xA 1.13 = 0.80 x lambda)	`scoreline_reset=true` `team_lambda=1.41` `squad_xg_sum=1.0814` `squad_xa_sum=0.9535` `team_expected_assists=1.128` `team_xa_to_xg_ratio=0.8`	shots_faced: 11.1 -> 15.0 expected_saves: 2.71 -> 3.68 high_pressure_saves: 0.94 -> 1.26
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 1 by projected rating, fifa_overall 77.0)	`match_squad_rank=1` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 1 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=1` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.0` `scorer_weight=0.0` `squad_weight_total=1.5061`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.0` `provider_weight=0.0` `squad_weight_total=1.0036`	-

Natan D 89 0 0.08 0.02 245% 1 0.2 6.9 51.7 33 FBref Prior D_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost D_service raw output over 7 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0072, quality 0.732)	`market_value_eur=32000000` `quality_factor=0.732` `boost_factor=1.0072` `reference_quality=0.505` `history_matches=42` `history_shrink=0.1064`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0092)	`blend_weight=0.0092` `history_matches=42` `season_overlap_discount=1.0` `per90_to_match_factor=0.9929` `prior_reliability=0.1333` `competition=La Liga`	wonTackle: 0.6913 -> 0.7032 interceptionWon: 0.8458 -> 0.838
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` `duel_context={'opponent_faced': 'Fernando Niño', 'matchup_zone': 'Striker vs Two Centre Backs — 1v2 Aerial/Physical Battle'}`	-
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 1 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=1`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.41 (squad xG sum 1.0814, xG scale 1.3039, xA scale 1.1830, team xA 1.13 = 0.80 x lambda)	`scoreline_reset=true` `team_lambda=1.41` `squad_xg_sum=1.0814` `squad_xa_sum=0.9535` `team_expected_assists=1.128` `team_xa_to_xg_ratio=0.8`	expected_goals: 0.0619 -> 0.0807
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 7 by projected rating, fifa_overall 79.0)	`match_squad_rank=7` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 1 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=1` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.0775` `goal_rate=0.0667` `scorer_weight=0.0842`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.013` `assist_rate=0.0` `provider_weight=0.0107`	-

Diego Llorente D 81 0 0.02 0.01 67% 0 0.1 6.8 49.8 28 FBref Prior D_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost D_service raw output over 7 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 0.9964, quality 0.437)	`market_value_eur=2600000` `quality_factor=0.4374` `boost_factor=0.9964` `reference_quality=0.505` `history_matches=23` `history_shrink=0.1786`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0077)	`blend_weight=0.0077` `history_matches=23` `season_overlap_discount=1.0` `per90_to_match_factor=0.8962` `prior_reliability=0.0667` `competition=La Liga`	interceptionWon: 1.2469 -> 1.2719 wonTackle: 0.7347 -> 0.729
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` `duel_context={'opponent_faced': 'Fernando Niño', 'matchup_zone': 'Striker vs Two Centre Backs — 1v2 Aerial/Physical Battle'}`	-
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 1 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=1`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.41 (squad xG sum 1.0814, xG scale 1.3039, xA scale 1.1830, team xA 1.13 = 0.80 x lambda)	`scoreline_reset=true` `team_lambda=1.41` `squad_xg_sum=1.0814` `squad_xa_sum=0.9535` `team_expected_assists=1.128` `team_xa_to_xg_ratio=0.8`	-
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 11 by projected rating, fifa_overall 79.0)	`match_squad_rank=11` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 1 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=1` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.0174` `goal_rate=0.0` `scorer_weight=0.0141`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.004` `assist_rate=0.0` `provider_weight=0.0037`	-

Fran García D 89 0 0.02 0.02 ok 1 0.2 6.7 33.1 33 FBref Prior D_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost D_service raw output over 7 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 0.9989, quality 0.485)	`market_value_eur=6900000` `quality_factor=0.4854` `boost_factor=0.9989` `reference_quality=0.505` `history_matches=22` `history_shrink=0.1852`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0193)	`blend_weight=0.0193` `history_matches=22` `season_overlap_discount=1.0` `per90_to_match_factor=0.9883` `prior_reliability=0.16` `competition=La Liga`	totalCross: 2.866 -> 2.898 totalShots: 0.5249 -> 0.5148 goalAssist: 0.0225 -> 0.03 wasFouled: 1.1303 -> 1.1244
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=true`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` `duel_context={'opponent_faced': 'Ali Houary', 'matchup_zone': 'Right Wing vs Left Back — Flank Battle'}`	-
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 1 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=1`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.41 (squad xG sum 1.0814, xG scale 1.3039, xA scale 1.1830, team xA 1.13 = 0.80 x lambda)	`scoreline_reset=true` `team_lambda=1.41` `squad_xg_sum=1.0814` `squad_xa_sum=0.9535` `team_expected_assists=1.128` `team_xa_to_xg_ratio=0.8`	expected_assists: 0.0472 -> 0.0558
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 17 by projected rating, fifa_overall 77.0)	`match_squad_rank=17` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 1 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=1` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.0197` `goal_rate=0.0` `scorer_weight=0.0159`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.0543` `assist_rate=0.03` `provider_weight=0.0459`	-

Junior Firpo D 79 0 0.03 0.00 888% 0 0.0 6.8 22.1 21 FBref Prior D_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost D_service raw output over 7 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 0.9951, quality 0.442)	`market_value_eur=3400000` `quality_factor=0.4424` `boost_factor=0.9951` `reference_quality=0.505` `history_matches=14` `history_shrink=0.2632`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0068)	`blend_weight=0.0068` `history_matches=14` `season_overlap_discount=1.0` `per90_to_match_factor=0.8747` `prior_reliability=0.04` `competition=La Liga`	totalCross: 1.4357 -> 1.4458 wonTackle: 1.2813 -> 1.2725 interceptionWon: 1.0311 -> 1.024
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` `duel_context={'opponent_faced': 'Germán Valera', 'matchup_zone': 'Left Wing vs Right Back — Flank Battle'}`	-
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 1 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=1`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.41 (squad xG sum 1.0814, xG scale 1.3039, xA scale 1.1830, team xA 1.13 = 0.80 x lambda)	`scoreline_reset=true` `team_lambda=1.41` `squad_xg_sum=1.0814` `squad_xa_sum=0.9535` `team_expected_assists=1.128` `team_xa_to_xg_ratio=0.8`	expected_goals: 0.0194 -> 0.0253 expected_assists: 0.054 -> 0.0639
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 15 by projected rating, fifa_overall 74.0)	`match_squad_rank=15` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 1 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=1` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.025` `goal_rate=0.0` `scorer_weight=0.0202`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.0619` `assist_rate=0.034` `provider_weight=0.0516`	-

Marc Roca M 86 0 0.06 0.02 153% 1 0.5 7.1 51.3 34 FBref Prior M_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost M_service raw output over 8 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 0.9981, quality 0.452)	`market_value_eur=4200000` `quality_factor=0.4523` `boost_factor=0.9981` `reference_quality=0.505` `history_matches=37` `history_shrink=0.119`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0088)	`blend_weight=0.0088` `history_matches=37` `season_overlap_discount=1.0` `per90_to_match_factor=0.9579` `prior_reliability=0.1133` `competition=La Liga`	totalShots: 0.5759 -> 0.5854 wonTackle: 0.8983 -> 0.9052 fouls: 1.6966 -> 1.6916
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` duel_context={'opponent_faced': 'Marc Aguado', 'matchup_zone': 'Holding Mid vs Central Mid — Midfield Anchor Duel', 'duel_win_rate': 0.429, 'pressing_faced': 'LOW', 'numerical_situation': '1v1', 'xg_multiplier': 1.0421, 'skill_edge': 0.0046, 'height_edge': 0.024, 'attacker_skill': 0.4615, 'defender_skill': 0.4512, 'attacker_role': 'CDM', 'attacker_mv_price_index': 0.85, 'defender_adjusted_parts': [{'player': 'Marc Aguado', 'market_value': 3700000, 'role': 'CM', 'price_index': 0.9, 'adjusted_quality': 0.4512}], 'xg_override': 0.0447}	aerialWon: 0.5233 -> 1.1 totalShots: 0.5854 -> 0.61 shotOffTarget: 0.4365 -> 0.4549 onTargetScoringAttempt: 0.1382 -> 0.144
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 1 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=1`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.41 (squad xG sum 1.0814, xG scale 1.3039, xA scale 1.1830, team xA 1.13 = 0.80 x lambda)	`scoreline_reset=true` `team_lambda=1.41` `squad_xg_sum=1.0814` `squad_xa_sum=0.9535` `team_expected_assists=1.128` `team_xa_to_xg_ratio=0.8`	expected_goals: 0.0447 -> 0.0583 expected_assists: 0.071 -> 0.084
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 5 by projected rating, fifa_overall 78.0)	`match_squad_rank=5` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 1 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=1` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.0566` `goal_rate=0.0469` `scorer_weight=0.0597`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.0806` `assist_rate=0.0653` `provider_weight=0.08`	-

Isco M 68 0 0.20 0.19 ok 2 1.2 7.2 58.9 20 FBref Prior M_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost M_service raw output over 8 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 0.9961, quality 0.445)	`market_value_eur=3600000` `quality_factor=0.4449` `boost_factor=0.9961` `reference_quality=0.505` `history_matches=18` `history_shrink=0.2174`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0123)	`blend_weight=0.0123` `history_matches=18` `season_overlap_discount=1.0` `per90_to_match_factor=0.7518` `prior_reliability=0.0867` `competition=La Liga`	fouls: 0.6204 -> 0.6411 onTargetScoringAttempt: 0.6292 -> 0.642 totalCross: 3.1144 -> 3.1258
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` duel_context={'opponent_faced': 'Pedro Bigas', 'matchup_zone': 'Left Wing vs Right Back — Flank Battle', 'duel_win_rate': 0.467, 'pressing_faced': 'LOW', 'numerical_situation': '1v1', 'xg_multiplier': 0.9814, 'skill_edge': 0.0034, 'height_edge': -0.016, 'attacker_skill': 0.4449, 'defender_skill': 0.4374, 'attacker_role': 'LW', 'attacker_mv_price_index': 1.0, 'defender_adjusted_parts': [{'player': 'Pedro Bigas', 'market_value': 310000, 'role': 'RB', 'price_index': 0.7, 'adjusted_quality': 0.4374}], 'xg_override': 0.153}	aerialWon: 0.0 -> 0.9 totalShots: 1.8279 -> 1.7939 shotOffTarget: 1.1859 -> 1.1638 onTargetScoringAttempt: 0.642 -> 0.6301 expectedGoalsOnTarget: 0.4169 -> 0.4091
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 1 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=1`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.41 (squad xG sum 1.0814, xG scale 1.3039, xA scale 1.1830, team xA 1.13 = 0.80 x lambda)	`scoreline_reset=true` `team_lambda=1.41` `squad_xg_sum=1.0814` `squad_xa_sum=0.9535` `team_expected_assists=1.128` `team_xa_to_xg_ratio=0.8`	expected_goals: 0.153 -> 0.1995 expected_assists: 0.1915 -> 0.2267
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 3 by projected rating, fifa_overall 83.0)	`match_squad_rank=3` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 1 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=1` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.1809` `goal_rate=0.1486` `scorer_weight=0.1952`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=1` `assist_probability=0.2028` `assist_rate=0.1938` `provider_weight=0.2289`	-

Pablo Fornals M 80 0 0.21 0.09 149% 2 1.1 7.2 62.4 34 FBref Prior M_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost M_service raw output over 8 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0000, quality 0.503)	`market_value_eur=8400000` `quality_factor=0.5035` `boost_factor=1.0` `reference_quality=0.505` `history_matches=47` `history_shrink=0.0962`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0092)	`blend_weight=0.0092` `history_matches=47` `season_overlap_discount=1.0` `per90_to_match_factor=0.8914` `prior_reliability=0.1467` `competition=La Liga`	totalShots: 1.5056 -> 1.5136 wasFouled: 0.8332 -> 0.8256 interceptionWon: 0.8785 -> 0.8853
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` duel_context={'opponent_faced': 'Víctor Chust', 'matchup_zone': 'Right Wing vs Left Back — Flank Battle', 'duel_win_rate': 0.489, 'pressing_faced': 'LOW', 'numerical_situation': '1v1', 'xg_multiplier': 1.0131, 'skill_edge': 0.0249, 'height_edge': -0.016, 'attacker_skill': 0.5035, 'defender_skill': 0.4481, 'attacker_role': 'RW', 'attacker_mv_price_index': 1.0, 'defender_adjusted_parts': [{'player': 'Víctor Chust', 'market_value': 2700000, 'role': 'LB', 'price_index': 0.7, 'adjusted_quality': 0.4481}], 'xg_override': 0.1632}	aerialWon: 0.2656 -> 1.0 totalShots: 1.5136 -> 1.5334 shotOffTarget: 1.0467 -> 1.0604 onTargetScoringAttempt: 0.4547 -> 0.4607
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 1 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=1`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.41 (squad xG sum 1.0814, xG scale 1.3039, xA scale 1.1830, team xA 1.13 = 0.80 x lambda)	`scoreline_reset=true` `team_lambda=1.41` `squad_xg_sum=1.0814` `squad_xa_sum=0.9535` `team_expected_assists=1.128` `team_xa_to_xg_ratio=0.8`	expected_goals: 0.1632 -> 0.2128 expected_assists: 0.1885 -> 0.223
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 4 by projected rating, fifa_overall 85.0)	`match_squad_rank=4` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv_filled` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 1 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=1` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.1917` `goal_rate=0.1435` `scorer_weight=0.1931`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.1999` `assist_rate=0.1326` `provider_weight=0.1784`	-

Facundo Bernal M 75 0 0.07 0.02 299% 1 0.5 6.8 48.4 11 FBref Prior M_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost M_service raw output over 8 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0016, quality 0.519)	`market_value_eur=9700000` `quality_factor=0.5189` `boost_factor=1.0016` `reference_quality=0.505` `history_matches=8` `history_shrink=0.3846`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0400)	`blend_weight=0.04` `history_matches=8` `season_overlap_discount=1.0` `per90_to_match_factor=0.8366` `prior_reliability=0.16` `competition=La Liga`	wonTackle: 0.9687 -> 0.9439 fouls: 0.8657 -> 0.845 interceptionWon: 1.2349 -> 1.2552 totalShots: 0.7182 -> 0.7306 onTargetScoringAttempt: 0.1867 -> 0.193
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` duel_context={'opponent_faced': 'Marc Aguado', 'matchup_zone': 'Holding Mid vs Central Mid — Midfield Anchor Duel', 'duel_win_rate': 0.475, 'pressing_faced': 'LOW', 'numerical_situation': '1v1', 'xg_multiplier': 1.1109, 'skill_edge': 0.0395, 'height_edge': 0.036, 'attacker_skill': 0.5389, 'defender_skill': 0.4512, 'attacker_role': 'CDM', 'attacker_mv_price_index': 0.85, 'defender_adjusted_parts': [{'player': 'Marc Aguado', 'market_value': 3700000, 'role': 'CM', 'price_index': 0.9, 'adjusted_quality': 0.4512}], 'xg_override': 0.0545}	aerialWon: 0.3811 -> 1.2 totalShots: 0.7306 -> 0.8116 shotOffTarget: 0.5315 -> 0.5904 onTargetScoringAttempt: 0.193 -> 0.2144 goals: 0.0568 -> 0.0631 expectedGoals: 0.0491 -> 0.0545
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 1 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=1`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.41 (squad xG sum 1.0814, xG scale 1.3039, xA scale 1.1830, team xA 1.13 = 0.80 x lambda)	`scoreline_reset=true` `team_lambda=1.41` `squad_xg_sum=1.0814` `squad_xa_sum=0.9535` `team_expected_assists=1.128` `team_xa_to_xg_ratio=0.8`	expected_goals: 0.0545 -> 0.0711 expected_assists: 0.0278 -> 0.0329
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 12 by projected rating, fifa_overall 74.0)	`match_squad_rank=12` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 1 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=1` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.0686` `goal_rate=0.0631` `scorer_weight=0.077`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.0324` `assist_rate=0.0432` `provider_weight=0.0394`	-

Giovani Lo Celso M 68 0 0.12 0.04 182% 1 0.5 6.9 43.9 20 XGBoost M_service

Step-by-step (9 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost M_service raw output over 8 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 0.9990, quality 0.484)	`market_value_eur=6800000` `quality_factor=0.4842` `boost_factor=0.999` `reference_quality=0.505` `history_matches=27` `history_shrink=0.1562`	-
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` duel_context={'opponent_faced': 'Javi Morcillo', 'matchup_zone': 'Attacking Mid vs Single Holding Mid — Space Creator Duel', 'duel_win_rate': 0.459, 'pressing_faced': 'LOW', 'numerical_situation': '1v1', 'xg_multiplier': 0.9985, 'skill_edge': 0.023, 'height_edge': -0.024, 'attacker_skill': 0.4885, 'defender_skill': 0.4374, 'attacker_role': 'CAM', 'attacker_mv_price_index': 0.95, 'defender_adjusted_parts': [{'player': 'Javi Morcillo', 'market_value': 2100000, 'role': 'CDM', 'price_index': 0.85, 'adjusted_quality': 0.4374}], 'xg_override': 0.0904}	aerialWon: 0.3946 -> 1.0
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 1 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=1`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.41 (squad xG sum 1.0814, xG scale 1.3039, xA scale 1.1830, team xA 1.13 = 0.80 x lambda)	`scoreline_reset=true` `team_lambda=1.41` `squad_xg_sum=1.0814` `squad_xa_sum=0.9535` `team_expected_assists=1.128` `team_xa_to_xg_ratio=0.8`	expected_goals: 0.0904 -> 0.1179 expected_assists: 0.1259 -> 0.1489
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 6 by projected rating, fifa_overall 80.0)	`match_squad_rank=6` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 1 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=1` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.1112` `goal_rate=0.0817` `scorer_weight=0.1106`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.1383` `assist_rate=0.1031` `provider_weight=0.1296`	-

Abdessamad Ezzalzouli F 85 1 0.48 0.24 99% 3 2.0 7.2 62.1 33 XGBoost F_service

Step-by-step (9 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost F_service raw output over 7 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0104, quality 0.796)	`market_value_eur=43000000` `quality_factor=0.7957` `boost_factor=1.0104` `reference_quality=0.505` `history_matches=37` `history_shrink=0.119`	-
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` duel_context={'opponent_faced': 'Buba Sangaré, Roy Revivo', 'matchup_zone': 'Striker vs Two Centre Backs — 1v2 Aerial/Physical Battle', 'duel_win_rate': 0.317, 'pressing_faced': 'LOW', 'numerical_situation': '1v2', 'xg_multiplier': 1.1716, 'skill_edge': 0.1427, 'height_edge': -0.026, 'attacker_skill': 0.7957, 'defender_skill': 0.4785, 'attacker_role': 'ST', 'attacker_mv_price_index': 1.0, 'defender_adjusted_parts': [{'player': 'Buba Sangaré', 'market_value': 4900000, 'role': 'CB', 'price_index': 0.75, 'adjusted_quality': 0.4809}, {'player': 'Roy Revivo', 'market_value': 4600000, 'role': 'CB', 'price_index': 0.75, 'adjusted_quality': 0.4761}], 'xg_override': 0.3716}	totalShots: 2.6976 -> 3.1605 shotOffTarget: 1.5849 -> 1.8569 onTargetScoringAttempt: 1.1127 -> 1.3036 goals: 0.4456 -> 0.5221 expectedGoalsOnTarget: 0.3696 -> 0.433 expectedGoals: 0.3172 -> 0.3716 bigChanceMissed: 0.3074 -> 0.3601
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 1 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=1`	goals: 1.0 -> 0
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.41 (squad xG sum 1.0814, xG scale 1.3039, xA scale 1.1830, team xA 1.13 = 0.80 x lambda)	`scoreline_reset=true` `team_lambda=1.41` `squad_xg_sum=1.0814` `squad_xa_sum=0.9535` `team_expected_assists=1.128` `team_xa_to_xg_ratio=0.8`	expected_goals: 0.3716 -> 0.4845 expected_assists: 0.184 -> 0.2177
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 2 by projected rating, fifa_overall 86.0)	`match_squad_rank=2` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv_filled` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 1 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=1` `allocation_applied=true` `goals_allocated=1` `goal_probability=0.384` `goal_rate=0.5221` `scorer_weight=0.6056`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.1956` `assist_rate=0.0808` `provider_weight=0.1742`	-

Team totals 1 1.41 14 7.0

Bench Candidates (7 modeled substitutes with empirical playing time)

Player	Pos	Exp Min	App Prob	Substitute Modeling Details
Diego Conde	G	4.4m	10.9%	Market-value substitute prior
Valentín Gómez	D	14.0m	22.5%	Market-value substitute prior
Angel Ortiz	D	13.8m	22.1%	Market-value substitute prior
Nelson Deossa	M	19.9m	57.0%	Market-value substitute prior
Dani Ceballos	M	22.1m	50.6%	Market-value substitute prior
Antony	F	26.6m	36.0%	Market-value substitute prior

Antagonist / tactical summary

Midfield is evenly contested — Marc Aguado vs Facundo Bernal will be the pivotal duel. Total antagonist xG — Home: 0.75, Away: 0.88. Home danger zone: Striker vs Two Centre Backs — 1v2 Aerial/Physical Battle. Away danger zone: Striker vs Two Centre Backs — 1v2 Aerial/Physical Battle.

Midfield — possession 48.9% / 51.1%, dominance ratio 0.45, key duel: Marc Aguado vs Facundo Bernal

Germán Valera xG 0.13Ali Houary xG 0.14Fernando Niño xG 0.34Isco xG 0.15Pablo Fornals xG 0.16Giovani Lo Celso xG 0.09

Summary

Antagonist matchup analysis for Elche vs Real Betis. Most likely scoreline: 2-1. Lambdas derived from 6 home and 6 away individual player duels. Head-to-head: Elche 2W-2D-6L vs Real Betis over 10 meetings.

Tactical parity between 4-3-3 and 4-2-3-1. Match outcome driven primarily by individual player quality. Midfield is evenly contested — Marc Aguado vs Facundo Bernal will be the pivotal duel. Total antagonist xG — Home: 0.75, Away: 0.88. Home danger zone: Striker vs Two Centre Backs — 1v2 Aerial/Physical Battle. Away danger zone: Striker vs Two Centre Backs — 1v2 Aerial/Physical Battle.

Ipswich vs Bournemouth

Premier League · 20261108 · 4-2-3-1 vs 4-2-3-1

Predicted 1-1Winner: DrawWinning H 1-0 (8.4%)Winning A 0-1 (8.6%)Confidence 0.64λ home/away 1.37 / 1.41Most-likely P 12.1%

Actual vs predicted: result not available yet (upcoming fixture).

XGBoost Reasoning & FBref Dataset Tracing

Antagonist + XGBoost match analysis for Ipswich vs Bournemouth. Player statistics regressed using position-specific XGBoost models (G, D, M, F) with 2026-2027 FBref priors. Micro-duels resolved across 4-2-3-1 vs 4-2-3-1 (duel lambdas: Home 1.210, Away 1.351). Final calibrated Poisson lambdas: Home 1.370, Away 1.411 (most likely score 1-1).

Home Starters 11/11 XGBoost (9 FBref) Away Starters 11/11 XGBoost (10 FBref) Bench Modeled H: 7 / A: 7 Duel λ (H/A) 1.21 / 1.35 Final λ (H/A) 1.37 / 1.41

FBref Team Priors: Ipswich Town (Gls/90: 2.0, GA/90: 2.0 → E[λ]: 1.379) vs Bournemouth (Gls/90: 1.0, GA/90: 1.0 → E[λ]: 1.379) | Blend Weight: 0.192

Modeling Pipeline Stages (5 layers)

1. XGBoost player regressions for volume & event rates with market-value quality scaling
2. Multi-source prior blending (FBref 2026-2027 tables + Sofascore season history)
3. Antagonist formation duel matrix resolving 1v1 and zonal matchups
4. Team-level prior calibration (FBref possession and attack/defense efficiencies)
5. Bivariate Poisson matrix with Dixon-Coles low-score correction

Standing & form context

Metric	Ipswich	Bournemouth
League position	13.00	7.00
Points per game	1.28	1.34
Attack index	1.11	1.01
Defense (concede) index	0.98	1.26
Form multiplier	1.00	1.00
Standing λ multiplier	1.03	1.10

Lambda decomposition (how the goal rates were built)

Stage	Ipswich λ	Bournemouth λ
FBref Squad Prior E[λ] (weight 0.192)	1.379	1.379
Pre-competitive-floor	1.210	1.351
Pre-calibration (Micro-duels)	1.210	1.351
Final λ	1.370	1.411

competition calibration scale 1.086competition observed total goals 2.781quality gap applied Truemismatch score pressure applied Falsecompetitive underdog floor applied Falsestage discount 1.0

Scoreline distribution & markets

Top 5 scorelines	P
1-1	12.1%
0-1	8.6%
1-2	8.5%
1-0	8.4%
2-1	8.2%

Home win	36.3%
Draw	25.6%
Away win	38.1%
Over 2.5	52.6%
BTTS yes	56.5%

Pregame fair line — favorite Bournemouth (38.1%), fair total 2.78, fair spread -0.04

Projected players vs season baseline

Proj = projected for this match · Season = this season's per-game average from match data · deviation badge compares projected xG to the player's season xG/game.

Ipswich — projected line vs season baseline (4-2-3-1)

Player Pos Min Proj G Proj xG Season xG/g Proj Sh Season Sh/g Proj Rtg Season Rtg Rec Model / FBref Prior

Christian Walton G 90 0 0.00 - n/a 0.0 7.0 22.9 14 FBref Prior G_service

Step-by-step (9 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost G_service raw output over 3 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 0.9899, quality 0.437)	`market_value_eur=2600000` `quality_factor=0.4374` `boost_factor=0.9899` `reference_quality=0.505` `history_matches=5` `history_shrink=0.5`	-
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` `duel_context={'opponent_faced': 'Rayan, Justin Kluivert', 'matchup_zone': 'Goalkeeper shot-stopping context', 'opponent_lambda': 1.3505}`	-
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 1 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=1`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.37 (squad xG sum 0.9028, xG scale 1.5176, xA scale 1.5176, team xA 0.98 = 0.72 x lambda)	`scoreline_reset=true` `team_lambda=1.37` `squad_xg_sum=0.9028` `squad_xa_sum=0.6471` `team_expected_assists=0.982` `team_xa_to_xg_ratio=0.7168`	shots_faced: 12.8 -> 13.4 expected_saves: 3.13 -> 3.28 high_pressure_saves: 1.08 -> 1.13
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 5 by projected rating, fifa_overall 73.0)	`match_squad_rank=5` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 1 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=1` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.0` `scorer_weight=0.0` `squad_weight_total=1.4124`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.0` `provider_weight=0.0` `squad_weight_total=1.0631`	-

Issa Diop D 90 0 0.06 0.03 100% 0 0.1 6.9 36.9 31 FBref Prior D_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost D_service raw output over 7 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0012, quality 0.526)	`market_value_eur=10300000` `quality_factor=0.526` `boost_factor=1.0012` `reference_quality=0.505` `history_matches=22` `history_shrink=0.1852`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0160)	`blend_weight=0.016` `history_matches=22` `season_overlap_discount=1.0` `per90_to_match_factor=0.998` `prior_reliability=0.1333` `competition=Premier League`	fouls: 0.8618 -> 0.872 interceptionWon: 1.0645 -> 1.0714 totalShots: 0.3714 -> 0.3654
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` `duel_context={'opponent_faced': 'Rayan', 'matchup_zone': 'Striker vs Two Centre Backs — 1v2 Aerial/Physical Battle'}`	-
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 1 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=1`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.37 (squad xG sum 0.9028, xG scale 1.5176, xA scale 1.5176, team xA 0.98 = 0.72 x lambda)	`scoreline_reset=true` `team_lambda=1.37` `squad_xg_sum=0.9028` `squad_xa_sum=0.6471` `team_expected_assists=0.982` `team_xa_to_xg_ratio=0.7168`	expected_goals: 0.0383 -> 0.0581 expected_assists: 0.0148 -> 0.0225
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 7 by projected rating, fifa_overall 75.0)	`match_squad_rank=7` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 1 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=1` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.0564` `goal_rate=0.0365` `scorer_weight=0.0569`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.0222` `assist_rate=0.0` `provider_weight=0.018`	-

Leif Davis D 90 0 0.21 0.12 79% 1 0.5 6.7 43.6 14 FBref Prior D_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost D_service raw output over 7 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0194, quality 0.686)	`market_value_eur=26000000` `quality_factor=0.6858` `boost_factor=1.0194` `reference_quality=0.505` `history_matches=9` `history_shrink=0.3571`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0309)	`blend_weight=0.0309` `history_matches=9` `season_overlap_discount=1.0` `per90_to_match_factor=1.0` `prior_reliability=0.1333` `competition=Premier League`	totalCross: 2.7087 -> 2.6713 onTargetScoringAttempt: 0.2159 -> 0.2402 wonTackle: 0.7174 -> 0.6952 fouls: 0.7127 -> 0.6906 totalShots: 0.5185 -> 0.5334 goals: 0.1521 -> 0.1629
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` `duel_context={'opponent_faced': 'Ryan Christie', 'matchup_zone': 'Right Wing vs Left Back — Flank Battle'}`	-
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 1 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=1`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.37 (squad xG sum 0.9028, xG scale 1.5176, xA scale 1.5176, team xA 0.98 = 0.72 x lambda)	`scoreline_reset=true` `team_lambda=1.37` `squad_xg_sum=0.9028` `squad_xa_sum=0.6471` `team_expected_assists=0.982` `team_xa_to_xg_ratio=0.7168`	expected_goals: 0.1386 -> 0.2103 expected_assists: 0.0644 -> 0.0977
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 12 by projected rating, fifa_overall 75.0)	`match_squad_rank=12` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 1 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=1` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.1897` `goal_rate=0.1629` `scorer_weight=0.2374`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.0931` `assist_rate=0.0856` `provider_weight=0.1158`	-

Dara O'Shea D 90 0 0.05 0.02 129% 0 0.1 6.8 43.3 14 FBref Prior D_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost D_service raw output over 7 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0145, quality 0.641)	`market_value_eur=21000000` `quality_factor=0.6408` `boost_factor=1.0145` `reference_quality=0.505` `history_matches=9` `history_shrink=0.3571`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0309)	`blend_weight=0.0309` `history_matches=9` `season_overlap_discount=1.0` `per90_to_match_factor=1.0` `prior_reliability=0.1333` `competition=Premier League`	fouls: 0.5768 -> 0.6054 interceptionWon: 1.2112 -> 1.1892 wonTackle: 0.9626 -> 0.9483 totalCross: 0.6867 -> 0.6964 wasFouled: 0.784 -> 0.7907
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` `duel_context={'opponent_faced': 'Justin Kluivert', 'matchup_zone': 'Left Wing vs Right Back — Flank Battle'}`	-
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 1 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=1`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.37 (squad xG sum 0.9028, xG scale 1.5176, xA scale 1.5176, team xA 0.98 = 0.72 x lambda)	`scoreline_reset=true` `team_lambda=1.37` `squad_xg_sum=0.9028` `squad_xa_sum=0.6471` `team_expected_assists=0.982` `team_xa_to_xg_ratio=0.7168`	expected_goals: 0.033 -> 0.0501 expected_assists: 0.0264 -> 0.0401
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 8 by projected rating, fifa_overall 75.0)	`match_squad_rank=8` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 1 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=1` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.0489` `goal_rate=0.1178` `scorer_weight=0.0626`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.0393` `assist_rate=0.0` `provider_weight=0.0321`	-

Jacob Greaves D 89 0 0.01 0.00 484% 0 0.1 6.8 22.4 14 FBref Prior D_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost D_service raw output over 7 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0132, quality 0.593)	`market_value_eur=16300000` `quality_factor=0.5932` `boost_factor=1.0132` `reference_quality=0.505` `history_matches=5` `history_shrink=0.5`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0433)	`blend_weight=0.0433` `history_matches=5` `season_overlap_discount=1.0` `per90_to_match_factor=0.9942` `prior_reliability=0.1333` `competition=Premier League`	wasFouled: 1.0914 -> 1.1303 interceptionWon: 0.8577 -> 0.8205 wonTackle: 0.8215 -> 0.8505 totalShots: 0.3316 -> 0.3172 fouls: 0.8568 -> 0.8628
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` `duel_context={'opponent_faced': 'Rayan', 'matchup_zone': 'Striker vs Two Centre Backs — 1v2 Aerial/Physical Battle'}`	-
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 1 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=1`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.37 (squad xG sum 0.9028, xG scale 1.5176, xA scale 1.5176, team xA 0.98 = 0.72 x lambda)	`scoreline_reset=true` `team_lambda=1.37` `squad_xg_sum=0.9028` `squad_xa_sum=0.6471` `team_expected_assists=0.982` `team_xa_to_xg_ratio=0.7168`	expected_assists: 0.0162 -> 0.0246
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 11 by projected rating, fifa_overall 74.0)	`match_squad_rank=11` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 1 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=1` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.0091` `goal_rate=0.0` `scorer_weight=0.0081`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.0243` `assist_rate=0.0` `provider_weight=0.0197`	-

Saša Lukić M 83 0 0.07 0.03 157% 1 0.4 7.0 47.3 24 FBref Prior M_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost M_service raw output over 8 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0004, quality 0.515)	`market_value_eur=9400000` `quality_factor=0.5154` `boost_factor=1.0004` `reference_quality=0.505` `history_matches=32` `history_shrink=0.1351`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0105)	`blend_weight=0.0105` `history_matches=32` `season_overlap_discount=1.0` `per90_to_match_factor=0.9228` `prior_reliability=0.12` `competition=Premier League`	wonTackle: 0.9597 -> 0.9658 wasFouled: 0.4797 -> 0.4746
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` duel_context={'opponent_faced': 'Adam Smith', 'matchup_zone': 'Left Wing vs Right Back — Flank Battle', 'duel_win_rate': 0.527, 'pressing_faced': 'LOW', 'numerical_situation': '1v1', 'xg_multiplier': 1.0692, 'skill_edge': 0.0351, 'height_edge': 0.012, 'attacker_skill': 0.5154, 'defender_skill': 0.4374, 'attacker_role': 'LW', 'attacker_mv_price_index': 1.0, 'defender_adjusted_parts': [{'player': 'Adam Smith', 'market_value': 310000, 'role': 'RB', 'price_index': 0.7, 'adjusted_quality': 0.4374}], 'xg_override': 0.048}	aerialWon: 0.4507 -> 1.1 totalShots: 0.5189 -> 0.5548 shotOffTarget: 0.3529 -> 0.3773 onTargetScoringAttempt: 0.1642 -> 0.1756
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 1 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=1`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.37 (squad xG sum 0.9028, xG scale 1.5176, xA scale 1.5176, team xA 0.98 = 0.72 x lambda)	`scoreline_reset=true` `team_lambda=1.37` `squad_xg_sum=0.9028` `squad_xa_sum=0.6471` `team_expected_assists=0.982` `team_xa_to_xg_ratio=0.7168`	expected_goals: 0.048 -> 0.0728 expected_assists: 0.1042 -> 0.1581
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 3 by projected rating, fifa_overall 78.0)	`match_squad_rank=3` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 1 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=1` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.0702` `goal_rate=0.0525` `scorer_weight=0.0761`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.1462` `assist_rate=0.1055` `provider_weight=0.1594`	-

Julio Enciso M 83 0 0.28 0.14 106% 3 2.0 7.0 45.8 14 FBref Prior M_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost M_service raw output over 8 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0211, quality 0.702)	`market_value_eur=28000000` `quality_factor=0.7022` `boost_factor=1.0211` `reference_quality=0.505` `history_matches=9` `history_shrink=0.3571`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0279)	`blend_weight=0.0279` `history_matches=9` `season_overlap_discount=1.0` `per90_to_match_factor=0.924` `prior_reliability=0.12` `competition=Premier League`	totalCross: 1.0371 -> 1.0082 goalAssist: 0.1513 -> 0.1757 totalShots: 2.5288 -> 2.5454 wasFouled: 1.4724 -> 1.4886 wonTackle: 0.631 -> 0.642 interceptionWon: 0.3487 -> 0.339 onTargetScoringAttempt: 0.8267 -> 0.8181 goals: 0.1953 -> 0.1899
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` duel_context={'opponent_faced': 'Adrien Truffert', 'matchup_zone': 'Right Wing vs Left Back — Flank Battle', 'duel_win_rate': 0.418, 'pressing_faced': 'HIGH', 'numerical_situation': '1v1', 'xg_multiplier': 0.9087, 'skill_edge': -0.0501, 'height_edge': -0.012, 'attacker_skill': 0.7022, 'defender_skill': 0.8135, 'attacker_role': 'RW', 'attacker_mv_price_index': 1.0, 'defender_adjusted_parts': [{'player': 'Adrien Truffert', 'market_value': 33000000, 'role': 'LB', 'price_index': 0.7, 'adjusted_quality': 0.8135}], 'xg_override': 0.1869}	aerialWon: 0.0884 -> 0.9 totalShots: 2.5454 -> 2.313 shotOffTarget: 1.7021 -> 1.5467 onTargetScoringAttempt: 0.8181 -> 0.7434 expectedGoalsOnTarget: 0.3533 -> 0.321 bigChanceMissed: 0.2842 -> 0.2583 expectedGoals: 0.2057 -> 0.1869 goals: 0.1899 -> 0.1726
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 1 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=1`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.37 (squad xG sum 0.9028, xG scale 1.5176, xA scale 1.5176, team xA 0.98 = 0.72 x lambda)	`scoreline_reset=true` `team_lambda=1.37` `squad_xg_sum=0.9028` `squad_xa_sum=0.6471` `team_expected_assists=0.982` `team_xa_to_xg_ratio=0.7168`	expected_goals: 0.1869 -> 0.2836 expected_assists: 0.152 -> 0.2306
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 2 by projected rating, fifa_overall 78.0)	`match_squad_rank=2` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 1 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=1` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.2469` `goal_rate=0.1726` `scorer_weight=0.2665`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=1` `assist_probability=0.2059` `assist_rate=0.1757` `provider_weight=0.2576`	-

Florentino Luís M 76 0 0.07 0.01 531% 1 0.2 6.8 47.2 31 XGBoost M_service

Step-by-step (9 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost M_service raw output over 8 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0028, quality 0.585)	`market_value_eur=15500000` `quality_factor=0.5846` `boost_factor=1.0028` `reference_quality=0.505` `history_matches=37` `history_shrink=0.119`	-
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` duel_context={'opponent_faced': 'Tyler Adams', 'matchup_zone': 'CDM vs CDM', 'duel_win_rate': 0.39, 'pressing_faced': 'HIGH', 'numerical_situation': '1v1', 'xg_multiplier': 0.9553, 'skill_edge': -0.0664, 'height_edge': 0.036, 'attacker_skill': 0.6134, 'defender_skill': 0.761, 'attacker_role': 'CDM', 'attacker_mv_price_index': 0.85, 'defender_adjusted_parts': [{'player': 'Tyler Adams', 'market_value': 31000000, 'role': 'CDM', 'price_index': 0.85, 'adjusted_quality': 0.761}], 'xg_override': 0.0473}	aerialWon: 0.942 -> 1.1 totalShots: 0.612 -> 0.5846 shotOffTarget: 0.4331 -> 0.4137 onTargetScoringAttempt: 0.1788 -> 0.1708
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 1 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=1`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.37 (squad xG sum 0.9028, xG scale 1.5176, xA scale 1.5176, team xA 0.98 = 0.72 x lambda)	`scoreline_reset=true` `team_lambda=1.37` `squad_xg_sum=0.9028` `squad_xa_sum=0.6471` `team_expected_assists=0.982` `team_xa_to_xg_ratio=0.7168`	expected_goals: 0.0473 -> 0.0718 expected_assists: 0.0376 -> 0.0571
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 9 by projected rating, fifa_overall 80.0)	`match_squad_rank=9` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv_filled` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 1 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=1` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.0693` `goal_rate=0.0499` `scorer_weight=0.0737`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.0555` `assist_rate=0.0502` `provider_weight=0.0653`	-

Exequiel Palacios M 87 0 0.05 - n/a 0 0.0 7.1 43.2 13 FBref Prior M_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost M_service raw output over 8 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0209, quality 0.686)	`market_value_eur=26000000` `quality_factor=0.6858` `boost_factor=1.0209` `reference_quality=0.505` `history_matches=8` `history_shrink=0.3846`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0050)	`blend_weight=0.005` `history_matches=8` `season_overlap_discount=1.0` `per90_to_match_factor=0.9671` `prior_reliability=0.02` `competition=Premier League`	interceptionWon: 1.0987 -> 1.1254 fouls: 1.1861 -> 1.1963 wonTackle: 1.2394 -> 1.2493 wasFouled: 1.2881 -> 1.2978
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` duel_context={'opponent_faced': 'Tyler Adams, Lewis Cook', 'matchup_zone': 'Attacking Mid vs Double Pivot — Tight Space Battle', 'duel_win_rate': 0.264, 'pressing_faced': 'MEDIUM', 'numerical_situation': '1v2', 'xg_multiplier': 1.0346, 'skill_edge': 0.0155, 'height_edge': 0.008, 'attacker_skill': 0.6971, 'defender_skill': 0.6626, 'attacker_role': 'CAM', 'attacker_mv_price_index': 0.95, 'defender_adjusted_parts': [{'player': 'Tyler Adams', 'market_value': 31000000, 'role': 'CDM', 'price_index': 0.85, 'adjusted_quality': 0.761}, {'player': 'Lewis Cook', 'market_value': 11600000, 'role': 'CDM', 'price_index': 0.85, 'adjusted_quality': 0.5643}], 'xg_override': 0.0347}	aerialWon: 0.2458 -> 1.0 totalShots: 0.4125 -> 0.4268 shotOffTarget: 0.307 -> 0.3176
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 1 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=1`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.37 (squad xG sum 0.9028, xG scale 1.5176, xA scale 1.5176, team xA 0.98 = 0.72 x lambda)	`scoreline_reset=true` `team_lambda=1.37` `squad_xg_sum=0.9028` `squad_xa_sum=0.6471` `team_expected_assists=0.982` `team_xa_to_xg_ratio=0.7168`	expected_goals: 0.0347 -> 0.0527 expected_assists: 0.059 -> 0.0895
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 1 by projected rating, fifa_overall 83.0)	`match_squad_rank=1` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 1 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=1` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.0513` `goal_rate=0.0441` `scorer_weight=0.0585`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.0856` `assist_rate=0.0553` `provider_weight=0.0865`	-

Abdoul Ouattara M 89 0 0.06 0.01 544% 1 0.2 6.9 48.5 21 FBref Prior M_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost M_service raw output over 8 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0097, quality 0.595)	`market_value_eur=16500000` `quality_factor=0.5953` `boost_factor=1.0097` `reference_quality=0.505` `history_matches=9` `history_shrink=0.3571`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0046)	`blend_weight=0.0046` `history_matches=9` `season_overlap_discount=1.0` `per90_to_match_factor=0.9873` `prior_reliability=0.02` `competition=Premier League`	totalCross: 1.9085 -> 1.8996 interceptionWon: 1.3234 -> 1.3173 wonTackle: 1.2916 -> 1.2856
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` duel_context={'opponent_faced': 'Tyler Adams', 'matchup_zone': 'CDM vs CDM', 'duel_win_rate': 0.379, 'pressing_faced': 'HIGH', 'numerical_situation': '1v1', 'xg_multiplier': 0.9396, 'skill_edge': -0.0611, 'height_edge': 0.02, 'attacker_skill': 0.6252, 'defender_skill': 0.761, 'attacker_role': 'CDM', 'attacker_mv_price_index': 0.85, 'defender_adjusted_parts': [{'player': 'Tyler Adams', 'market_value': 31000000, 'role': 'CDM', 'price_index': 0.85, 'adjusted_quality': 0.761}], 'xg_override': 0.0415}	aerialWon: 0.3639 -> 1.1 totalShots: 0.5401 -> 0.5075 shotOffTarget: 0.3944 -> 0.3706 onTargetScoringAttempt: 0.1457 -> 0.1369
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 1 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=1`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.37 (squad xG sum 0.9028, xG scale 1.5176, xA scale 1.5176, team xA 0.98 = 0.72 x lambda)	`scoreline_reset=true` `team_lambda=1.37` `squad_xg_sum=0.9028` `squad_xa_sum=0.6471` `team_expected_assists=0.982` `team_xa_to_xg_ratio=0.7168`	expected_goals: 0.0415 -> 0.063 expected_assists: 0.0794 -> 0.1205
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 6 by projected rating, fifa_overall 73.0)	`match_squad_rank=6` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 1 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=1` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.0611` `goal_rate=0.0432` `scorer_weight=0.0642`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.1135` `assist_rate=0.0921` `provider_weight=0.1323`	-

Emersonn F 71 1 0.42 0.19 118% 2 1.1 6.7 43.4 14 FBref Prior F_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost F_service raw output over 7 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0175, quality 0.669)	`market_value_eur=24000000` `quality_factor=0.6685` `boost_factor=1.0175` `reference_quality=0.505` `history_matches=9` `history_shrink=0.3571`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0186)	`blend_weight=0.0186` `history_matches=9` `season_overlap_discount=1.0` `per90_to_match_factor=0.7926` `prior_reliability=0.08` `competition=Premier League`	totalOffside: 0.8892 -> 0.8727 onTargetScoringAttempt: 0.6919 -> 0.7032 totalCross: 0.0541 -> 0.0654 wonTackle: 0.2327 -> 0.2406 goals: 0.2919 -> 0.2987 wasFouled: 1.6671 -> 1.6607
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` duel_context={'opponent_faced': 'James Hill, António Silva', 'matchup_zone': 'Striker vs Two Centre Backs — 1v2 Aerial/Physical Battle', 'duel_win_rate': 0.18, 'pressing_faced': 'HIGH', 'numerical_situation': '1v2', 'xg_multiplier': 0.9199, 'skill_edge': -0.0565, 'height_edge': 0.002, 'attacker_skill': 0.6685, 'defender_skill': 0.7941, 'attacker_role': 'ST', 'attacker_mv_price_index': 1.0, 'defender_adjusted_parts': [{'player': 'James Hill', 'market_value': 32000000, 'role': 'CB', 'price_index': 0.75, 'adjusted_quality': 0.7941}, {'player': 'António Silva', 'market_value': 32000000, 'role': 'CB', 'price_index': 0.75, 'adjusted_quality': 0.7941}], 'xg_override': 0.2741}	totalShots: 2.111 -> 1.9419 shotOffTarget: 1.4078 -> 1.295 onTargetScoringAttempt: 0.7032 -> 0.6469 goals: 0.2987 -> 0.2748 expectedGoals: 0.298 -> 0.2741 bigChanceMissed: 0.2866 -> 0.2636 expectedGoalsOnTarget: 0.1338 -> 0.1231
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 1 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=1`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.37 (squad xG sum 0.9028, xG scale 1.5176, xA scale 1.5176, team xA 0.98 = 0.72 x lambda)	`scoreline_reset=true` `team_lambda=1.37` `squad_xg_sum=0.9028` `squad_xa_sum=0.6471` `team_expected_assists=0.982` `team_xa_to_xg_ratio=0.7168`	expected_goals: 0.2741 -> 0.416 expected_assists: 0.0435 -> 0.066
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 13 by projected rating, fifa_overall 74.0)	`match_squad_rank=13` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 1 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=1` `allocation_applied=true` `goals_allocated=1` `goal_probability=0.3403` `goal_rate=0.2748` `scorer_weight=0.4169`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.0639` `assist_rate=0.0865` `provider_weight=0.0825`	-

Team totals 1 1.37 10 6.9

Bench Candidates (7 modeled substitutes with empirical playing time)

Player	Pos	Exp Min	App Prob	Substitute Modeling Details
Kjell Scherpen	G	4.8m	14.0%	Market-value substitute prior
Cédric Kipré	D	13.6m	29.9%	Market-value substitute prior
Darnell Furlong	D	16.0m	60.2%	Market-value substitute prior
Marcelino Núñez	M	22.5m	31.8%	Market-value substitute prior
Azor Matusiwa	M	22.4m	52.0%	Market-value substitute prior
Abdul Fatawu Issahaku	F	26.2m	35.3%	Market-value substitute prior

Bournemouth — projected line vs season baseline (4-2-3-1)

Player Pos Min Proj G Proj xG Season xG/g Proj Sh Season Sh/g Proj Rtg Season Rtg Rec Model / FBref Prior

Đorđe Petrović G 90 0 0.00 - n/a 0.0 6.9 58.5 33 FBref Prior G_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost G_service raw output over 3 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0059, quality 0.710)	`market_value_eur=29000000` `quality_factor=0.71` `boost_factor=1.0059` `reference_quality=0.505` `history_matches=47` `history_shrink=0.0962`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0083)	`blend_weight=0.0083` `history_matches=47` `season_overlap_discount=1.0` `per90_to_match_factor=1.0` `prior_reliability=0.1333` `competition=Premier League`	saves: 2.7329 -> 2.7476
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` `duel_context={'opponent_faced': 'Emersonn, Julio Enciso', 'matchup_zone': 'Goalkeeper shot-stopping context', 'opponent_lambda': 1.2104}`	-
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 1 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=1`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.41 (squad xG sum 1.1126, xG scale 1.2682, xA scale 1.2682, team xA 1.12 = 0.80 x lambda)	`scoreline_reset=true` `team_lambda=1.411` `squad_xg_sum=1.1126` `squad_xa_sum=0.8871` `team_expected_assists=1.125` `team_xa_to_xg_ratio=0.7973`	shots_faced: 11.5 -> 13.0 expected_saves: 2.81 -> 3.18 high_pressure_saves: 0.97 -> 1.1
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 7 by projected rating, fifa_overall 79.0)	`match_squad_rank=7` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 1 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=1` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.0` `scorer_weight=0.0` `squad_weight_total=1.4986`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.0` `provider_weight=0.0` `squad_weight_total=1.1331`	-

Adrien Truffert D 89 0 0.04 0.04 ok 1 0.4 6.9 58.5 33 FBref Prior D_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost D_service raw output over 7 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0067, quality 0.739)	`market_value_eur=33000000` `quality_factor=0.7389` `boost_factor=1.0067` `reference_quality=0.505` `history_matches=47` `history_shrink=0.0962`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0083)	`blend_weight=0.0083` `history_matches=47` `season_overlap_discount=1.0` `per90_to_match_factor=0.9938` `prior_reliability=0.1333` `competition=Premier League`	totalCross: 3.6485 -> 3.6388 wonTackle: 1.3299 -> 1.323 totalShots: 0.6246 -> 0.6194
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` `duel_context={'opponent_faced': 'Julio Enciso', 'matchup_zone': 'Right Wing vs Left Back — Flank Battle'}`	-
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 1 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=1`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.41 (squad xG sum 1.1126, xG scale 1.2682, xA scale 1.2682, team xA 1.12 = 0.80 x lambda)	`scoreline_reset=true` `team_lambda=1.411` `squad_xg_sum=1.1126` `squad_xa_sum=0.8871` `team_expected_assists=1.125` `team_xa_to_xg_ratio=0.7973`	expected_goals: 0.0309 -> 0.0392 expected_assists: 0.1253 -> 0.1589
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 8 by projected rating, fifa_overall 80.0)	`match_squad_rank=8` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 1 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=1` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.0384` `goal_rate=0.0158` `scorer_weight=0.0319`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=1` `assist_probability=0.1469` `assist_rate=0.2079` `provider_weight=0.1986`	-

James Hill D 90 0 0.02 0.02 ok 1 0.4 7.2 60.9 33 FBref Prior D_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost D_service raw output over 7 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0079, quality 0.732)	`market_value_eur=32000000` `quality_factor=0.732` `boost_factor=1.0079` `reference_quality=0.505` `history_matches=38` `history_shrink=0.1163`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0101)	`blend_weight=0.0101` `history_matches=38` `season_overlap_discount=1.0` `per90_to_match_factor=0.9996` `prior_reliability=0.1333` `competition=Premier League`	totalShots: 0.6993 -> 0.6923 fouls: 0.8455 -> 0.8521
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` `duel_context={'opponent_faced': 'Emersonn', 'matchup_zone': 'Striker vs Two Centre Backs — 1v2 Aerial/Physical Battle'}`	-
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 1 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=1`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.41 (squad xG sum 1.1126, xG scale 1.2682, xA scale 1.2682, team xA 1.12 = 0.80 x lambda)	`scoreline_reset=true` `team_lambda=1.411` `squad_xg_sum=1.1126` `squad_xa_sum=0.8871` `team_expected_assists=1.125` `team_xa_to_xg_ratio=0.7973`	expected_assists: 0.0603 -> 0.0765
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 1 by projected rating, fifa_overall 78.0)	`match_squad_rank=1` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 1 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=1` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.0201` `goal_rate=0.0` `scorer_weight=0.0162`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.0736` `assist_rate=0.0123` `provider_weight=0.0612`	-

Adam Smith D 85 0 0.08 0.02 357% 0 0.1 6.7 37.8 33 FBref Prior D_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost D_service raw output over 7 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 0.9972, quality 0.437)	`market_value_eur=310000` `quality_factor=0.4374` `boost_factor=0.9972` `reference_quality=0.505` `history_matches=31` `history_shrink=0.1389`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0120)	`blend_weight=0.012` `history_matches=31` `season_overlap_discount=1.0` `per90_to_match_factor=0.9391` `prior_reliability=0.1333` `competition=Premier League`	interceptionWon: 0.8741 -> 0.8805
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` `duel_context={'opponent_faced': 'Saša Lukić', 'matchup_zone': 'Left Wing vs Right Back — Flank Battle'}`	-
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 1 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=1`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.41 (squad xG sum 1.1126, xG scale 1.2682, xA scale 1.2682, team xA 1.12 = 0.80 x lambda)	`scoreline_reset=true` `team_lambda=1.411` `squad_xg_sum=1.1126` `squad_xa_sum=0.8871` `team_expected_assists=1.125` `team_xa_to_xg_ratio=0.7973`	expected_goals: 0.0604 -> 0.0766 expected_assists: 0.0487 -> 0.0618
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 14 by projected rating, fifa_overall 73.0)	`match_squad_rank=14` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 1 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=1` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.0737` `goal_rate=0.0` `scorer_weight=0.0613`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.0599` `assist_rate=0.0366` `provider_weight=0.0542`	-

António Silva D 90 0 0.16 0.09 70% 1 0.4 7.0 44.7 14 FBref Prior D_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost D_service raw output over 7 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0243, quality 0.732)	`market_value_eur=32000000` `quality_factor=0.732` `boost_factor=1.0243` `reference_quality=0.505` `history_matches=9` `history_shrink=0.3571`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0309)	`blend_weight=0.0309` `history_matches=9` `season_overlap_discount=1.0` `per90_to_match_factor=1.0` `prior_reliability=0.1333` `competition=Premier League`	fouls: 0.8511 -> 0.8402 interceptionWon: 1.2634 -> 1.2552 totalShots: 0.7379 -> 0.7305 wonTackle: 0.7893 -> 0.7958 onTargetScoringAttempt: 0.1982 -> 0.1921
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` `duel_context={'opponent_faced': 'Emersonn', 'matchup_zone': 'Striker vs Two Centre Backs — 1v2 Aerial/Physical Battle'}`	-
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 1 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=1`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.41 (squad xG sum 1.1126, xG scale 1.2682, xA scale 1.2682, team xA 1.12 = 0.80 x lambda)	`scoreline_reset=true` `team_lambda=1.411` `squad_xg_sum=1.1126` `squad_xa_sum=0.8871` `team_expected_assists=1.125` `team_xa_to_xg_ratio=0.7973`	expected_goals: 0.1242 -> 0.1575 expected_assists: 0.034 -> 0.0431
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 4 by projected rating, fifa_overall 77.0)	`match_squad_rank=4` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 1 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=1` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.1457` `goal_rate=0.0` `scorer_weight=0.126`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.0422` `assist_rate=0.0` `provider_weight=0.0345`	-

Alex Scott M 89 0 0.13 0.12 ok 1 1.1 7.2 60.0 28 FBref Prior M_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost M_service raw output over 8 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0119, quality 0.877)	`market_value_eur=75000000` `quality_factor=0.877` `boost_factor=1.0119` `reference_quality=0.505` `history_matches=42` `history_shrink=0.1064`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0092)	`blend_weight=0.0092` `history_matches=42` `season_overlap_discount=1.0` `per90_to_match_factor=0.9889` `prior_reliability=0.1333` `competition=Premier League`	wonTackle: 1.0163 -> 1.0252 wasFouled: 2.7468 -> 2.7397 totalCross: 3.199 -> 3.206
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` duel_context={'opponent_faced': 'Florentino Luís, Abdoul Ouattara', 'matchup_zone': 'Attacking Mid vs Double Pivot — Tight Space Battle', 'duel_win_rate': 0.342, 'pressing_faced': 'LOW', 'numerical_situation': '1v2', 'xg_multiplier': 1.1496, 'skill_edge': 0.1178, 'height_edge': -0.016, 'attacker_skill': 0.8811, 'defender_skill': 0.6193, 'attacker_role': 'CAM', 'attacker_mv_price_index': 0.95, 'defender_adjusted_parts': [{'player': 'Florentino Luís', 'market_value': 15500000, 'role': 'CDM', 'price_index': 0.85, 'adjusted_quality': 0.6134}, {'player': 'Abdoul Ouattara', 'market_value': 16500000, 'role': 'CDM', 'price_index': 0.85, 'adjusted_quality': 0.6252}], 'xg_override': 0.1004}	totalShots: 1.1684 -> 1.3432 aerialWon: 0.8709 -> 1.0 shotOffTarget: 0.8235 -> 0.9467 onTargetScoringAttempt: 0.3449 -> 0.3965 bigChanceMissed: 0.149 -> 0.1713 expectedGoalsOnTarget: 0.118 -> 0.1357 goals: 0.1115 -> 0.1282 expectedGoals: 0.0873 -> 0.1004
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 1 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=1`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.41 (squad xG sum 1.1126, xG scale 1.2682, xA scale 1.2682, team xA 1.12 = 0.80 x lambda)	`scoreline_reset=true` `team_lambda=1.411` `squad_xg_sum=1.1126` `squad_xa_sum=0.8871` `team_expected_assists=1.125` `team_xa_to_xg_ratio=0.7973`	expected_goals: 0.1004 -> 0.1273 expected_assists: 0.1513 -> 0.1919
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 2 by projected rating, fifa_overall 81.0)	`match_squad_rank=2` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 1 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=1` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.1195` `goal_rate=0.1282` `scorer_weight=0.1508`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.1746` `assist_rate=0.0984` `provider_weight=0.1535`	-

Tyler Adams M 84 0 0.06 0.01 497% 1 0.2 6.9 51.4 25 FBref Prior M_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost M_service raw output over 8 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0085, quality 0.725)	`market_value_eur=31000000` `quality_factor=0.7249` `boost_factor=1.0085` `reference_quality=0.505` `history_matches=34` `history_shrink=0.1282`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0022)	`blend_weight=0.0022` `history_matches=34` `season_overlap_discount=1.0` `per90_to_match_factor=0.9381` `prior_reliability=0.0267` `competition=Premier League`	interceptionWon: 1.3629 -> 1.3703
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` duel_context={'opponent_faced': 'Florentino Luís', 'matchup_zone': 'CDM vs CDM', 'duel_win_rate': 0.45, 'pressing_faced': 'LOW', 'numerical_situation': '1v1', 'xg_multiplier': 1.0447, 'skill_edge': 0.0664, 'height_edge': -0.036, 'attacker_skill': 0.761, 'defender_skill': 0.6134, 'attacker_role': 'CDM', 'attacker_mv_price_index': 0.85, 'defender_adjusted_parts': [{'player': 'Florentino Luís', 'market_value': 15500000, 'role': 'CDM', 'price_index': 0.85, 'adjusted_quality': 0.6134}], 'xg_override': 0.0438}	aerialWon: 1.0965 -> 0.9 totalShots: 0.507 -> 0.5297 shotOffTarget: 0.3812 -> 0.3982 onTargetScoringAttempt: 0.1258 -> 0.1314
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 1 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=1`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.41 (squad xG sum 1.1126, xG scale 1.2682, xA scale 1.2682, team xA 1.12 = 0.80 x lambda)	`scoreline_reset=true` `team_lambda=1.411` `squad_xg_sum=1.1126` `squad_xa_sum=0.8871` `team_expected_assists=1.125` `team_xa_to_xg_ratio=0.7973`	expected_goals: 0.0438 -> 0.0555 expected_assists: 0.0335 -> 0.0425
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 6 by projected rating, fifa_overall 79.0)	`match_squad_rank=6` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 1 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=1` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.054` `goal_rate=0.0545` `scorer_weight=0.0618`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.0416` `assist_rate=0.0481` `provider_weight=0.0499`	-

Justin Kluivert M 73 0 0.21 0.10 117% 2 1.1 6.8 41.4 16 FBref Prior M_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost M_service raw output over 8 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0077, quality 0.669)	`market_value_eur=24000000` `quality_factor=0.6685` `boost_factor=1.0077` `reference_quality=0.505` `history_matches=27` `history_shrink=0.1562`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0115)	`blend_weight=0.0115` `history_matches=27` `season_overlap_discount=1.0` `per90_to_match_factor=0.8144` `prior_reliability=0.1133` `competition=Premier League`	onTargetScoringAttempt: 0.5704 -> 0.5802 totalCross: 2.2455 -> 2.2362 fouls: 1.7088 -> 1.7002 totalShots: 1.718 -> 1.7256 wonTackle: 0.4093 -> 0.4156
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` duel_context={'opponent_faced': "Dara O'Shea", 'matchup_zone': 'Left Wing vs Right Back — Flank Battle', 'duel_win_rate': 0.398, 'pressing_faced': 'MEDIUM', 'numerical_situation': '1v1', 'xg_multiplier': 0.8794, 'skill_edge': -0.0221, 'height_edge': -0.06, 'attacker_skill': 0.6685, 'defender_skill': 0.7176, 'attacker_role': 'LW', 'attacker_mv_price_index': 1.0, 'defender_adjusted_parts': [{'player': "Dara O'Shea", 'market_value': 21000000, 'role': 'RB', 'price_index': 0.7, 'adjusted_quality': 0.7176}], 'xg_override': 0.1626}	aerialWon: 0.4679 -> 0.8 totalShots: 1.7278 -> 1.5194 shotOffTarget: 1.1476 -> 1.0092 onTargetScoringAttempt: 0.5802 -> 0.5102 expectedGoalsOnTarget: 0.2487 -> 0.2187 goals: 0.1866 -> 0.1641 expectedGoals: 0.1849 -> 0.1626 bigChanceMissed: 0.0927 -> 0.0815
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 1 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=1`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.41 (squad xG sum 1.1126, xG scale 1.2682, xA scale 1.2682, team xA 1.12 = 0.80 x lambda)	`scoreline_reset=true` `team_lambda=1.411` `squad_xg_sum=1.1126` `squad_xa_sum=0.8871` `team_expected_assists=1.125` `team_xa_to_xg_ratio=0.7973`	expected_goals: 0.1626 -> 0.2062 expected_assists: 0.1067 -> 0.1353
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 12 by projected rating, fifa_overall 79.0)	`match_squad_rank=12` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 1 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=1` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.1863` `goal_rate=0.1641` `scorer_weight=0.2077`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.1265` `assist_rate=0.1417` `provider_weight=0.1655`	-

Ryan Christie M 73 0 0.09 0.06 49% 1 0.9 6.9 53.3 28 FBref Prior M_service

Step-by-step (9 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost M_service raw output over 8 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0000, quality 0.506)	`market_value_eur=8600000` `quality_factor=0.5059` `boost_factor=1.0` `reference_quality=0.505` `history_matches=34` `history_shrink=0.1282`	-
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` duel_context={'opponent_faced': 'Leif Davis', 'matchup_zone': 'Right Wing vs Left Back — Flank Battle', 'duel_win_rate': 0.347, 'pressing_faced': 'MEDIUM', 'numerical_situation': '1v1', 'xg_multiplier': 0.8051, 'skill_edge': -0.1166, 'height_edge': -0.016, 'attacker_skill': 0.5059, 'defender_skill': 0.765, 'attacker_role': 'RW', 'attacker_mv_price_index': 1.0, 'defender_adjusted_parts': [{'player': 'Leif Davis', 'market_value': 26000000, 'role': 'LB', 'price_index': 0.7, 'adjusted_quality': 0.765}], 'xg_override': 0.0728}	totalShots: 1.1548 -> 0.9297 shotOffTarget: 0.8289 -> 0.6673 onTargetScoringAttempt: 0.3259 -> 0.2624 aerialWon: 1.0612 -> 1.0 bigChanceMissed: 0.1805 -> 0.1453 expectedGoals: 0.0904 -> 0.0728 goals: 0.0796 -> 0.0641 expectedGoalsOnTarget: 0.0433 -> 0.0349
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 1 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=1`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.41 (squad xG sum 1.1126, xG scale 1.2682, xA scale 1.2682, team xA 1.12 = 0.80 x lambda)	`scoreline_reset=true` `team_lambda=1.411` `squad_xg_sum=1.1126` `squad_xa_sum=0.8871` `team_expected_assists=1.125` `team_xa_to_xg_ratio=0.7973`	expected_goals: 0.0728 -> 0.0923 expected_assists: 0.0852 -> 0.108
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 5 by projected rating, fifa_overall 78.0)	`match_squad_rank=5` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 1 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=1` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.0882` `goal_rate=0.0641` `scorer_weight=0.0858`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.1024` `assist_rate=0.078` `provider_weight=0.1022`	-

Lewis Cook M 80 0 0.04 0.00 1023% 1 0.0 6.8 50.7 23 FBref Prior M_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost M_service raw output over 8 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0017, quality 0.541)	`market_value_eur=11600000` `quality_factor=0.5411` `boost_factor=1.0017` `reference_quality=0.505` `history_matches=27` `history_shrink=0.1562`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0108)	`blend_weight=0.0108` `history_matches=27` `season_overlap_discount=1.0` `per90_to_match_factor=0.8884` `prior_reliability=0.1067` `competition=Premier League`	totalCross: 2.6407 -> 2.6121 wonTackle: 0.9905 -> 0.9798 fouls: 0.6912 -> 0.6837 totalShots: 0.6043 -> 0.5978 wasFouled: 0.5697 -> 0.5635
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` duel_context={'opponent_faced': 'Florentino Luís', 'matchup_zone': 'CDM vs CDM', 'duel_win_rate': 0.362, 'pressing_faced': 'LOW', 'numerical_situation': '1v1', 'xg_multiplier': 0.9146, 'skill_edge': -0.0221, 'height_edge': -0.036, 'attacker_skill': 0.5643, 'defender_skill': 0.6134, 'attacker_role': 'CDM', 'attacker_mv_price_index': 0.85, 'defender_adjusted_parts': [{'player': 'Florentino Luís', 'market_value': 15500000, 'role': 'CDM', 'price_index': 0.85, 'adjusted_quality': 0.6134}], 'xg_override': 0.0325}	aerialWon: 0.8463 -> 0.9 totalShots: 0.6029 -> 0.5514 shotOffTarget: 0.4728 -> 0.4324 onTargetScoringAttempt: 0.1301 -> 0.119 goals: 0.0593 -> 0.0542
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 1 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=1`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.41 (squad xG sum 1.1126, xG scale 1.2682, xA scale 1.2682, team xA 1.12 = 0.80 x lambda)	`scoreline_reset=true` `team_lambda=1.411` `squad_xg_sum=1.1126` `squad_xa_sum=0.8871` `team_expected_assists=1.125` `team_xa_to_xg_ratio=0.7973`	expected_goals: 0.0325 -> 0.0412 expected_assists: 0.0674 -> 0.0855
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 9 by projected rating, fifa_overall 78.0)	`match_squad_rank=9` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 1 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=1` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.0404` `goal_rate=0.0542` `scorer_weight=0.0515`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.0819` `assist_rate=0.0762` `provider_weight=0.0919`	-

Rayan F 86 1 0.42 0.17 142% 3 1.8 7.1 50.6 41 FBref Prior F_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost F_service raw output over 7 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0177, quality 0.848)	`market_value_eur=58000000` `quality_factor=0.8478` `boost_factor=1.0177` `reference_quality=0.505` `history_matches=24` `history_shrink=0.1724`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0149)	`blend_weight=0.0149` `history_matches=24` `season_overlap_discount=1.0` `per90_to_match_factor=0.9583` `prior_reliability=0.1333` `competition=Premier League`	totalCross: 1.5508 -> 1.5276 wasFouled: 1.1721 -> 1.1546 fouls: 1.1236 -> 1.1068 wonTackle: 0.7131 -> 0.7024 totalShots: 2.2305 -> 2.2406 interceptionWon: 1.1515 -> 1.1415 onTargetScoringAttempt: 0.8147 -> 0.8243 goals: 0.3817 -> 0.376
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` duel_context={'opponent_faced': 'Issa Diop, Jacob Greaves', 'matchup_zone': 'Striker vs Two Centre Backs — 1v2 Aerial/Physical Battle', 'duel_win_rate': 0.275, 'pressing_faced': 'LOW', 'numerical_situation': '1v2', 'xg_multiplier': 1.1097, 'skill_edge': 0.1086, 'height_edge': -0.034, 'attacker_skill': 0.8478, 'defender_skill': 0.6065, 'attacker_role': 'ST', 'attacker_mv_price_index': 1.0, 'defender_adjusted_parts': [{'player': 'Issa Diop', 'market_value': 10300000, 'role': 'CB', 'price_index': 0.75, 'adjusted_quality': 0.5652}, {'player': 'Jacob Greaves', 'market_value': 16300000, 'role': 'CB', 'price_index': 0.75, 'adjusted_quality': 0.6478}], 'xg_override': 0.3331}	totalShots: 2.2406 -> 2.4864 shotOffTarget: 1.4158 -> 1.5711 onTargetScoringAttempt: 0.8243 -> 0.9147 bigChanceMissed: 0.7808 -> 0.8665 goals: 0.376 -> 0.4172 expectedGoals: 0.3002 -> 0.3331 expectedGoalsOnTarget: 0.2802 -> 0.3109
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 1 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=1`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.41 (squad xG sum 1.1126, xG scale 1.2682, xA scale 1.2682, team xA 1.12 = 0.80 x lambda)	`scoreline_reset=true` `team_lambda=1.411` `squad_xg_sum=1.1126` `squad_xa_sum=0.8871` `team_expected_assists=1.125` `team_xa_to_xg_ratio=0.7973`	expected_goals: 0.3331 -> 0.4225 expected_assists: 0.087 -> 0.1103
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 3 by projected rating, fifa_overall 79.0)	`match_squad_rank=3` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 1 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=1` `allocation_applied=true` `goals_allocated=1` `goal_probability=0.3446` `goal_rate=0.4172` `scorer_weight=0.5152`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.1044` `assist_rate=0.0187` `provider_weight=0.0882`	-

Team totals 1 1.41 15 6.9

Bench Candidates (7 modeled substitutes with empirical playing time)

Player	Pos	Exp Min	App Prob	Substitute Modeling Details
Michele Di Gregorio	G	4.6m	11.9%	Market-value substitute prior
Bafodé Diakité	D	14.3m	33.4%	Market-value substitute prior
Veljko Milosavljević	D	14.3m	33.1%	Market-value substitute prior
Alex Tóth	M	18.9m	70.2%	Market-value substitute prior
Juanlu Sánchez	M	22.3m	51.3%	Market-value substitute prior
Eli Junior Kroupi	F	27.0m	63.7%	Market-value substitute prior

Antagonist / tactical summary

Midfield is evenly contested — Julio Enciso vs Alex Scott will be the pivotal duel. Total antagonist xG — Home: 0.63, Away: 0.75. Home danger zone: Striker vs Two Centre Backs — 1v2 Aerial/Physical Battle. Away danger zone: Striker vs Two Centre Backs — 1v2 Aerial/Physical Battle.

Midfield — possession 50.1% / 49.9%, dominance ratio 0.48, key duel: Julio Enciso vs Alex Scott

Saša Lukić xG 0.05Julio Enciso xG 0.19Exequiel Palacios xG 0.03Justin Kluivert xG 0.16Ryan Christie xG 0.07Alex Scott xG 0.10

Summary

Antagonist matchup analysis for Ipswich vs Bournemouth. Most likely scoreline: 1-1. Lambdas derived from 6 home and 6 away individual player duels. Head-to-head: Ipswich 1W-4D-1L vs Bournemouth over 6 meetings.

Tactical parity between 4-2-3-1 and 4-2-3-1. Match outcome driven primarily by individual player quality. Midfield is evenly contested — Julio Enciso vs Alex Scott will be the pivotal duel. Total antagonist xG — Home: 0.63, Away: 0.75. Home danger zone: Striker vs Two Centre Backs — 1v2 Aerial/Physical Battle. Away danger zone: Striker vs Two Centre Backs — 1v2 Aerial/Physical Battle.

Brighton vs Brentford

Premier League · 20261108 · 4-2-3-1 vs 4-2-3-1

Predicted 2-1Winner: BrightonWinning H 2-1 (8.8%)Winning A 1-2 (7.9%)Confidence 0.66λ home/away 1.65 / 1.47Most-likely P 8.8%

Actual vs predicted: result not available yet (upcoming fixture).

XGBoost Reasoning & FBref Dataset Tracing

Antagonist + XGBoost match analysis for Brighton vs Brentford. Player statistics regressed using position-specific XGBoost models (G, D, M, F) with 2026-2027 FBref priors. Micro-duels resolved across 4-2-3-1 vs 4-2-3-1 (duel lambdas: Home 1.878, Away 1.462). Final calibrated Poisson lambdas: Home 1.649, Away 1.472 (most likely score 2-1).

Home Starters 11/11 XGBoost (9 FBref) Away Starters 11/11 XGBoost (11 FBref) Bench Modeled H: 7 / A: 7 Duel λ (H/A) 1.88 / 1.46 Final λ (H/A) 1.65 / 1.47

FBref Team Priors: Brighton & Hove Albion (Gls/90: 2.5, GA/90: 3.5 → E[λ]: 3.448) vs Brentford (Gls/90: 2.0, GA/90: 2.0 → E[λ]: 3.5) | Blend Weight: 0.192

Modeling Pipeline Stages (5 layers)

1. XGBoost player regressions for volume & event rates with market-value quality scaling
2. Multi-source prior blending (FBref 2026-2027 tables + Sofascore season history)
3. Antagonist formation duel matrix resolving 1v1 and zonal matchups
4. Team-level prior calibration (FBref possession and attack/defense efficiencies)
5. Bivariate Poisson matrix with Dixon-Coles low-score correction

Standing & form context

Metric	Brighton	Brentford
League position	3.00	6.00
Points per game	1.81	1.48
Attack index	1.48	1.08
Defense (concede) index	0.84	0.86
Form multiplier	1.02	0.98
Standing λ multiplier	1.10	0.96

Lambda decomposition (how the goal rates were built)

Stage	Brighton λ	Brentford λ
FBref Squad Prior E[λ] (weight 0.192)	3.448	3.500
Pre-competitive-floor	1.930	1.742
Pre-calibration (Micro-duels)	1.878	1.462
Final λ	1.649	1.472

competition calibration scale 0.85competition observed total goals 2.781quality gap applied Falsemismatch score pressure applied Falsecompetitive underdog floor applied Falsestage discount 1.0

Scoreline distribution & markets

Top 5 scorelines	P
1-1	10.8%
2-1	8.8%
1-2	7.9%
1-0	7.2%
2-2	6.5%

Home win	41.9%
Draw	23.8%
Away win	34.3%
Over 2.5	60.3%
BTTS yes	62.3%

Pregame fair line — favorite Brighton (41.9%), fair total 3.12, fair spread 0.18

Projected players vs season baseline

Proj = projected for this match · Season = this season's per-game average from match data · deviation badge compares projected xG to the player's season xG/game.

Brighton — projected line vs season baseline (4-2-3-1)

Player Pos Min Proj G Proj xG Season xG/g Proj Sh Season Sh/g Proj Rtg Season Rtg Rec Model / FBref Prior

Bart Verbruggen G 90 0 0.00 - n/a 0.0 7.1 60.6 33 FBref Prior G_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost G_service raw output over 3 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0075, quality 0.764)	`market_value_eur=37000000` `quality_factor=0.7641` `boost_factor=1.0075` `reference_quality=0.505` `history_matches=47` `history_shrink=0.0962`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0083)	`blend_weight=0.0083` `history_matches=47` `season_overlap_discount=1.0` `per90_to_match_factor=1.0` `prior_reliability=0.1333` `competition=Premier League`	saves: 3.076 -> 3.0545
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` `duel_context={'opponent_faced': 'Igor Thiago, Mikkel Damsgaard', 'matchup_zone': 'Goalkeeper shot-stopping context', 'opponent_lambda': 1.4619}`	-
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 2 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=2`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.65 (squad xG sum 1.3905, xG scale 1.1859, xA scale 1.2086, team xA 1.07 = 0.65 x lambda)	`scoreline_reset=true` `team_lambda=1.649` `squad_xg_sum=1.3905` `squad_xa_sum=0.8869` `team_expected_assists=1.0718` `team_xa_to_xg_ratio=0.65`	shots_faced: 13.9 -> 14.0 expected_saves: 3.4 -> 3.43 high_pressure_saves: 1.17 -> 1.18
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 1 by projected rating, fifa_overall 81.0)	`match_squad_rank=1` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 2 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=2` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.0` `scorer_weight=0.0` `squad_weight_total=1.6715`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.0` `provider_weight=0.0` `squad_weight_total=1.1039`	-

Ferdi Kadıoğlu D 76 0 0.04 0.05 ok 1 0.7 6.8 55.3 31 XGBoost D_service

Step-by-step (9 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost D_service raw output over 7 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0086, quality 0.786)	`market_value_eur=41000000` `quality_factor=0.7859` `boost_factor=1.0086` `reference_quality=0.505` `history_matches=44` `history_shrink=0.102`	-
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` `duel_context={'opponent_faced': 'Igor Thiago', 'matchup_zone': 'Striker vs Two Centre Backs — 1v2 Aerial/Physical Battle'}`	-
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 2 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=2`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.65 (squad xG sum 1.3905, xG scale 1.1859, xA scale 1.2086, team xA 1.07 = 0.65 x lambda)	`scoreline_reset=true` `team_lambda=1.649` `squad_xg_sum=1.3905` `squad_xa_sum=0.8869` `team_expected_assists=1.0718` `team_xa_to_xg_ratio=0.65`	expected_goals: 0.0311 -> 0.0369 expected_assists: 0.0536 -> 0.0648
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 13 by projected rating, fifa_overall 80.0)	`match_squad_rank=13` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 2 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=2` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.0362` `goal_rate=0.0256` `scorer_weight=0.0344`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.0627` `assist_rate=0.0` `provider_weight=0.0518`	-

Maxim De Cuyper D 80 0 0.17 0.10 67% 1 0.7 6.8 50.1 39 FBref Prior D_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost D_service raw output over 7 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0070, quality 0.710)	`market_value_eur=29000000` `quality_factor=0.71` `boost_factor=1.007` `reference_quality=0.505` `history_matches=39` `history_shrink=0.1136`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0094)	`blend_weight=0.0094` `history_matches=39` `season_overlap_discount=1.0` `per90_to_match_factor=0.8934` `prior_reliability=0.1267` `competition=Premier League`	totalCross: 3.938 -> 3.9584 totalShots: 0.7959 -> 0.8109 wasFouled: 1.3743 -> 1.3614 fouls: 1.2782 -> 1.2838
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` `duel_context={'opponent_faced': 'Mikkel Damsgaard', 'matchup_zone': 'Right Wing vs Left Back — Flank Battle'}`	-
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 2 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=2`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.65 (squad xG sum 1.3905, xG scale 1.1859, xA scale 1.2086, team xA 1.07 = 0.65 x lambda)	`scoreline_reset=true` `team_lambda=1.649` `squad_xg_sum=1.3905` `squad_xa_sum=0.8869` `team_expected_assists=1.0718` `team_xa_to_xg_ratio=0.65`	expected_goals: 0.1468 -> 0.1741 expected_assists: 0.1577 -> 0.1906
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 8 by projected rating, fifa_overall 78.0)	`match_squad_rank=8` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 2 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=2` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.1598` `goal_rate=0.1463` `scorer_weight=0.1737`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=1` `assist_probability=0.1735` `assist_rate=0.3273` `provider_weight=0.2382`	-

Olivier Boscagli D 77 0 0.02 0.01 138% 1 0.2 6.8 39.1 34 FBref Prior D_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost D_service raw output over 7 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0064, quality 0.616)	`market_value_eur=18500000` `quality_factor=0.6161` `boost_factor=1.0064` `reference_quality=0.505` `history_matches=21` `history_shrink=0.1923`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0167)	`blend_weight=0.0167` `history_matches=21` `season_overlap_discount=1.0` `per90_to_match_factor=0.8527` `prior_reliability=0.1333` `competition=Premier League`	fouls: 0.8919 -> 0.9055 onTargetScoringAttempt: 0.0511 -> 0.0574 wasFouled: 0.8992 -> 0.9055 totalCross: 2.4376 -> 2.4325
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` `duel_context={'opponent_faced': 'Mathias Jensen', 'matchup_zone': 'Left Wing vs Right Back — Flank Battle'}`	-
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 2 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=2`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.65 (squad xG sum 1.3905, xG scale 1.1859, xA scale 1.2086, team xA 1.07 = 0.65 x lambda)	`scoreline_reset=true` `team_lambda=1.649` `squad_xg_sum=1.3905` `squad_xa_sum=0.8869` `team_expected_assists=1.0718` `team_xa_to_xg_ratio=0.65`	expected_assists: 0.0666 -> 0.0805
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 6 by projected rating, fifa_overall 78.0)	`match_squad_rank=6` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 2 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=2` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.0158` `goal_rate=0.0` `scorer_weight=0.0127`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.0773` `assist_rate=0.127` `provider_weight=0.1006`	-

Luka Vušković D 89 0 0.17 0.09 95% 1 1.1 7.0 47.4 14 FBref Prior D_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost D_service raw output over 7 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0390, quality 0.869)	`market_value_eur=69000000` `quality_factor=0.8692` `boost_factor=1.039` `reference_quality=0.505` `history_matches=9` `history_shrink=0.3571`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0309)	`blend_weight=0.0309` `history_matches=9` `season_overlap_discount=1.0` `per90_to_match_factor=0.9906` `prior_reliability=0.1333` `competition=Premier League`	interceptionWon: 1.33 -> 1.2888 wasFouled: 0.8317 -> 0.806 totalShots: 1.2364 -> 1.2594 totalCross: 0.3254 -> 0.3153 fouls: 1.2324 -> 1.2249 wonTackle: 0.7207 -> 0.7137 onTargetScoringAttempt: 0.2231 -> 0.2162
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` `duel_context={'opponent_faced': 'Igor Thiago', 'matchup_zone': 'Striker vs Two Centre Backs — 1v2 Aerial/Physical Battle'}`	-
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 2 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=2`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.65 (squad xG sum 1.3905, xG scale 1.1859, xA scale 1.2086, team xA 1.07 = 0.65 x lambda)	`scoreline_reset=true` `team_lambda=1.649` `squad_xg_sum=1.3905` `squad_xa_sum=0.8869` `team_expected_assists=1.0718` `team_xa_to_xg_ratio=0.65`	expected_goals: 0.1463 -> 0.1735
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 3 by projected rating, fifa_overall 81.0)	`match_squad_rank=3` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 2 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=2` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.1593` `goal_rate=0.0741` `scorer_weight=0.1388`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.0263` `assist_rate=0.0` `provider_weight=0.0214`	-

Diego Gómez M 84 1 0.20 0.13 56% 2 1.4 7.0 56.5 29 FBref Prior M_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost M_service raw output over 8 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0074, quality 0.732)	`market_value_eur=32000000` `quality_factor=0.732` `boost_factor=1.0074` `reference_quality=0.505` `history_matches=41` `history_shrink=0.1087`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0090)	`blend_weight=0.009` `history_matches=41` `season_overlap_discount=1.0` `per90_to_match_factor=0.9312` `prior_reliability=0.1267` `competition=Premier League`	totalShots: 1.8663 -> 1.8839 wonTackle: 1.1319 -> 1.1393 onTargetScoringAttempt: 0.6873 -> 0.694
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` duel_context={'opponent_faced': 'Michael Kayode', 'matchup_zone': 'Left Wing vs Right Back — Flank Battle', 'duel_win_rate': 0.455, 'pressing_faced': 'HIGH', 'numerical_situation': '1v1', 'xg_multiplier': 0.9626, 'skill_edge': -0.0494, 'height_edge': 0.024, 'attacker_skill': 0.732, 'defender_skill': 0.8419, 'attacker_role': 'LW', 'attacker_mv_price_index': 1.0, 'defender_adjusted_parts': [{'player': 'Michael Kayode', 'market_value': 39000000, 'role': 'RB', 'price_index': 0.7, 'adjusted_quality': 0.8419}], 'xg_override': 0.1698}	aerialWon: 3.4181 -> 1.2 totalShots: 1.8839 -> 1.8134 shotOffTarget: 1.179 -> 1.1349 onTargetScoringAttempt: 0.694 -> 0.668 bigChanceMissed: 0.3824 -> 0.3681 goals: 0.222 -> 0.2137 expectedGoals: 0.1764 -> 0.1698
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 2 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=2`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.65 (squad xG sum 1.3905, xG scale 1.1859, xA scale 1.2086, team xA 1.07 = 0.65 x lambda)	`scoreline_reset=true` `team_lambda=1.649` `squad_xg_sum=1.3905` `squad_xa_sum=0.8869` `team_expected_assists=1.0718` `team_xa_to_xg_ratio=0.65`	expected_goals: 0.1698 -> 0.2014 expected_assists: 0.1099 -> 0.1328
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 4 by projected rating, fifa_overall 78.0)	`match_squad_rank=4` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 2 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=2` `allocation_applied=true` `goals_allocated=1` `goal_probability=0.1824` `goal_rate=0.2137` `scorer_weight=0.2416`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.1244` `assist_rate=0.114` `provider_weight=0.1362`	-

Pascal Groß M 88 0 0.14 0.11 32% 1 0.9 7.1 60.1 33 FBref Prior M_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost M_service raw output over 8 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 0.9969, quality 0.437)	`market_value_eur=2300000` `quality_factor=0.4374` `boost_factor=0.9969` `reference_quality=0.505` `history_matches=28` `history_shrink=0.1515`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0131)	`blend_weight=0.0131` `history_matches=28` `season_overlap_discount=1.0` `per90_to_match_factor=0.9752` `prior_reliability=0.1333` `competition=Premier League`	totalCross: 5.9665 -> 6.0034 totalShots: 1.2941 -> 1.2835 interceptionWon: 0.3873 -> 0.3822 wonTackle: 0.5949 -> 0.5999
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` duel_context={'opponent_faced': 'Keane Lewis-Potter', 'matchup_zone': 'Right Wing vs Left Back — Flank Battle', 'duel_win_rate': 0.366, 'pressing_faced': 'MEDIUM', 'numerical_situation': '1v1', 'xg_multiplier': 0.8327, 'skill_edge': -0.1578, 'height_edge': 0.044, 'attacker_skill': 0.4374, 'defender_skill': 0.7881, 'attacker_role': 'RW', 'attacker_mv_price_index': 1.0, 'defender_adjusted_parts': [{'player': 'Keane Lewis-Potter', 'market_value': 29000000, 'role': 'LB', 'price_index': 0.7, 'adjusted_quality': 0.7881}], 'xg_override': 0.1218}	aerialWon: 0.6008 -> 1.1 totalShots: 1.2936 -> 1.0772 shotOffTarget: 0.7666 -> 0.6383 onTargetScoringAttempt: 0.527 -> 0.4388 goals: 0.1727 -> 0.1438 expectedGoals: 0.1463 -> 0.1218 expectedGoalsOnTarget: 0.09 -> 0.0749 bigChanceMissed: 0.0596 -> 0.0496
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 2 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=2`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.65 (squad xG sum 1.3905, xG scale 1.1859, xA scale 1.2086, team xA 1.07 = 0.65 x lambda)	`scoreline_reset=true` `team_lambda=1.649` `squad_xg_sum=1.3905` `squad_xa_sum=0.8869` `team_expected_assists=1.0718` `team_xa_to_xg_ratio=0.65`	expected_goals: 0.1218 -> 0.1444 expected_assists: 0.191 -> 0.2308
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 2 by projected rating, fifa_overall 79.0)	`match_squad_rank=2` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 2 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=2` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.1345` `goal_rate=0.1438` `scorer_weight=0.1629`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.2061` `assist_rate=0.1392` `provider_weight=0.1846`	-

Yasin Ayari M 85 0 0.19 0.09 116% 2 1.0 6.9 51.3 27 FBref Prior M_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost M_service raw output over 8 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0100, quality 0.786)	`market_value_eur=41000000` `quality_factor=0.7859` `boost_factor=1.01` `reference_quality=0.505` `history_matches=37` `history_shrink=0.119`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0036)	`blend_weight=0.0036` `history_matches=37` `season_overlap_discount=1.0` `per90_to_match_factor=0.9446` `prior_reliability=0.0467` `competition=Premier League`	fouls: 0.4628 -> 0.4709
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` duel_context={'opponent_faced': 'Vitaly Janelt', 'matchup_zone': 'CDM vs CDM', 'duel_win_rate': 0.436, 'pressing_faced': 'MEDIUM', 'numerical_situation': '1v1', 'xg_multiplier': 1.024, 'skill_edge': 0.0643, 'height_edge': -0.048, 'attacker_skill': 0.8177, 'defender_skill': 0.6747, 'attacker_role': 'CDM', 'attacker_mv_price_index': 0.85, 'defender_adjusted_parts': [{'player': 'Vitaly Janelt', 'market_value': 21000000, 'role': 'CDM', 'price_index': 0.85, 'adjusted_quality': 0.6747}], 'xg_override': 0.1606}	aerialWon: 0.4093 -> 0.8 totalShots: 1.6406 -> 1.68 shotOffTarget: 1.1141 -> 1.1408 onTargetScoringAttempt: 0.5208 -> 0.5333 expectedGoalsOnTarget: 0.3026 -> 0.3099
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 2 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=2`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.65 (squad xG sum 1.3905, xG scale 1.1859, xA scale 1.2086, team xA 1.07 = 0.65 x lambda)	`scoreline_reset=true` `team_lambda=1.649` `squad_xg_sum=1.3905` `squad_xa_sum=0.8869` `team_expected_assists=1.0718` `team_xa_to_xg_ratio=0.65`	expected_goals: 0.1606 -> 0.1905 expected_assists: 0.0644 -> 0.0778
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 5 by projected rating, fifa_overall 79.0)	`match_squad_rank=5` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 2 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=2` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.1735` `goal_rate=0.1502` `scorer_weight=0.1811`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.0749` `assist_rate=0.0722` `provider_weight=0.0831`	-

Chema Andrés M 77 0 0.08 0.04 68% 1 0.3 6.8 35.1 12 XGBoost M_service

Step-by-step (9 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost M_service raw output over 8 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0211, quality 0.660)	`market_value_eur=23000000` `quality_factor=0.6595` `boost_factor=1.0211` `reference_quality=0.505` `history_matches=6` `history_shrink=0.4545`	-
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` duel_context={'opponent_faced': 'Vitaly Janelt', 'matchup_zone': 'CDM vs CDM', 'duel_win_rate': 0.453, 'pressing_faced': 'MEDIUM', 'numerical_situation': '1v1', 'xg_multiplier': 1.0484, 'skill_edge': 0.0089, 'height_edge': 0.024, 'attacker_skill': 0.6946, 'defender_skill': 0.6747, 'attacker_role': 'CDM', 'attacker_mv_price_index': 0.85, 'defender_adjusted_parts': [{'player': 'Vitaly Janelt', 'market_value': 21000000, 'role': 'CDM', 'price_index': 0.85, 'adjusted_quality': 0.6747}], 'xg_override': 0.0636}	aerialWon: 0.9176 -> 1.3 totalShots: 0.6162 -> 0.646 shotOffTarget: 0.4349 -> 0.4559 onTargetScoringAttempt: 0.1813 -> 0.1901 expectedGoalsOnTarget: 0.1336 -> 0.1401
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 2 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=2`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.65 (squad xG sum 1.3905, xG scale 1.1859, xA scale 1.2086, team xA 1.07 = 0.65 x lambda)	`scoreline_reset=true` `team_lambda=1.649` `squad_xg_sum=1.3905` `squad_xa_sum=0.8869` `team_expected_assists=1.0718` `team_xa_to_xg_ratio=0.65`	expected_goals: 0.0636 -> 0.0754 expected_assists: 0.0308 -> 0.0372
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 7 by projected rating, fifa_overall 86.0)	`match_squad_rank=7` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv_filled` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 2 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=2` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.0726` `goal_rate=0.0746` `scorer_weight=0.0827`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.0365` `assist_rate=0.0466` `provider_weight=0.0445`	-

Jack Hinshelwood M 79 0 0.22 0.26 ok 2 1.3 6.8 50.6 27 FBref Prior M_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost M_service raw output over 8 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0100, quality 0.746)	`market_value_eur=34000000` `quality_factor=0.7455` `boost_factor=1.01` `reference_quality=0.505` `history_matches=31` `history_shrink=0.1389`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0042)	`blend_weight=0.0042` `history_matches=31` `season_overlap_discount=1.0` `per90_to_match_factor=0.8776` `prior_reliability=0.0467` `competition=Premier League`	goals: 0.1583 -> 0.1682 totalShots: 1.8318 -> 1.8399 onTargetScoringAttempt: 0.6814 -> 0.6891 wonTackle: 0.7615 -> 0.7689 wasFouled: 0.9029 -> 0.9097
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` duel_context={'opponent_faced': 'Vitaly Janelt, Mamadou Sangare', 'matchup_zone': 'Attacking Mid vs Double Pivot — Tight Space Battle', 'duel_win_rate': 0.236, 'pressing_faced': 'HIGH', 'numerical_situation': '1v2', 'xg_multiplier': 0.9944, 'skill_edge': -0.0018, 'height_edge': -0.002, 'attacker_skill': 0.7569, 'defender_skill': 0.7609, 'attacker_role': 'CAM', 'attacker_mv_price_index': 0.95, 'defender_adjusted_parts': [{'player': 'Vitaly Janelt', 'market_value': 21000000, 'role': 'CDM', 'price_index': 0.85, 'adjusted_quality': 0.6747}, {'player': 'Mamadou Sangare', 'market_value': 49000000, 'role': 'CDM', 'price_index': 0.85, 'adjusted_quality': 0.847}], 'xg_override': 0.1839}	aerialWon: 0.5215 -> 1.1 totalShots: 1.8399 -> 1.8296 shotOffTarget: 1.1504 -> 1.144
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 2 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=2`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.65 (squad xG sum 1.3905, xG scale 1.1859, xA scale 1.2086, team xA 1.07 = 0.65 x lambda)	`scoreline_reset=true` `team_lambda=1.649` `squad_xg_sum=1.3905` `squad_xa_sum=0.8869` `team_expected_assists=1.0718` `team_xa_to_xg_ratio=0.65`	expected_goals: 0.1839 -> 0.2181 expected_assists: 0.0653 -> 0.0789
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 9 by projected rating, fifa_overall 77.0)	`match_squad_rank=9` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 2 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=2` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.196` `goal_rate=0.1673` `scorer_weight=0.2026`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.0759` `assist_rate=0.0646` `provider_weight=0.0784`	-

Charalampos Kostoulas F 77 1 0.31 0.13 148% 2 1.1 6.8 45.4 30 FBref Prior F_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost F_service raw output over 7 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0125, quality 0.796)	`market_value_eur=43000000` `quality_factor=0.7957` `boost_factor=1.0125` `reference_quality=0.505` `history_matches=30` `history_shrink=0.1429`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0074)	`blend_weight=0.0074` `history_matches=30` `season_overlap_discount=1.0` `per90_to_match_factor=0.8606` `prior_reliability=0.08` `competition=Premier League`	wasFouled: 1.3645 -> 1.3544
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` duel_context={'opponent_faced': 'Nathan Collins, Kristoffer Ajer', 'matchup_zone': 'Striker vs Two Centre Backs — 1v2 Aerial/Physical Battle', 'duel_win_rate': 0.18, 'pressing_faced': 'MEDIUM', 'numerical_situation': '1v2', 'xg_multiplier': 0.9638, 'skill_edge': 0.0134, 'height_edge': -0.038, 'attacker_skill': 0.7957, 'defender_skill': 0.7659, 'attacker_role': 'ST', 'attacker_mv_price_index': 1.0, 'defender_adjusted_parts': [{'player': 'Nathan Collins', 'market_value': 29000000, 'role': 'CB', 'price_index': 0.75, 'adjusted_quality': 0.7736}, {'player': 'Kristoffer Ajer', 'market_value': 27000000, 'role': 'CB', 'price_index': 0.75, 'adjusted_quality': 0.7581}], 'xg_override': 0.2654}	totalShots: 2.0809 -> 2.0056 shotOffTarget: 1.3439 -> 1.2953 onTargetScoringAttempt: 0.7362 -> 0.7095 goals: 0.2895 -> 0.279 expectedGoals: 0.2754 -> 0.2654 expectedGoalsOnTarget: 0.269 -> 0.2593 bigChanceMissed: 0.2596 -> 0.2502
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 2 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=2`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.65 (squad xG sum 1.3905, xG scale 1.1859, xA scale 1.2086, team xA 1.07 = 0.65 x lambda)	`scoreline_reset=true` `team_lambda=1.649` `squad_xg_sum=1.3905` `squad_xa_sum=0.8869` `team_expected_assists=1.0718` `team_xa_to_xg_ratio=0.65`	expected_goals: 0.2654 -> 0.3147 expected_assists: 0.0622 -> 0.0752
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 12 by projected rating, fifa_overall 73.0)	`match_squad_rank=12` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 2 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=2` `allocation_applied=true` `goals_allocated=1` `goal_probability=0.27` `goal_rate=0.279` `scorer_weight=0.3283`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.0724` `assist_rate=0.277` `provider_weight=0.094`	-

Team totals 2 1.65 16 6.9

Bench Candidates (7 modeled substitutes with empirical playing time)

Player	Pos	Exp Min	App Prob	Substitute Modeling Details
Tom McGill	G	4.4m	10.9%	Market-value substitute prior
Mats Wieffer	D	14.5m	23.6%	Market-value substitute prior
Pascal Struijk	D	14.4m	33.9%	Market-value substitute prior
Malick Yalcouyé	M	13.0m	51.7%	Market-value substitute prior
Matt O'Riley	M	22.6m	52.7%	Market-value substitute prior
Yankuba Minteh	F	26.8m	62.9%	Market-value substitute prior

Brentford — projected line vs season baseline (4-2-3-1)

Player Pos Min Proj G Proj xG Season xG/g Proj Sh Season Sh/g Proj Rtg Season Rtg Rec Model / FBref Prior

Caoimhin Kelleher G 90 0 0.00 - n/a 0.0 6.8 59.4 33 FBref Prior G_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost G_service raw output over 3 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0060, quality 0.710)	`market_value_eur=29000000` `quality_factor=0.71` `boost_factor=1.006` `reference_quality=0.505` `history_matches=46` `history_shrink=0.098`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0085)	`blend_weight=0.0085` `history_matches=46` `season_overlap_discount=1.0` `per90_to_match_factor=1.0` `prior_reliability=0.1333` `competition=Premier League`	-
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` `duel_context={'opponent_faced': 'Charalampos Kostoulas, Jack Hinshelwood', 'matchup_zone': 'Goalkeeper shot-stopping context', 'opponent_lambda': 1.8777}`	-
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 1 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=1`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.47 (squad xG sum 1.0237, xG scale 1.4379, xA scale 1.4069, team xA 1.18 = 0.80 x lambda)	`scoreline_reset=true` `team_lambda=1.472` `squad_xg_sum=1.0237` `squad_xa_sum=0.837` `team_expected_assists=1.1776` `team_xa_to_xg_ratio=0.8`	shots_faced: 17.8 -> 15.7 expected_saves: 4.35 -> 3.85 high_pressure_saves: 1.5 -> 1.32
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 10 by projected rating, fifa_overall 80.0)	`match_squad_rank=10` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 1 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=1` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.0` `scorer_weight=0.0` `squad_weight_total=1.6042`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.0` `provider_weight=0.0` `squad_weight_total=1.0952`	-

Keane Lewis-Potter D 82 0 0.15 0.10 49% 1 1.0 6.7 57.0 34 FBref Prior D_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost D_service raw output over 7 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0061, quality 0.710)	`market_value_eur=29000000` `quality_factor=0.71` `boost_factor=1.0061` `reference_quality=0.505` `history_matches=45` `history_shrink=0.1`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0087)	`blend_weight=0.0087` `history_matches=45` `season_overlap_discount=1.0` `per90_to_match_factor=0.9078` `prior_reliability=0.1333` `competition=Premier League`	totalCross: 2.9299 -> 2.9438 wasFouled: 1.2226 -> 1.212
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` `duel_context={'opponent_faced': 'Pascal Groß', 'matchup_zone': 'Right Wing vs Left Back — Flank Battle'}`	-
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 1 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=1`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.47 (squad xG sum 1.0237, xG scale 1.4379, xA scale 1.4069, team xA 1.18 = 0.80 x lambda)	`scoreline_reset=true` `team_lambda=1.472` `squad_xg_sum=1.0237` `squad_xa_sum=0.837` `team_expected_assists=1.1776` `team_xa_to_xg_ratio=0.8`	expected_goals: 0.1063 -> 0.1528 expected_assists: 0.0832 -> 0.1171
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 13 by projected rating, fifa_overall 77.0)	`match_squad_rank=13` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 1 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=1` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.1417` `goal_rate=0.1719` `scorer_weight=0.191`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.1105` `assist_rate=0.0831` `provider_weight=0.117`	-

Nathan Collins D 80 0 0.14 0.06 149% 1 0.5 6.8 43.7 31 FBref Prior D_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost D_service raw output over 7 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0072, quality 0.710)	`market_value_eur=29000000` `quality_factor=0.71` `boost_factor=1.0072` `reference_quality=0.505` `history_matches=38` `history_shrink=0.1163`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0101)	`blend_weight=0.0101` `history_matches=38` `season_overlap_discount=1.0` `per90_to_match_factor=0.8846` `prior_reliability=0.1333` `competition=Premier League`	totalShots: 0.8522 -> 0.8659
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` `duel_context={'opponent_faced': 'Charalampos Kostoulas', 'matchup_zone': 'Striker vs Two Centre Backs — 1v2 Aerial/Physical Battle'}`	-
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 1 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=1`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.47 (squad xG sum 1.0237, xG scale 1.4379, xA scale 1.4069, team xA 1.18 = 0.80 x lambda)	`scoreline_reset=true` `team_lambda=1.472` `squad_xg_sum=1.0237` `squad_xa_sum=0.837` `team_expected_assists=1.1776` `team_xa_to_xg_ratio=0.8`	expected_goals: 0.0956 -> 0.1375 expected_assists: 0.0466 -> 0.0656
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 11 by projected rating, fifa_overall 79.0)	`match_squad_rank=11` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 1 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=1` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.1285` `goal_rate=0.0124` `scorer_weight=0.11`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.0635` `assist_rate=0.0243` `provider_weight=0.0525`	-

Kristoffer Ajer D 89 0 0.04 0.01 241% 0 0.2 6.8 49.3 33 FBref Prior D_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost D_service raw output over 7 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0069, quality 0.694)	`market_value_eur=27000000` `quality_factor=0.6941` `boost_factor=1.0069` `reference_quality=0.505` `history_matches=36` `history_shrink=0.122`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0106)	`blend_weight=0.0106` `history_matches=36` `season_overlap_discount=1.0` `per90_to_match_factor=0.9866` `prior_reliability=0.1333` `competition=Premier League`	wonTackle: 0.9715 -> 0.9664
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` `duel_context={'opponent_faced': 'Charalampos Kostoulas', 'matchup_zone': 'Striker vs Two Centre Backs — 1v2 Aerial/Physical Battle'}`	-
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 1 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=1`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.47 (squad xG sum 1.0237, xG scale 1.4379, xA scale 1.4069, team xA 1.18 = 0.80 x lambda)	`scoreline_reset=true` `team_lambda=1.472` `squad_xg_sum=1.0237` `squad_xa_sum=0.837` `team_expected_assists=1.1776` `team_xa_to_xg_ratio=0.8`	expected_goals: 0.0266 -> 0.0382 expected_assists: 0.0312 -> 0.0439
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 9 by projected rating, fifa_overall 77.0)	`match_squad_rank=9` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 1 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=1` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.0375` `goal_rate=0.0931` `scorer_weight=0.0478`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.043` `assist_rate=0.0` `provider_weight=0.0351`	-

Michael Kayode D 85 0 0.07 0.05 45% 1 0.3 6.7 58.3 33 FBref Prior D_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost D_service raw output over 7 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0080, quality 0.775)	`market_value_eur=39000000` `quality_factor=0.7754` `boost_factor=1.008` `reference_quality=0.505` `history_matches=46` `history_shrink=0.098`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0076)	`blend_weight=0.0076` `history_matches=46` `season_overlap_discount=1.0` `per90_to_match_factor=0.9412` `prior_reliability=0.12` `competition=Premier League`	totalShots: 0.5697 -> 0.5854 wasFouled: 0.9025 -> 0.9116
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` `duel_context={'opponent_faced': 'Diego Gómez', 'matchup_zone': 'Left Wing vs Right Back — Flank Battle'}`	-
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 1 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=1`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.47 (squad xG sum 1.0237, xG scale 1.4379, xA scale 1.4069, team xA 1.18 = 0.80 x lambda)	`scoreline_reset=true` `team_lambda=1.472` `squad_xg_sum=1.0237` `squad_xa_sum=0.837` `team_expected_assists=1.1776` `team_xa_to_xg_ratio=0.8`	expected_goals: 0.0485 -> 0.0697 expected_assists: 0.0679 -> 0.0955
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 12 by projected rating, fifa_overall 81.0)	`match_squad_rank=12` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 1 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=1` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.0673` `goal_rate=0.0659` `scorer_weight=0.082`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.0911` `assist_rate=0.0154` `provider_weight=0.0764`	-

Vitaly Janelt M 87 0 0.07 0.04 52% 1 0.4 7.0 55.0 26 FBref Prior M_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost M_service raw output over 8 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0054, quality 0.641)	`market_value_eur=21000000` `quality_factor=0.6408` `boost_factor=1.0054` `reference_quality=0.505` `history_matches=33` `history_shrink=0.1316`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0114)	`blend_weight=0.0114` `history_matches=33` `season_overlap_discount=1.0` `per90_to_match_factor=0.9709` `prior_reliability=0.1333` `competition=Premier League`	totalCross: 2.1767 -> 2.2017 fouls: 1.3028 -> 1.3156 interceptionWon: 2.891 -> 2.9023 wonTackle: 1.3832 -> 1.3896
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` duel_context={'opponent_faced': 'Yasin Ayari', 'matchup_zone': 'CDM vs CDM', 'duel_win_rate': 0.404, 'pressing_faced': 'HIGH', 'numerical_situation': '1v1', 'xg_multiplier': 0.976, 'skill_edge': -0.0643, 'height_edge': 0.048, 'attacker_skill': 0.6747, 'defender_skill': 0.8177, 'attacker_role': 'CDM', 'attacker_mv_price_index': 0.85, 'defender_adjusted_parts': [{'player': 'Yasin Ayari', 'market_value': 41000000, 'role': 'CDM', 'price_index': 0.85, 'adjusted_quality': 0.8177}], 'xg_override': 0.0464}	aerialWon: 0.7577 -> 1.1 totalShots: 0.5908 -> 0.5766 shotOffTarget: 0.4326 -> 0.4222
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 1 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=1`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.47 (squad xG sum 1.0237, xG scale 1.4379, xA scale 1.4069, team xA 1.18 = 0.80 x lambda)	`scoreline_reset=true` `team_lambda=1.472` `squad_xg_sum=1.0237` `squad_xa_sum=0.837` `team_expected_assists=1.1776` `team_xa_to_xg_ratio=0.8`	expected_goals: 0.0464 -> 0.0667 expected_assists: 0.0744 -> 0.1047
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 3 by projected rating, fifa_overall 78.0)	`match_squad_rank=3` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 1 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=1` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.0645` `goal_rate=0.0918` `scorer_weight=0.0834`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.0994` `assist_rate=0.0605` `provider_weight=0.093`	-

Mathias Jensen M 76 0 0.10 0.06 65% 1 0.9 6.9 58.8 30 FBref Prior M_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost M_service raw output over 8 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 0.9999, quality 0.502)	`market_value_eur=8300000` `quality_factor=0.5023` `boost_factor=0.9999` `reference_quality=0.505` `history_matches=41` `history_shrink=0.1087`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0033)	`blend_weight=0.0033` `history_matches=41` `season_overlap_discount=1.0` `per90_to_match_factor=0.8488` `prior_reliability=0.0467` `competition=Premier League`	fouls: 0.7784 -> 0.7878 totalCross: 4.0992 -> 4.1057
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` duel_context={'opponent_faced': 'Olivier Boscagli', 'matchup_zone': 'Left Wing vs Right Back — Flank Battle', 'duel_win_rate': 0.392, 'pressing_faced': 'LOW', 'numerical_situation': '1v1', 'xg_multiplier': 0.8703, 'skill_edge': -0.0842, 'height_edge': -0.004, 'attacker_skill': 0.5023, 'defender_skill': 0.6894, 'attacker_role': 'LW', 'attacker_mv_price_index': 1.0, 'defender_adjusted_parts': [{'player': 'Olivier Boscagli', 'market_value': 18500000, 'role': 'RB', 'price_index': 0.7, 'adjusted_quality': 0.6894}], 'xg_override': 0.0678}	aerialWon: 0.1476 -> 1.1 totalShots: 1.0992 -> 0.9566 shotOffTarget: 0.7398 -> 0.6438 onTargetScoringAttempt: 0.358 -> 0.3116 bigChanceMissed: 0.111 -> 0.0966 expectedGoals: 0.0779 -> 0.0678 goals: 0.0774 -> 0.0674
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 1 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=1`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.47 (squad xG sum 1.0237, xG scale 1.4379, xA scale 1.4069, team xA 1.18 = 0.80 x lambda)	`scoreline_reset=true` `team_lambda=1.472` `squad_xg_sum=1.0237` `squad_xa_sum=0.837` `team_expected_assists=1.1776` `team_xa_to_xg_ratio=0.8`	expected_goals: 0.0678 -> 0.0975 expected_assists: 0.1377 -> 0.1937
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 5 by projected rating, fifa_overall 80.0)	`match_squad_rank=5` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 1 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=1` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.0929` `goal_rate=0.0674` `scorer_weight=0.0972`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.1761` `assist_rate=0.1071` `provider_weight=0.1621`	-

Yehor Yarmoliuk M 75 0 0.07 0.03 134% 1 0.3 6.8 56.5 31 FBref Prior M_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost M_service raw output over 8 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0065, quality 0.718)	`market_value_eur=30000000` `quality_factor=0.7176` `boost_factor=1.0065` `reference_quality=0.505` `history_matches=44` `history_shrink=0.102`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0022)	`blend_weight=0.0022` `history_matches=44` `season_overlap_discount=1.0` `per90_to_match_factor=0.8307` `prior_reliability=0.0333` `competition=Premier League`	-
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` duel_context={'opponent_faced': 'Yasin Ayari, Chema Andrés', 'matchup_zone': 'Attacking Mid vs Double Pivot — Tight Space Battle', 'duel_win_rate': 0.224, 'pressing_faced': 'HIGH', 'numerical_situation': '1v2', 'xg_multiplier': 0.9762, 'skill_edge': -0.0122, 'height_edge': -0.004, 'attacker_skill': 0.7291, 'defender_skill': 0.7562, 'attacker_role': 'CAM', 'attacker_mv_price_index': 0.95, 'defender_adjusted_parts': [{'player': 'Yasin Ayari', 'market_value': 41000000, 'role': 'CDM', 'price_index': 0.85, 'adjusted_quality': 0.8177}, {'player': 'Chema Andrés', 'market_value': 23000000, 'role': 'CDM', 'price_index': 0.85, 'adjusted_quality': 0.6946}], 'xg_override': 0.0483}	aerialWon: 1.1343 -> 1.1 totalShots: 0.7063 -> 0.6895 shotOffTarget: 0.4951 -> 0.4833 onTargetScoringAttempt: 0.2087 -> 0.2037
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 1 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=1`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.47 (squad xG sum 1.0237, xG scale 1.4379, xA scale 1.4069, team xA 1.18 = 0.80 x lambda)	`scoreline_reset=true` `team_lambda=1.472` `squad_xg_sum=1.0237` `squad_xa_sum=0.837` `team_expected_assists=1.1776` `team_xa_to_xg_ratio=0.8`	expected_goals: 0.0483 -> 0.0694 expected_assists: 0.0686 -> 0.0965
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 8 by projected rating, fifa_overall 78.0)	`match_squad_rank=8` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 1 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=1` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.067` `goal_rate=0.0557` `scorer_weight=0.0746`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.092` `assist_rate=0.067` `provider_weight=0.0952`	-

Mamadou Sangare M 78 0 0.09 0.03 159% 1 0.6 6.9 44.6 14 FBref Prior M_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost M_service raw output over 8 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0338, quality 0.821)	`market_value_eur=49000000` `quality_factor=0.8206` `boost_factor=1.0338` `reference_quality=0.505` `history_matches=9` `history_shrink=0.3571`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0279)	`blend_weight=0.0279` `history_matches=9` `season_overlap_discount=1.0` `per90_to_match_factor=0.8648` `prior_reliability=0.12` `competition=Premier League`	totalShots: 0.8625 -> 0.891 interceptionWon: 1.9338 -> 1.9602 totalCross: 0.2405 -> 0.2606 wasFouled: 0.7333 -> 0.753 goalAssist: 0.0543 -> 0.0662 wonTackle: 1.7547 -> 1.746 onTargetScoringAttempt: 0.2075 -> 0.215 fouls: 1.1889 -> 1.1959
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` duel_context={'opponent_faced': 'Yasin Ayari', 'matchup_zone': 'CDM vs CDM', 'duel_win_rate': 0.469, 'pressing_faced': 'HIGH', 'numerical_situation': '1v1', 'xg_multiplier': 1.0723, 'skill_edge': 0.0132, 'height_edge': 0.036, 'attacker_skill': 0.847, 'defender_skill': 0.8177, 'attacker_role': 'CDM', 'attacker_mv_price_index': 0.85, 'defender_adjusted_parts': [{'player': 'Yasin Ayari', 'market_value': 41000000, 'role': 'CDM', 'price_index': 0.85, 'adjusted_quality': 0.8177}], 'xg_override': 0.0618}	aerialWon: 1.4109 -> 1.1 totalShots: 0.891 -> 0.9554 shotOffTarget: 0.655 -> 0.7024 onTargetScoringAttempt: 0.215 -> 0.2305
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 1 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=1`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.47 (squad xG sum 1.0237, xG scale 1.4379, xA scale 1.4069, team xA 1.18 = 0.80 x lambda)	`scoreline_reset=true` `team_lambda=1.472` `squad_xg_sum=1.0237` `squad_xa_sum=0.837` `team_expected_assists=1.1776` `team_xa_to_xg_ratio=0.8`	expected_goals: 0.0618 -> 0.0889 expected_assists: 0.0477 -> 0.0671
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 6 by projected rating, fifa_overall 81.0)	`match_squad_rank=6` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 1 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=1` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.0851` `goal_rate=0.0661` `scorer_weight=0.0923`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.0649` `assist_rate=0.0662` `provider_weight=0.0798`	-

Mikkel Damsgaard M 80 0 0.12 0.06 105% 1 0.8 7.0 55.5 32 FBref Prior M_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost M_service raw output over 8 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0067, quality 0.710)	`market_value_eur=29000000` `quality_factor=0.71` `boost_factor=1.0067` `reference_quality=0.505` `history_matches=41` `history_shrink=0.1087`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0042)	`blend_weight=0.0042` `history_matches=41` `season_overlap_discount=1.0` `per90_to_match_factor=0.8942` `prior_reliability=0.06` `competition=Premier League`	totalCross: 3.3114 -> 3.3311
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` duel_context={'opponent_faced': 'Maxim De Cuyper', 'matchup_zone': 'Right Wing vs Left Back — Flank Battle', 'duel_win_rate': 0.433, 'pressing_faced': 'MEDIUM', 'numerical_situation': '1v1', 'xg_multiplier': 0.9307, 'skill_edge': -0.0351, 'height_edge': -0.012, 'attacker_skill': 0.71, 'defender_skill': 0.7881, 'attacker_role': 'RW', 'attacker_mv_price_index': 1.0, 'defender_adjusted_parts': [{'player': 'Maxim De Cuyper', 'market_value': 29000000, 'role': 'LB', 'price_index': 0.7, 'adjusted_quality': 0.7881}], 'xg_override': 0.0855}	aerialWon: 0.8051 -> 1.1 totalShots: 1.2208 -> 1.1362 shotOffTarget: 0.822 -> 0.765 onTargetScoringAttempt: 0.3981 -> 0.3705 goals: 0.0998 -> 0.0929 expectedGoals: 0.0919 -> 0.0855
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 1 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=1`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.47 (squad xG sum 1.0237, xG scale 1.4379, xA scale 1.4069, team xA 1.18 = 0.80 x lambda)	`scoreline_reset=true` `team_lambda=1.472` `squad_xg_sum=1.0237` `squad_xa_sum=0.837` `team_expected_assists=1.1776` `team_xa_to_xg_ratio=0.8`	expected_goals: 0.0855 -> 0.1229 expected_assists: 0.1427 -> 0.2008
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 1 by projected rating, fifa_overall 80.0)	`match_squad_rank=1` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 1 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=1` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.1156` `goal_rate=0.0929` `scorer_weight=0.1296`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=1` `assist_probability=0.1819` `assist_rate=0.1073` `provider_weight=0.1639`	-

Igor Thiago F 89 1 0.47 0.46 ok 2 2.0 7.0 58.0 34 FBref Prior F_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost F_service raw output over 7 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0105, quality 0.868)	`market_value_eur=68000000` `quality_factor=0.8677` `boost_factor=1.0105` `reference_quality=0.505` `history_matches=47` `history_shrink=0.0962`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0079)	`blend_weight=0.0079` `history_matches=47` `season_overlap_discount=1.0` `per90_to_match_factor=0.992` `prior_reliability=0.1267` `competition=Premier League`	totalShots: 2.0495 -> 2.062 fouls: 2.0664 -> 2.0542 wasFouled: 1.2622 -> 1.2522 wonTackle: 0.7118 -> 0.7062
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` duel_context={'opponent_faced': 'Ferdi Kadıoğlu, Luka Vušković', 'matchup_zone': 'Striker vs Two Centre Backs — 1v2 Aerial/Physical Battle', 'duel_win_rate': 0.231, 'pressing_faced': 'ULTRA_HIGH', 'numerical_situation': '1v2', 'xg_multiplier': 1.0462, 'skill_edge': 0.0014, 'height_edge': 0.03, 'attacker_skill': 0.8677, 'defender_skill': 0.8645, 'attacker_role': 'ST', 'attacker_mv_price_index': 1.0, 'defender_adjusted_parts': [{'player': 'Ferdi Kadıoğlu', 'market_value': 41000000, 'role': 'CB', 'price_index': 0.75, 'adjusted_quality': 0.839}, {'player': 'Luka Vušković', 'market_value': 69000000, 'role': 'CB', 'price_index': 0.75, 'adjusted_quality': 0.89}], 'xg_override': 0.324}	totalShots: 2.062 -> 2.1573 shotOffTarget: 1.3359 -> 1.3976 onTargetScoringAttempt: 0.7162 -> 0.7493 bigChanceMissed: 0.5143 -> 0.5381 goals: 0.3501 -> 0.3663 expectedGoals: 0.3097 -> 0.324 expectedGoalsOnTarget: 0.291 -> 0.3044
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 1 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=1`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.47 (squad xG sum 1.0237, xG scale 1.4379, xA scale 1.4069, team xA 1.18 = 0.80 x lambda)	`scoreline_reset=true` `team_lambda=1.472` `squad_xg_sum=1.0237` `squad_xa_sum=0.837` `team_expected_assists=1.1776` `team_xa_to_xg_ratio=0.8`	expected_goals: 0.324 -> 0.466 expected_assists: 0.0814 -> 0.1145
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 4 by projected rating, fifa_overall 82.0)	`match_squad_rank=4` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 1 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=1` `allocation_applied=true` `goals_allocated=1` `goal_probability=0.3725` `goal_rate=0.3663` `scorer_weight=0.5187`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.1082` `assist_rate=0.238` `provider_weight=0.1431`	-

Team totals 1 1.47 12 6.9

Bench Candidates (7 modeled substitutes with empirical playing time)

Player	Pos	Exp Min	App Prob	Substitute Modeling Details
Hákon Rafn Valdimarsson	G	4.4m	10.9%	Market-value substitute prior
El Hadji Malick Diouf	D	14.8m	35.6%	Market-value substitute prior
Sepp van den Berg	D	14.8m	35.3%	Market-value substitute prior
Antoni Milambo	M	22.6m	52.5%	Market-value substitute prior
Josh Dasilva	M	22.0m	49.9%	Market-value substitute prior
Kevin Schade	F	26.9m	36.6%	Market-value substitute prior

Antagonist / tactical summary

Midfield is evenly contested — Yasin Ayari vs Mamadou Sangare will be the pivotal duel. Total antagonist xG — Home: 0.97, Away: 0.63. Home danger zone: Striker vs Two Centre Backs — 1v2 Aerial/Physical Battle. Away danger zone: Striker vs Two Centre Backs — 1v2 Aerial/Physical Battle.

Midfield — possession 49.6% / 50.4%, dominance ratio 0.50, key duel: Yasin Ayari vs Mamadou Sangare

Diego Gómez xG 0.17Pascal Groß xG 0.12Jack Hinshelwood xG 0.18Mathias Jensen xG 0.07Mikkel Damsgaard xG 0.09Yehor Yarmoliuk xG 0.05

Summary

Antagonist matchup analysis for Brighton vs Brentford. Most likely scoreline: 2-1. Lambdas derived from 6 home and 6 away individual player duels. Head-to-head: Brighton 4W-3D-3L vs Brentford over 10 meetings.

Tactical parity between 4-2-3-1 and 4-2-3-1. Match outcome driven primarily by individual player quality. Midfield is evenly contested — Yasin Ayari vs Mamadou Sangare will be the pivotal duel. Total antagonist xG — Home: 0.97, Away: 0.63. Home danger zone: Striker vs Two Centre Backs — 1v2 Aerial/Physical Battle. Away danger zone: Striker vs Two Centre Backs — 1v2 Aerial/Physical Battle.

Sunderland vs Chelsea

Premier League · 20261108 · 4-2-3-1 vs 4-2-3-1

Predicted 1-2Winner: ChelseaWinning H 2-1 (6.7%)Winning A 1-2 (9.6%)Confidence 0.68λ home/away 1.20 / 1.73Most-likely P 9.6%

Actual vs predicted: result not available yet (upcoming fixture).

XGBoost Reasoning & FBref Dataset Tracing

Antagonist + XGBoost match analysis for Sunderland vs Chelsea. Player statistics regressed using position-specific XGBoost models (G, D, M, F) with 2026-2027 FBref priors. Micro-duels resolved across 4-2-3-1 vs 4-2-3-1 (duel lambdas: Home 1.388, Away 1.919). Final calibrated Poisson lambdas: Home 1.201, Away 1.733 (most likely score 1-2).

Home Starters 11/11 XGBoost (9 FBref) Away Starters 11/11 XGBoost (10 FBref) Bench Modeled H: 7 / A: 7 Duel λ (H/A) 1.39 / 1.92 Final λ (H/A) 1.20 / 1.73

FBref Team Priors: Sunderland (Gls/90: 1.0, GA/90: 1.0 → E[λ]: 2.414) vs Chelsea (Gls/90: 3.5, GA/90: 3.5 → E[λ]: 2.414) | Blend Weight: 0.192

Modeling Pipeline Stages (5 layers)

1. XGBoost player regressions for volume & event rates with market-value quality scaling
2. Multi-source prior blending (FBref 2026-2027 tables + Sofascore season history)
3. Antagonist formation duel matrix resolving 1v1 and zonal matchups
4. Team-level prior calibration (FBref possession and attack/defense efficiencies)
5. Bivariate Poisson matrix with Dixon-Coles low-score correction

Standing & form context

Metric	Sunderland	Chelsea
League position	20.00	4.00
Points per game	0.88	1.61
Attack index	0.86	1.18
Defense (concede) index	1.04	1.22
Form multiplier	0.96	1.04
Standing λ multiplier	0.95	1.17

Lambda decomposition (how the goal rates were built)

Stage	Sunderland λ	Chelsea λ
FBref Squad Prior E[λ] (weight 0.192)	2.414	2.414
Pre-mismatch	1.356	2.096
Pre-competitive-floor	1.327	2.268
Pre-calibration (Micro-duels)	1.388	1.919
Final λ	1.201	1.733

competition calibration scale 0.85competition observed total goals 2.781quality gap applied Truemismatch score pressure applied Truecompetitive underdog floor applied Falsestage discount 1.0

Scoreline distribution & markets

Top 5 scorelines	P
1-1	11.1%
1-2	9.6%
0-1	9.1%
0-2	8.0%
2-1	6.7%

Home win	26.4%
Draw	23.8%
Away win	49.8%
Over 2.5	56.2%
BTTS yes	57.6%

Pregame fair line — favorite Chelsea (49.8%), fair total 2.93, fair spread -0.53

Projected players vs season baseline

Proj = projected for this match · Season = this season's per-game average from match data · deviation badge compares projected xG to the player's season xG/game.

Sunderland — projected line vs season baseline (4-2-3-1)

Player Pos Min Proj G Proj xG Season xG/g Proj Sh Season Sh/g Proj Rtg Season Rtg Rec Model / FBref Prior

Robin Roefs G 90 0 0.00 - n/a 0.0 7.0 55.7 31 FBref Prior G_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost G_service raw output over 3 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0067, quality 0.725)	`market_value_eur=31000000` `quality_factor=0.7249` `boost_factor=1.0067` `reference_quality=0.505` `history_matches=44` `history_shrink=0.102`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0088)	`blend_weight=0.0088` `history_matches=44` `season_overlap_discount=1.0` `per90_to_match_factor=1.0` `prior_reliability=0.1333` `competition=Premier League`	saves: 2.7914 -> 2.7711
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` `duel_context={'opponent_faced': 'João Pedro, Cole Palmer', 'matchup_zone': 'Goalkeeper shot-stopping context', 'opponent_lambda': 1.9191}`	-
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 1 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=1`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.20 (squad xG sum 1.1254, xG scale 1.0672, xA scale 1.0672, team xA 0.82 = 0.68 x lambda)	`scoreline_reset=true` `team_lambda=1.201` `squad_xg_sum=1.1254` `squad_xa_sum=0.7648` `team_expected_assists=0.8162` `team_xa_to_xg_ratio=0.6796`	shots_faced: 18.2 -> 16.5 expected_saves: 4.45 -> 4.04 high_pressure_saves: 1.54 -> 1.39
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 2 by projected rating, fifa_overall 81.0)	`match_squad_rank=2` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 1 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=1` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.0` `scorer_weight=0.0` `squad_weight_total=1.165`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.0` `provider_weight=0.0` `squad_weight_total=0.7435`	-

Kevin Danso D 79 0 0.07 0.03 121% 1 0.5 6.6 36.7 28 XGBoost D_service

Step-by-step (9 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost D_service raw output over 7 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0053, quality 0.625)	`market_value_eur=19400000` `quality_factor=0.6251` `boost_factor=1.0053` `reference_quality=0.505` `history_matches=29` `history_shrink=0.1471`	-
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` `duel_context={'opponent_faced': 'João Pedro', 'matchup_zone': 'Striker vs Two Centre Backs — 1v2 Aerial/Physical Battle'}`	-
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 1 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=1`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.20 (squad xG sum 1.1254, xG scale 1.0672, xA scale 1.0672, team xA 0.82 = 0.68 x lambda)	`scoreline_reset=true` `team_lambda=1.201` `squad_xg_sum=1.1254` `squad_xa_sum=0.7648` `team_expected_assists=0.8162` `team_xa_to_xg_ratio=0.6796`	-
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 18 by projected rating, fifa_overall 79.0)	`match_squad_rank=18` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 1 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=1` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.0656` `goal_rate=0.0` `scorer_weight=0.0543`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.0255` `assist_rate=0.0` `provider_weight=0.0206`	-

Nordi Mukiele D 83 0 0.10 0.07 53% 1 0.8 6.9 51.4 31 FBref Prior D_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost D_service raw output over 7 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0033, quality 0.602)	`market_value_eur=17100000` `quality_factor=0.6016` `boost_factor=1.0033` `reference_quality=0.505` `history_matches=39` `history_shrink=0.1136`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0049)	`blend_weight=0.0049` `history_matches=39` `season_overlap_discount=1.0` `per90_to_match_factor=0.9248` `prior_reliability=0.0667` `competition=Premier League`	totalCross: 2.6647 -> 2.6561 wonTackle: 1.2388 -> 1.2327 totalShots: 1.1747 -> 1.1689
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` `duel_context={'opponent_faced': 'Jordan Henderson', 'matchup_zone': 'Right Wing vs Left Back — Flank Battle'}`	-
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 1 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=1`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.20 (squad xG sum 1.1254, xG scale 1.0672, xA scale 1.0672, team xA 0.82 = 0.68 x lambda)	`scoreline_reset=true` `team_lambda=1.201` `squad_xg_sum=1.1254` `squad_xa_sum=0.7648` `team_expected_assists=0.8162` `team_xa_to_xg_ratio=0.6796`	expected_goals: 0.0937 -> 0.1
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 5 by projected rating, fifa_overall 81.0)	`match_squad_rank=5` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 1 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=1` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.0952` `goal_rate=0.0506` `scorer_weight=0.08`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.0672` `assist_rate=0.0255` `provider_weight=0.0557`	-

Reinildo Mandava D 87 0 0.01 0.00 61% 0 0.1 6.7 50.8 26 FBref Prior D_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost D_service raw output over 7 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 0.9979, quality 0.451)	`market_value_eur=4099999` `quality_factor=0.4511` `boost_factor=0.9979` `reference_quality=0.505` `history_matches=33` `history_shrink=0.1316`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0114)	`blend_weight=0.0114` `history_matches=33` `season_overlap_discount=1.0` `per90_to_match_factor=0.967` `prior_reliability=0.1333` `competition=Premier League`	fouls: 1.0369 -> 1.0526 interceptionWon: 1.1636 -> 1.1724 totalCross: 0.8673 -> 0.8739
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` `duel_context={'opponent_faced': 'Cole Palmer', 'matchup_zone': 'Left Wing vs Right Back — Flank Battle'}`	-
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 1 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=1`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.20 (squad xG sum 1.1254, xG scale 1.0672, xA scale 1.0672, team xA 0.82 = 0.68 x lambda)	`scoreline_reset=true` `team_lambda=1.201` `squad_xg_sum=1.1254` `squad_xa_sum=0.7648` `team_expected_assists=0.8162` `team_xa_to_xg_ratio=0.6796`	-
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 12 by projected rating, fifa_overall 80.0)	`match_squad_rank=12` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv_filled` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 1 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=1` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.0059` `goal_rate=0.0` `scorer_weight=0.0053`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.0317` `assist_rate=0.0111` `provider_weight=0.0258`	-

Daniel Ballard D 84 0 0.07 0.06 ok 1 0.6 6.8 52.3 24 FBref Prior D_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost D_service raw output over 7 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0073, quality 0.694)	`market_value_eur=27000000` `quality_factor=0.6941` `boost_factor=1.0073` `reference_quality=0.505` `history_matches=34` `history_shrink=0.1282`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0111)	`blend_weight=0.0111` `history_matches=34` `season_overlap_discount=1.0` `per90_to_match_factor=0.938` `prior_reliability=0.1333` `competition=Premier League`	interceptionWon: 1.073 -> 1.0663 wonTackle: 0.9431 -> 0.9378
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` `duel_context={'opponent_faced': 'João Pedro', 'matchup_zone': 'Striker vs Two Centre Backs — 1v2 Aerial/Physical Battle'}`	-
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 1 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=1`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.20 (squad xG sum 1.1254, xG scale 1.0672, xA scale 1.0672, team xA 0.82 = 0.68 x lambda)	`scoreline_reset=true` `team_lambda=1.201` `squad_xg_sum=1.1254` `squad_xa_sum=0.7648` `team_expected_assists=0.8162` `team_xa_to_xg_ratio=0.6796`	-
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 7 by projected rating, fifa_overall 78.0)	`match_squad_rank=7` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 1 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=1` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.0709` `goal_rate=0.0052` `scorer_weight=0.0588`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.0284` `assist_rate=0.0` `provider_weight=0.023`	-

Noah Sadiki M 84 0 0.06 0.05 ok 1 0.6 6.8 56.6 31 FBref Prior M_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost M_service raw output over 8 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0091, quality 0.791)	`market_value_eur=42000000` `quality_factor=0.7909` `boost_factor=1.0091` `reference_quality=0.505` `history_matches=42` `history_shrink=0.1064`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0092)	`blend_weight=0.0092` `history_matches=42` `season_overlap_discount=1.0` `per90_to_match_factor=0.9368` `prior_reliability=0.1333` `competition=Premier League`	wonTackle: 0.5861 -> 0.5936 wasFouled: 1.3029 -> 1.3082
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` duel_context={'opponent_faced': 'Moisés Caicedo', 'matchup_zone': 'CDM vs CDM', 'duel_win_rate': 0.366, 'pressing_faced': 'ULTRA_HIGH', 'numerical_situation': '1v1', 'xg_multiplier': 0.9205, 'skill_edge': -0.0341, 'height_edge': -0.02, 'attacker_skill': 0.8221, 'defender_skill': 0.8978, 'attacker_role': 'CDM', 'attacker_mv_price_index': 0.85, 'defender_adjusted_parts': [{'player': 'Moisés Caicedo', 'market_value': 104000000, 'role': 'CDM', 'price_index': 0.85, 'adjusted_quality': 0.8978}], 'xg_override': 0.0607}	aerialWon: 0.4365 -> 0.9 totalShots: 0.6433 -> 0.5922 shotOffTarget: 0.38 -> 0.3498 onTargetScoringAttempt: 0.2625 -> 0.2416 bigChanceMissed: 0.1271 -> 0.117 goals: 0.0684 -> 0.063 expectedGoals: 0.0659 -> 0.0607
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 1 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=1`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.20 (squad xG sum 1.1254, xG scale 1.0672, xA scale 1.0672, team xA 0.82 = 0.68 x lambda)	`scoreline_reset=true` `team_lambda=1.201` `squad_xg_sum=1.1254` `squad_xa_sum=0.7648` `team_expected_assists=0.8162` `team_xa_to_xg_ratio=0.6796`	-
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 6 by projected rating, fifa_overall 80.0)	`match_squad_rank=6` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 1 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=1` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.0627` `goal_rate=0.063` `scorer_weight=0.0661`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.0695` `assist_rate=0.0874` `provider_weight=0.0842`	-

Enzo Le Fée M 87 0 0.18 0.18 ok 1 1.2 6.9 59.5 33 FBref Prior M_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost M_service raw output over 8 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0059, quality 0.702)	`market_value_eur=28000000` `quality_factor=0.7022` `boost_factor=1.0059` `reference_quality=0.505` `history_matches=45` `history_shrink=0.1`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0082)	`blend_weight=0.0082` `history_matches=45` `season_overlap_discount=1.0` `per90_to_match_factor=0.9687` `prior_reliability=0.1267` `competition=Premier League`	totalCross: 2.7197 -> 2.7141
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` duel_context={'opponent_faced': 'Moisés Caicedo, Roméo Lavia', 'matchup_zone': 'Attacking Mid vs Double Pivot — Tight Space Battle', 'duel_win_rate': 0.18, 'pressing_faced': 'HIGH', 'numerical_situation': '1v2', 'xg_multiplier': 0.9071, 'skill_edge': -0.0372, 'height_edge': -0.026, 'attacker_skill': 0.7136, 'defender_skill': 0.7962, 'attacker_role': 'CAM', 'attacker_mv_price_index': 0.95, 'defender_adjusted_parts': [{'player': 'Moisés Caicedo', 'market_value': 104000000, 'role': 'CDM', 'price_index': 0.85, 'adjusted_quality': 0.8978}, {'player': 'Roméo Lavia', 'market_value': 23000000, 'role': 'CDM', 'price_index': 0.85, 'adjusted_quality': 0.6946}], 'xg_override': 0.1645}	aerialWon: 0.3769 -> 0.9 totalShots: 1.5349 -> 1.3923 shotOffTarget: 1.0651 -> 0.9662 onTargetScoringAttempt: 0.4657 -> 0.4224 expectedGoalsOnTarget: 0.2686 -> 0.2436 expectedGoals: 0.1814 -> 0.1645 goals: 0.1439 -> 0.1305 bigChanceMissed: 0.1221 -> 0.1108
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 1 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=1`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.20 (squad xG sum 1.1254, xG scale 1.0672, xA scale 1.0672, team xA 0.82 = 0.68 x lambda)	`scoreline_reset=true` `team_lambda=1.201` `squad_xg_sum=1.1254` `squad_xa_sum=0.7648` `team_expected_assists=0.8162` `team_xa_to_xg_ratio=0.6796`	expected_goals: 0.1645 -> 0.1756 expected_assists: 0.1506 -> 0.1607
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 4 by projected rating, fifa_overall 80.0)	`match_squad_rank=4` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 1 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=1` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.161` `goal_rate=0.1305` `scorer_weight=0.1478`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.1485` `assist_rate=0.1176` `provider_weight=0.1343`	-

Granit Xhaka M 90 0 0.04 0.02 69% 1 0.5 7.3 61.5 30 FBref Prior M_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost M_service raw output over 8 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0000, quality 0.505)	`market_value_eur=8500000` `quality_factor=0.5047` `boost_factor=1.0` `reference_quality=0.505` `history_matches=43` `history_shrink=0.1042`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0090)	`blend_weight=0.009` `history_matches=43` `season_overlap_discount=1.0` `per90_to_match_factor=0.9963` `prior_reliability=0.1333` `competition=Premier League`	wasFouled: 1.6575 -> 1.674 totalShots: 0.651 -> 0.6451
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` duel_context={'opponent_faced': 'Wesley Fofana', 'matchup_zone': 'Left Wing vs Right Back — Flank Battle', 'duel_win_rate': 0.337, 'pressing_faced': 'HIGH', 'numerical_situation': '1v1', 'xg_multiplier': 0.7898, 'skill_edge': -0.139, 'height_edge': -0.004, 'attacker_skill': 0.5047, 'defender_skill': 0.8135, 'attacker_role': 'LW', 'attacker_mv_price_index': 1.0, 'defender_adjusted_parts': [{'player': 'Wesley Fofana', 'market_value': 33000000, 'role': 'RB', 'price_index': 0.7, 'adjusted_quality': 0.8135}], 'xg_override': 0.0334}	aerialWon: 0.7615 -> 1.2 totalShots: 0.6492 -> 0.5127 shotOffTarget: 0.4519 -> 0.3569 onTargetScoringAttempt: 0.1973 -> 0.1558 goals: 0.0487 -> 0.0385 expectedGoalsOnTarget: 0.0433 -> 0.0342 expectedGoals: 0.0423 -> 0.0334
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 1 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=1`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.20 (squad xG sum 1.1254, xG scale 1.0672, xA scale 1.0672, team xA 0.82 = 0.68 x lambda)	`scoreline_reset=true` `team_lambda=1.201` `squad_xg_sum=1.1254` `squad_xa_sum=0.7648` `team_expected_assists=0.8162` `team_xa_to_xg_ratio=0.6796`	expected_assists: 0.16 -> 0.1709
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 1 by projected rating, fifa_overall 85.0)	`match_squad_rank=1` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 1 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=1` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.035` `goal_rate=0.0385` `scorer_weight=0.0378`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=1` `assist_probability=0.1571` `assist_rate=0.0916` `provider_weight=0.1367`	-

Chris Rigg M 70 0 0.10 0.02 341% 1 0.4 6.8 27.4 34 FBref Prior M_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost M_service raw output over 8 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0088, quality 0.669)	`market_value_eur=24000000` `quality_factor=0.6685` `boost_factor=1.0088` `reference_quality=0.505` `history_matches=23` `history_shrink=0.1786`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0015)	`blend_weight=0.0015` `history_matches=23` `season_overlap_discount=1.0` `per90_to_match_factor=0.783` `prior_reliability=0.0133` `competition=Premier League`	-
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` duel_context={'opponent_faced': 'Moisés Caicedo', 'matchup_zone': 'CDM vs CDM', 'duel_win_rate': 0.329, 'pressing_faced': 'ULTRA_HIGH', 'numerical_situation': '1v1', 'xg_multiplier': 0.8659, 'skill_edge': -0.0872, 'height_edge': -0.004, 'attacker_skill': 0.704, 'defender_skill': 0.8978, 'attacker_role': 'CDM', 'attacker_mv_price_index': 0.85, 'defender_adjusted_parts': [{'player': 'Moisés Caicedo', 'market_value': 104000000, 'role': 'CDM', 'price_index': 0.85, 'adjusted_quality': 0.8978}], 'xg_override': 0.094}	aerialWon: 0.5704 -> 1.0 totalShots: 1.2769 -> 1.1057 shotOffTarget: 0.837 -> 0.7248 onTargetScoringAttempt: 0.4399 -> 0.3809 expectedGoals: 0.1085 -> 0.094 goals: 0.0965 -> 0.0836 expectedGoalsOnTarget: 0.0883 -> 0.0765
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 1 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=1`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.20 (squad xG sum 1.1254, xG scale 1.0672, xA scale 1.0672, team xA 0.82 = 0.68 x lambda)	`scoreline_reset=true` `team_lambda=1.201` `squad_xg_sum=1.1254` `squad_xa_sum=0.7648` `team_expected_assists=0.8162` `team_xa_to_xg_ratio=0.6796`	expected_goals: 0.094 -> 0.1003
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 11 by projected rating, fifa_overall 74.0)	`match_squad_rank=11` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 1 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=1` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.0954` `goal_rate=0.0836` `scorer_weight=0.0931`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.067` `assist_rate=0.0712` `provider_weight=0.0731`	-

Dayann Methalie M 87 0 0.05 0.00 1266% 0 0.1 6.8 11.5 15 XGBoost M_service

Step-by-step (9 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost M_service raw output over 8 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0405, quality 0.694)	`market_value_eur=27000000` `quality_factor=0.6941` `boost_factor=1.0405` `reference_quality=0.505` `history_matches=2` `history_shrink=0.7143`	-
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=true`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` duel_context={'opponent_faced': 'Levi Colwill', 'matchup_zone': 'Right Wing vs Left Back — Flank Battle', 'duel_win_rate': 0.392, 'pressing_faced': 'ULTRA_HIGH', 'numerical_situation': '1v1', 'xg_multiplier': 0.8707, 'skill_edge': -0.084, 'height_edge': -0.004, 'attacker_skill': 0.6941, 'defender_skill': 0.8807, 'attacker_role': 'RW', 'attacker_mv_price_index': 1.0, 'defender_adjusted_parts': [{'player': 'Levi Colwill', 'market_value': 55000000, 'role': 'LB', 'price_index': 0.7, 'adjusted_quality': 0.8807}], 'xg_override': 0.0459}	aerialWon: 0.9561 -> 1.2 totalShots: 0.5176 -> 0.4507 shotOffTarget: 0.3663 -> 0.3189 onTargetScoringAttempt: 0.1513 -> 0.1317 expectedGoals: 0.0527 -> 0.0459 goals: 0.0459 -> 0.04
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 1 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=1`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.20 (squad xG sum 1.1254, xG scale 1.0672, xA scale 1.0672, team xA 0.82 = 0.68 x lambda)	`scoreline_reset=true` `team_lambda=1.201` `squad_xg_sum=1.1254` `squad_xa_sum=0.7648` `team_expected_assists=0.8162` `team_xa_to_xg_ratio=0.6796`	-
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 9 by projected rating, fifa_overall 75.0)	`match_squad_rank=9` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 1 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=1` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.0478` `goal_rate=0.04` `scorer_weight=0.046`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.0546` `assist_rate=0.071` `provider_weight=0.0662`	-

Brian Brobbey F 80 1 0.39 0.28 37% 2 1.4 6.8 55.5 30 FBref Prior F_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost F_service raw output over 7 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0115, quality 0.834)	`market_value_eur=53000000` `quality_factor=0.834` `boost_factor=1.0115` `reference_quality=0.505` `history_matches=38` `history_shrink=0.1163`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0071)	`blend_weight=0.0071` `history_matches=38` `season_overlap_discount=1.0` `per90_to_match_factor=0.8918` `prior_reliability=0.0933` `competition=Premier League`	wasFouled: 1.9355 -> 1.9263 totalOffside: 0.6403 -> 0.6493
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` duel_context={'opponent_faced': 'Jorrel Hato, Maxence Lacroix', 'matchup_zone': 'Striker vs Two Centre Backs — 1v2 Aerial/Physical Battle', 'duel_win_rate': 0.18, 'pressing_faced': 'ULTRA_HIGH', 'numerical_situation': '1v2', 'xg_multiplier': 0.9336, 'skill_edge': -0.0172, 'height_edge': -0.028, 'attacker_skill': 0.834, 'defender_skill': 0.8722, 'attacker_role': 'ST', 'attacker_mv_price_index': 1.0, 'defender_adjusted_parts': [{'player': 'Jorrel Hato', 'market_value': 47000000, 'role': 'CB', 'price_index': 0.75, 'adjusted_quality': 0.8582}, {'player': 'Maxence Lacroix', 'market_value': 64000000, 'role': 'CB', 'price_index': 0.75, 'adjusted_quality': 0.8862}], 'xg_override': 0.3615}	totalShots: 2.4402 -> 2.2782 shotOffTarget: 1.2774 -> 1.1926 onTargetScoringAttempt: 1.1628 -> 1.0856 expectedGoalsOnTarget: 0.4491 -> 0.4193 goals: 0.4331 -> 0.4043 bigChanceMissed: 0.4081 -> 0.381 expectedGoals: 0.3872 -> 0.3615
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 1 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=1`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.20 (squad xG sum 1.1254, xG scale 1.0672, xA scale 1.0672, team xA 0.82 = 0.68 x lambda)	`scoreline_reset=true` `team_lambda=1.201` `squad_xg_sum=1.1254` `squad_xa_sum=0.7648` `team_expected_assists=0.8162` `team_xa_to_xg_ratio=0.6796`	expected_goals: 0.3615 -> 0.3857
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 8 by projected rating, fifa_overall 78.0)	`match_squad_rank=8` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 1 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=1` `allocation_applied=true` `goals_allocated=1` `goal_probability=0.32` `goal_rate=0.4043` `scorer_weight=0.4258`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.0408` `assist_rate=0.0` `provider_weight=0.0334`	-

Team totals 1 1.20 12 6.9

Bench Candidates (7 modeled substitutes with empirical playing time)

Player	Pos	Exp Min	App Prob	Substitute Modeling Details
Melker Ellborg	G	4.4m	10.9%	Market-value substitute prior
Trai Hume	D	14.3m	23.3%	Market-value substitute prior
Omar Alderete	D	19.5m	60.9%	Market-value substitute prior
Habib Diarra	M	18.7m	71.8%	Market-value substitute prior
Jules Ahoka	M	22.2m	50.7%	Market-value substitute prior
Malick Fofana	F	26.4m	60.8%	Market-value substitute prior

Chelsea — projected line vs season baseline (4-2-3-1)

Player Pos Min Proj G Proj xG Season xG/g Proj Sh Season Sh/g Proj Rtg Season Rtg Rec Model / FBref Prior

Emiliano Martínez G 88 0 0.00 - n/a 0.0 7.0 58.8 28 FBref Prior G_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost G_service raw output over 3 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0004, quality 0.518)	`market_value_eur=9600000` `quality_factor=0.5177` `boost_factor=1.0004` `reference_quality=0.505` `history_matches=38` `history_shrink=0.1163`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0050)	`blend_weight=0.005` `history_matches=38` `season_overlap_discount=1.0` `per90_to_match_factor=0.9739` `prior_reliability=0.0667` `competition=Premier League`	wasFouled: 0.5351 -> 0.5422
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` `duel_context={'opponent_faced': 'Brian Brobbey, Enzo Le Fée', 'matchup_zone': 'Goalkeeper shot-stopping context', 'opponent_lambda': 1.3881}`	-
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 2 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=2`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.73 (squad xG sum 1.3618, xG scale 1.2726, xA scale 1.5089, team xA 1.13 = 0.65 x lambda)	`scoreline_reset=true` `team_lambda=1.733` `squad_xg_sum=1.3618` `squad_xa_sum=0.7465` `team_expected_assists=1.1265` `team_xa_to_xg_ratio=0.65`	shots_faced: 13.2 -> 11.4 expected_saves: 3.23 -> 2.79 high_pressure_saves: 1.11 -> 0.96
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 5 by projected rating, fifa_overall 85.0)	`match_squad_rank=5` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 2 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=2` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.0` `scorer_weight=0.0` `squad_weight_total=1.8217`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.0` `provider_weight=0.0` `squad_weight_total=1.1285`	-

Levi Colwill D 89 0 0.00 0.00 ok 0 0.1 6.8 43.8 17 FBref Prior D_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost D_service raw output over 7 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0314, quality 0.840)	`market_value_eur=55000000` `quality_factor=0.8399` `boost_factor=1.0314` `reference_quality=0.505` `history_matches=11` `history_shrink=0.3125`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0244)	`blend_weight=0.0244` `history_matches=11` `season_overlap_discount=1.0` `per90_to_match_factor=0.9882` `prior_reliability=0.12` `competition=Premier League`	wonTackle: 0.7511 -> 0.7729 fouls: 1.1945 -> 1.1788
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` `duel_context={'opponent_faced': 'Dayann Methalie', 'matchup_zone': 'Right Wing vs Left Back — Flank Battle'}`	-
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 2 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=2`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.73 (squad xG sum 1.3618, xG scale 1.2726, xA scale 1.5089, team xA 1.13 = 0.65 x lambda)	`scoreline_reset=true` `team_lambda=1.733` `squad_xg_sum=1.3618` `squad_xa_sum=0.7465` `team_expected_assists=1.1265` `team_xa_to_xg_ratio=0.65`	expected_assists: 0.0541 -> 0.0816
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 12 by projected rating, fifa_overall 80.0)	`match_squad_rank=12` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 2 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=2` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.001` `goal_rate=0.0` `scorer_weight=0.001`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.0784` `assist_rate=0.0` `provider_weight=0.0653`	-

Jorrel Hato D 74 0 0.00 0.01 76% 0 0.2 6.8 51.1 35 FBref Prior D_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost D_service raw output over 7 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0140, quality 0.813)	`market_value_eur=47000000` `quality_factor=0.8129` `boost_factor=1.014` `reference_quality=0.505` `history_matches=28` `history_shrink=0.1515`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0112)	`blend_weight=0.0112` `history_matches=28` `season_overlap_discount=1.0` `per90_to_match_factor=0.8236` `prior_reliability=0.1133` `competition=Premier League`	wasFouled: 0.1883 -> 0.197 wonTackle: 1.2483 -> 1.2398 fouls: 0.6436 -> 0.6364 interceptionWon: 1.0013 -> 0.9955
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` `duel_context={'opponent_faced': 'Brian Brobbey', 'matchup_zone': 'Striker vs Two Centre Backs — 1v2 Aerial/Physical Battle'}`	-
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 2 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=2`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.73 (squad xG sum 1.3618, xG scale 1.2726, xA scale 1.5089, team xA 1.13 = 0.65 x lambda)	`scoreline_reset=true` `team_lambda=1.733` `squad_xg_sum=1.3618` `squad_xa_sum=0.7465` `team_expected_assists=1.1265` `team_xa_to_xg_ratio=0.65`	expected_assists: 0.0336 -> 0.0507
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 14 by projected rating, fifa_overall 78.0)	`match_squad_rank=14` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 2 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=2` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.0023` `goal_rate=0.0` `scorer_weight=0.0022`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.0494` `assist_rate=0.186` `provider_weight=0.0634`	-

Wesley Fofana D 81 0 0.04 0.02 110% 1 0.3 7.0 45.7 30 FBref Prior D_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost D_service raw output over 7 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0100, quality 0.739)	`market_value_eur=33000000` `quality_factor=0.7389` `boost_factor=1.01` `reference_quality=0.505` `history_matches=30` `history_shrink=0.1429`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0043)	`blend_weight=0.0043` `history_matches=30` `season_overlap_discount=1.0` `per90_to_match_factor=0.8954` `prior_reliability=0.0467` `competition=Premier League`	wasFouled: 1.3271 -> 1.3213
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` `duel_context={'opponent_faced': 'Granit Xhaka', 'matchup_zone': 'Left Wing vs Right Back — Flank Battle'}`	-
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 2 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=2`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.73 (squad xG sum 1.3618, xG scale 1.2726, xA scale 1.5089, team xA 1.13 = 0.65 x lambda)	`scoreline_reset=true` `team_lambda=1.733` `squad_xg_sum=1.3618` `squad_xa_sum=0.7465` `team_expected_assists=1.1265` `team_xa_to_xg_ratio=0.65`	expected_goals: 0.0335 -> 0.0426 expected_assists: 0.0133 -> 0.0201
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 7 by projected rating, fifa_overall 79.0)	`match_squad_rank=7` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 2 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=2` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.0417` `goal_rate=0.0` `scorer_weight=0.0341`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.0199` `assist_rate=0.0` `provider_weight=0.0161`	-

Maxence Lacroix D 88 0 0.09 0.07 30% 1 0.6 6.8 55.1 31 FBref Prior D_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost D_service raw output over 7 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0111, quality 0.861)	`market_value_eur=64000000` `quality_factor=0.8608` `boost_factor=1.0111` `reference_quality=0.505` `history_matches=43` `history_shrink=0.1042`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0090)	`blend_weight=0.009` `history_matches=43` `season_overlap_discount=1.0` `per90_to_match_factor=0.9804` `prior_reliability=0.1333` `competition=Premier League`	wonTackle: 0.9113 -> 0.9208
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` `duel_context={'opponent_faced': 'Brian Brobbey', 'matchup_zone': 'Striker vs Two Centre Backs — 1v2 Aerial/Physical Battle'}`	-
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 2 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=2`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.73 (squad xG sum 1.3618, xG scale 1.2726, xA scale 1.5089, team xA 1.13 = 0.65 x lambda)	`scoreline_reset=true` `team_lambda=1.733` `squad_xg_sum=1.3618` `squad_xa_sum=0.7465` `team_expected_assists=1.1265` `team_xa_to_xg_ratio=0.65`	expected_goals: 0.0686 -> 0.0873 expected_assists: 0.0204 -> 0.0308
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 13 by projected rating, fifa_overall 82.0)	`match_squad_rank=13` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 2 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=2` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.0836` `goal_rate=0.0138` `scorer_weight=0.0698`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.0303` `assist_rate=0.0625` `provider_weight=0.0385`	-

Morgan Rogers M 88 0 0.35 0.22 56% 3 2.3 7.1 61.0 37 FBref Prior M_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost M_service raw output over 8 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0115, quality 0.897)	`market_value_eur=115000000` `quality_factor=0.8968` `boost_factor=1.0115` `reference_quality=0.505` `history_matches=46` `history_shrink=0.098`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0081)	`blend_weight=0.0081` `history_matches=46` `season_overlap_discount=1.0` `per90_to_match_factor=0.983` `prior_reliability=0.1267` `competition=Premier League`	totalShots: 2.2659 -> 2.2727
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` duel_context={'opponent_faced': 'Noah Sadiki, Chris Rigg', 'matchup_zone': 'Attacking Mid vs Double Pivot — Tight Space Battle', 'duel_win_rate': 0.349, 'pressing_faced': 'HIGH', 'numerical_situation': '1v2', 'xg_multiplier': 1.1596, 'skill_edge': 0.0606, 'height_edge': 0.048, 'attacker_skill': 0.8977, 'defender_skill': 0.763, 'attacker_role': 'CAM', 'attacker_mv_price_index': 0.95, 'defender_adjusted_parts': [{'player': 'Noah Sadiki', 'market_value': 42000000, 'role': 'CDM', 'price_index': 0.85, 'adjusted_quality': 0.8221}, {'player': 'Chris Rigg', 'market_value': 24000000, 'role': 'CDM', 'price_index': 0.85, 'adjusted_quality': 0.704}], 'xg_override': 0.2718}	totalShots: 2.2727 -> 2.6354 shotOffTarget: 1.4058 -> 1.6302 onTargetScoringAttempt: 0.8574 -> 0.9942 aerialWon: 1.0662 -> 1.2 goals: 0.2564 -> 0.2973 expectedGoals: 0.2344 -> 0.2718 expectedGoalsOnTarget: 0.2125 -> 0.2464 bigChanceMissed: 0.2096 -> 0.2431
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 2 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=2`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.73 (squad xG sum 1.3618, xG scale 1.2726, xA scale 1.5089, team xA 1.13 = 0.65 x lambda)	`scoreline_reset=true` `team_lambda=1.733` `squad_xg_sum=1.3618` `squad_xa_sum=0.7465` `team_expected_assists=1.1265` `team_xa_to_xg_ratio=0.65`	expected_goals: 0.2718 -> 0.3459 expected_assists: 0.1388 -> 0.2094
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 1 by projected rating, fifa_overall 84.0)	`match_squad_rank=1` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 2 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=2` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.2924` `goal_rate=0.2973` `scorer_weight=0.3733`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.1889` `assist_rate=0.1262` `provider_weight=0.1954`	-

Cole Palmer M 87 1 0.36 0.34 ok 3 2.5 7.0 61.7 35 FBref Prior M_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost M_service raw output over 8 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0146, quality 0.894)	`market_value_eur=101000000` `quality_factor=0.8936` `boost_factor=1.0146` `reference_quality=0.505` `history_matches=35` `history_shrink=0.125`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0103)	`blend_weight=0.0103` `history_matches=35` `season_overlap_discount=1.0` `per90_to_match_factor=0.9706` `prior_reliability=0.1267` `competition=Premier League`	totalShots: 2.4641 -> 2.4858 goals: 0.2482 -> 0.2562 onTargetScoringAttempt: 0.8618 -> 0.8686 fouls: 1.1314 -> 1.125
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` duel_context={'opponent_faced': 'Reinildo Mandava', 'matchup_zone': 'Left Wing vs Right Back — Flank Battle', 'duel_win_rate': 0.689, 'pressing_faced': 'LOW', 'numerical_situation': '1v1', 'xg_multiplier': 1.3078, 'skill_edge': 0.1894, 'height_edge': 0.02, 'attacker_skill': 0.8936, 'defender_skill': 0.4727, 'attacker_role': 'LW', 'attacker_mv_price_index': 1.0, 'defender_adjusted_parts': [{'player': 'Reinildo Mandava', 'market_value': 4099999, 'role': 'RB', 'price_index': 0.7, 'adjusted_quality': 0.4727}], 'xg_override': 0.2864}	aerialWon: 0.3319 -> 1.2 totalShots: 2.4858 -> 3.2509 shotOffTarget: 1.6023 -> 2.0955 onTargetScoringAttempt: 0.8686 -> 1.136 goals: 0.2562 -> 0.3351 expectedGoalsOnTarget: 0.2559 -> 0.3347 expectedGoals: 0.219 -> 0.2864 bigChanceMissed: 0.1256 -> 0.1643
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 2 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=2`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.73 (squad xG sum 1.3618, xG scale 1.2726, xA scale 1.5089, team xA 1.13 = 0.65 x lambda)	`scoreline_reset=true` `team_lambda=1.733` `squad_xg_sum=1.3618` `squad_xa_sum=0.7465` `team_expected_assists=1.1265` `team_xa_to_xg_ratio=0.65`	expected_goals: 0.2864 -> 0.3645 expected_assists: 0.1631 -> 0.2462
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 3 by projected rating, fifa_overall 85.0)	`match_squad_rank=3` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 2 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=2` `allocation_applied=true` `goals_allocated=1` `goal_probability=0.3055` `goal_rate=0.3351` `scorer_weight=0.4173`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=1` `assist_probability=0.2182` `assist_rate=0.1615` `provider_weight=0.2444`	-

Moisés Caicedo M 86 0 0.05 0.03 66% 1 0.6 7.0 58.0 28 FBref Prior M_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost M_service raw output over 8 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0136, quality 0.895)	`market_value_eur=104000000` `quality_factor=0.8945` `boost_factor=1.0136` `reference_quality=0.505` `history_matches=38` `history_shrink=0.1163`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0010)	`blend_weight=0.001` `history_matches=38` `season_overlap_discount=1.0` `per90_to_match_factor=0.9553` `prior_reliability=0.0133` `competition=Premier League`	-
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` duel_context={'opponent_faced': 'Noah Sadiki', 'matchup_zone': 'CDM vs CDM', 'duel_win_rate': 0.474, 'pressing_faced': 'HIGH', 'numerical_situation': '1v1', 'xg_multiplier': 1.0795, 'skill_edge': 0.0341, 'height_edge': 0.02, 'attacker_skill': 0.8978, 'defender_skill': 0.8221, 'attacker_role': 'CDM', 'attacker_mv_price_index': 0.85, 'defender_adjusted_parts': [{'player': 'Noah Sadiki', 'market_value': 42000000, 'role': 'CDM', 'price_index': 0.85, 'adjusted_quality': 0.8221}], 'xg_override': 0.0388}	aerialWon: 0.4661 -> 1.0 totalShots: 0.5108 -> 0.5514 shotOffTarget: 0.3937 -> 0.425 onTargetScoringAttempt: 0.1171 -> 0.1264
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 2 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=2`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.73 (squad xG sum 1.3618, xG scale 1.2726, xA scale 1.5089, team xA 1.13 = 0.65 x lambda)	`scoreline_reset=true` `team_lambda=1.733` `squad_xg_sum=1.3618` `squad_xa_sum=0.7465` `team_expected_assists=1.1265` `team_xa_to_xg_ratio=0.65`	expected_goals: 0.0388 -> 0.0494 expected_assists: 0.0359 -> 0.0542
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 6 by projected rating, fifa_overall 86.0)	`match_squad_rank=6` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 2 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=2` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.0482` `goal_rate=0.0481` `scorer_weight=0.0546`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.0528` `assist_rate=0.051` `provider_weight=0.0637`	-

Jordan Henderson M 83 0 0.08 0.02 339% 1 0.2 7.0 52.2 25 XGBoost M_service

Step-by-step (9 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost M_service raw output over 8 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 0.9975, quality 0.437)	`market_value_eur=1300000` `quality_factor=0.4374` `boost_factor=0.9975` `reference_quality=0.505` `history_matches=36` `history_shrink=0.122`	-
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=true`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` duel_context={'opponent_faced': 'Nordi Mukiele', 'matchup_zone': 'Right Wing vs Left Back — Flank Battle', 'duel_win_rate': 0.366, 'pressing_faced': 'LOW', 'numerical_situation': '1v1', 'xg_multiplier': 0.8329, 'skill_edge': -0.1057, 'height_edge': -0.008, 'attacker_skill': 0.4374, 'defender_skill': 0.6723, 'attacker_role': 'RW', 'attacker_mv_price_index': 1.0, 'defender_adjusted_parts': [{'player': 'Nordi Mukiele', 'market_value': 17100000, 'role': 'LB', 'price_index': 0.7, 'adjusted_quality': 0.6723}], 'xg_override': 0.0667}	aerialWon: 0.6958 -> 1.1 totalShots: 0.9953 -> 0.829 shotOffTarget: 0.6619 -> 0.5513 onTargetScoringAttempt: 0.3334 -> 0.2777 goals: 0.103 -> 0.0858 expectedGoals: 0.0801 -> 0.0667 expectedGoalsOnTarget: 0.0772 -> 0.0643
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 2 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=2`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.73 (squad xG sum 1.3618, xG scale 1.2726, xA scale 1.5089, team xA 1.13 = 0.65 x lambda)	`scoreline_reset=true` `team_lambda=1.733` `squad_xg_sum=1.3618` `squad_xa_sum=0.7465` `team_expected_assists=1.1265` `team_xa_to_xg_ratio=0.65`	expected_goals: 0.0667 -> 0.0849 expected_assists: 0.1205 -> 0.1818
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 4 by projected rating, fifa_overall 78.0)	`match_squad_rank=4` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 2 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=2` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.0814` `goal_rate=0.0858` `scorer_weight=0.0988`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.1662` `assist_rate=0.1012` `provider_weight=0.1612`	-

Roméo Lavia M 72 0 0.06 0.04 50% 1 0.2 6.8 40.2 24 FBref Prior M_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost M_service raw output over 8 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0097, quality 0.660)	`market_value_eur=23000000` `quality_factor=0.6595` `boost_factor=1.0097` `reference_quality=0.505` `history_matches=19` `history_shrink=0.2083`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0135)	`blend_weight=0.0135` `history_matches=19` `season_overlap_discount=1.0` `per90_to_match_factor=0.7976` `prior_reliability=0.1` `competition=Premier League`	onTargetScoringAttempt: 0.14 -> 0.1529 interceptionWon: 1.7743 -> 1.7647 totalShots: 0.5631 -> 0.5703 goals: 0.058 -> 0.0644 fouls: 0.9228 -> 0.9175
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` duel_context={'opponent_faced': 'Noah Sadiki', 'matchup_zone': 'CDM vs CDM', 'duel_win_rate': 0.395, 'pressing_faced': 'HIGH', 'numerical_situation': '1v1', 'xg_multiplier': 0.9627, 'skill_edge': -0.0573, 'height_edge': 0.032, 'attacker_skill': 0.6946, 'defender_skill': 0.8221, 'attacker_role': 'CDM', 'attacker_mv_price_index': 0.85, 'defender_adjusted_parts': [{'player': 'Noah Sadiki', 'market_value': 42000000, 'role': 'CDM', 'price_index': 0.85, 'adjusted_quality': 0.8221}], 'xg_override': 0.047}	aerialWon: 0.2299 -> 1.1 totalShots: 0.576 -> 0.5545 shotOffTarget: 0.4231 -> 0.4073 onTargetScoringAttempt: 0.1529 -> 0.1472 expectedGoalsOnTarget: 0.1518 -> 0.1461
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 2 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=2`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.73 (squad xG sum 1.3618, xG scale 1.2726, xA scale 1.5089, team xA 1.13 = 0.65 x lambda)	`scoreline_reset=true` `team_lambda=1.733` `squad_xg_sum=1.3618` `squad_xa_sum=0.7465` `team_expected_assists=1.1265` `team_xa_to_xg_ratio=0.65`	expected_goals: 0.047 -> 0.0598 expected_assists: 0.0388 -> 0.0585
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 11 by projected rating, fifa_overall 78.0)	`match_squad_rank=11` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 2 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=2` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.058` `goal_rate=0.062` `scorer_weight=0.069`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.0568` `assist_rate=0.0555` `provider_weight=0.0695`	-

João Pedro F 85 1 0.56 0.43 31% 3 2.1 7.1 60.0 27 FBref Prior F_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost F_service raw output over 7 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0126, quality 0.890)	`market_value_eur=92000000` `quality_factor=0.89` `boost_factor=1.0126` `reference_quality=0.505` `history_matches=41` `history_shrink=0.1087`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0094)	`blend_weight=0.0094` `history_matches=41` `season_overlap_discount=1.0` `per90_to_match_factor=0.9492` `prior_reliability=0.1333` `competition=Premier League`	totalShots: 2.6388 -> 2.6542 goalAssist: 0.2805 -> 0.2868 goals: 0.3926 -> 0.3978 totalCross: 0.5376 -> 0.5325
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` duel_context={'opponent_faced': 'Kevin Danso, Daniel Ballard', 'matchup_zone': 'Striker vs Two Centre Backs — 1v2 Aerial/Physical Battle', 'duel_win_rate': 0.266, 'pressing_faced': 'MEDIUM', 'numerical_situation': '1v2', 'xg_multiplier': 1.0968, 'skill_edge': 0.0759, 'height_edge': -0.01, 'attacker_skill': 0.89, 'defender_skill': 0.7214, 'attacker_role': 'ST', 'attacker_mv_price_index': 1.0, 'defender_adjusted_parts': [{'player': 'Kevin Danso', 'market_value': 19400000, 'role': 'CB', 'price_index': 0.75, 'adjusted_quality': 0.6847}, {'player': 'Daniel Ballard', 'market_value': 27000000, 'role': 'CB', 'price_index': 0.75, 'adjusted_quality': 0.7581}], 'xg_override': 0.4394}	totalShots: 2.6542 -> 2.9111 shotOffTarget: 1.5459 -> 1.6955 onTargetScoringAttempt: 1.0959 -> 1.202 expectedGoalsOnTarget: 0.5532 -> 0.6067 bigChanceMissed: 0.4122 -> 0.4521 expectedGoals: 0.4006 -> 0.4394 goals: 0.3978 -> 0.4363
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 2 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=2`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.73 (squad xG sum 1.3618, xG scale 1.2726, xA scale 1.5089, team xA 1.13 = 0.65 x lambda)	`scoreline_reset=true` `team_lambda=1.733` `squad_xg_sum=1.3618` `squad_xa_sum=0.7465` `team_expected_assists=1.1265` `team_xa_to_xg_ratio=0.65`	expected_goals: 0.4394 -> 0.5592 expected_assists: 0.0612 -> 0.0923
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 2 by projected rating, fifa_overall 83.0)	`match_squad_rank=2` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 2 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=2` `allocation_applied=true` `goals_allocated=1` `goal_probability=0.4283` `goal_rate=0.4363` `scorer_weight=0.5557`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.0882` `assist_rate=0.2868` `provider_weight=0.1154`	-

Team totals 2 1.73 15 7.0

Bench Candidates (7 modeled substitutes with empirical playing time)

Player	Pos	Exp Min	App Prob	Substitute Modeling Details
Mike Penders	G	5.3m	15.0%	Market-value substitute prior
Reece James	D	19.5m	44.4%	Market-value substitute prior
Malo Gusto	D	14.8m	24.3%	Market-value substitute prior
Marco Palestra	M	23.5m	56.6%	Market-value substitute prior
Valentín Barco	M	15.5m	72.1%	Market-value substitute prior
Estêvão	F	27.1m	63.9%	Market-value substitute prior

Antagonist / tactical summary

Midfield is evenly contested — Noah Sadiki vs Morgan Rogers will be the pivotal duel. Total antagonist xG — Home: 0.76, Away: 1.15. Home danger zone: Striker vs Two Centre Backs — 1v2 Aerial/Physical Battle. Away danger zone: Striker vs Two Centre Backs — 1v2 Aerial/Physical Battle.

Midfield — possession 49.3% / 50.7%, dominance ratio 0.47, key duel: Noah Sadiki vs Morgan Rogers

Granit Xhaka xG 0.03Dayann Methalie xG 0.05Enzo Le Fée xG 0.16Cole Palmer xG 0.29Jordan Henderson xG 0.07Morgan Rogers xG 0.27

Summary

Antagonist matchup analysis for Sunderland vs Chelsea. Most likely scoreline: 1-2. Lambdas derived from 6 home and 6 away individual player duels. Head-to-head: Sunderland 5W-1D-4L vs Chelsea over 10 meetings.

Tactical parity between 4-2-3-1 and 4-2-3-1. Match outcome driven primarily by individual player quality. Midfield is evenly contested — Noah Sadiki vs Morgan Rogers will be the pivotal duel. Total antagonist xG — Home: 0.76, Away: 1.15. Home danger zone: Striker vs Two Centre Backs — 1v2 Aerial/Physical Battle. Away danger zone: Striker vs Two Centre Backs — 1v2 Aerial/Physical Battle.

Villarreal vs Getafe

LaLiga · 20261108 · 4-4-2 vs 5-4-1

Predicted 2-0Winner: VillarrealWinning H 2-0 (11.6%)Winning A 0-1 (4.2%)Confidence 0.73λ home/away 2.14 / 0.84Most-likely P 11.6%

Actual vs predicted: result not available yet (upcoming fixture).

XGBoost Reasoning & FBref Dataset Tracing

Antagonist + XGBoost match analysis for Villarreal vs Getafe. Player statistics regressed using position-specific XGBoost models (G, D, M, F) with 2026-2027 FBref priors. Micro-duels resolved across 4-4-2 vs 5-4-1 (duel lambdas: Home 1.604, Away 0.740). Final calibrated Poisson lambdas: Home 2.143, Away 0.840 (most likely score 2-0).

Home Starters 11/11 XGBoost (8 FBref) Away Starters 11/11 XGBoost (11 FBref) Bench Modeled H: 7 / A: 7 Duel λ (H/A) 1.60 / 0.74 Final λ (H/A) 2.14 / 0.84

FBref Team Priors: Villarreal (Gls/90: 1.3333333333333333, GA/90: 1.33 → E[λ]: 0.326) vs Getafe (Gls/90: 0.3333333333333333, GA/90: 0.33 → E[λ]: 0.328) | Blend Weight: 0.228

Modeling Pipeline Stages (5 layers)

1. XGBoost player regressions for volume & event rates with market-value quality scaling
2. Multi-source prior blending (FBref 2026-2027 tables + Sofascore season history)
3. Antagonist formation duel matrix resolving 1v1 and zonal matchups
4. Team-level prior calibration (FBref possession and attack/defense efficiencies)
5. Bivariate Poisson matrix with Dixon-Coles low-score correction

Standing & form context

Metric	Villarreal	Getafe
League position	8.00	18.00
Points per game	1.19	1.08
Attack index	1.00	0.70
Defense (concede) index	1.00	1.00
Form multiplier	1.01	0.99
Standing λ multiplier	1.00	0.87

Lambda decomposition (how the goal rates were built)

Stage	Villarreal λ	Getafe λ
FBref Squad Prior E[λ] (weight 0.228)	0.326	0.328
Pre-mismatch	1.815	0.698
Pre-calibration (Micro-duels)	1.604	0.740
Final λ	2.143	0.840

competition calibration scale 1.187competition observed total goals 2.984quality gap applied Truemismatch score pressure applied Truecompetitive underdog floor applied Falsestage discount 1.0

Scoreline distribution & markets

Top 5 scorelines	P
2-0	11.6%
1-0	10.8%
2-1	9.8%
1-1	9.2%
3-0	8.3%

Home win	67.3%
Draw	19.3%
Away win	13.5%
Over 2.5	57.3%
BTTS yes	50.2%

Pregame fair line — favorite Villarreal (67.3%), fair total 2.98, fair spread 1.30

Projected players vs season baseline

Proj = projected for this match · Season = this season's per-game average from match data · deviation badge compares projected xG to the player's season xG/game.

Villarreal — projected line vs season baseline (4-4-2)

Player Pos Min Proj G Proj xG Season xG/g Proj Sh Season Sh/g Proj Rtg Season Rtg Rec Model / FBref Prior

Luiz Júnior G 90 0 0.00 - n/a 0.0 7.0 36.5 35 XGBoost G_service

Step-by-step (9 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost G_service raw output over 3 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0018, quality 0.550)	`market_value_eur=12400000` `quality_factor=0.5502` `boost_factor=1.0018` `reference_quality=0.505` `history_matches=32` `history_shrink=0.1351`	-
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` `duel_context={'opponent_faced': 'Martín Satriano, Ramón Terrats', 'matchup_zone': 'Goalkeeper shot-stopping context', 'opponent_lambda': 0.7396}`	-
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 2 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=2`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 2.14 (squad xG sum 1.0517, xG scale 2.0376, xA scale 1.7540, team xA 1.71 = 0.80 x lambda)	`scoreline_reset=true` `team_lambda=2.143` `squad_xg_sum=1.0517` `squad_xa_sum=0.9774` `team_expected_assists=1.7144` `team_xa_to_xg_ratio=0.8`	shots_faced: 7.0 -> 8.0 expected_saves: 1.71 -> 1.96 high_pressure_saves: 0.59 -> 0.67
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 5 by projected rating, fifa_overall 78.0)	`match_squad_rank=5` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 2 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=2` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.0` `scorer_weight=0.0` `squad_weight_total=2.3628`	-
`integer_assist_projection`	Team scoreline allocation: 2 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=2` `assists_allocated=0` `assist_probability=0.0` `provider_weight=0.0` `squad_weight_total=1.701`	-

Renato Veiga D 90 0 0.03 0.01 200% 1 0.2 7.1 58.2 33 FBref Prior D_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost D_service raw output over 7 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0065, quality 0.710)	`market_value_eur=29000000` `quality_factor=0.71` `boost_factor=1.0065` `reference_quality=0.505` `history_matches=42` `history_shrink=0.1064`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0138)	`blend_weight=0.0138` `history_matches=42` `season_overlap_discount=1.0` `per90_to_match_factor=0.9988` `prior_reliability=0.2` `competition=La Liga`	wonTackle: 0.6117 -> 0.6032
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` `duel_context={'opponent_faced': 'Martín Satriano', 'matchup_zone': 'Striker vs Two Centre Backs — 1v2 Aerial/Physical Battle'}`	-
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 2 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=2`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 2.14 (squad xG sum 1.0517, xG scale 2.0376, xA scale 1.7540, team xA 1.71 = 0.80 x lambda)	`scoreline_reset=true` `team_lambda=2.143` `squad_xg_sum=1.0517` `squad_xa_sum=0.9774` `team_expected_assists=1.7144` `team_xa_to_xg_ratio=0.8`	expected_goals: 0.0135 -> 0.0275 expected_assists: 0.0994 -> 0.1743
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 3 by projected rating, fifa_overall 78.0)	`match_squad_rank=3` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 2 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=2` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.0271` `goal_rate=0.0` `scorer_weight=0.022`	-
`integer_assist_projection`	Team scoreline allocation: 2 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=2` `assists_allocated=0` `assist_probability=0.16` `assist_rate=0.0662` `provider_weight=0.1394`	-

Juan Foyth D 78 0 0.07 0.02 176% 1 0.3 7.0 53.9 13 FBref Prior D_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost D_service raw output over 7 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0001, quality 0.506)	`market_value_eur=8600000` `quality_factor=0.5059` `boost_factor=1.0001` `reference_quality=0.505` `history_matches=19` `history_shrink=0.2083`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0271)	`blend_weight=0.0271` `history_matches=19` `season_overlap_discount=1.0` `per90_to_match_factor=0.8703` `prior_reliability=0.2` `competition=La Liga`	interceptionWon: 1.1421 -> 1.1662 totalShots: 0.5069 -> 0.5245 wonTackle: 0.9478 -> 0.9614 fouls: 0.6301 -> 0.6209 onTargetScoringAttempt: 0.0973 -> 0.1024 wasFouled: 0.3964 -> 0.4014
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` `duel_context={'opponent_faced': 'Francho Serrano', 'matchup_zone': 'Right Mid vs Left Back — Wide Channel'}`	-
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 2 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=2`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 2.14 (squad xG sum 1.0517, xG scale 2.0376, xA scale 1.7540, team xA 1.71 = 0.80 x lambda)	`scoreline_reset=true` `team_lambda=2.143` `squad_xg_sum=1.0517` `squad_xa_sum=0.9774` `team_expected_assists=1.7144` `team_xa_to_xg_ratio=0.8`	expected_goals: 0.0334 -> 0.0681 expected_assists: 0.0391 -> 0.0686
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 7 by projected rating, fifa_overall 79.0)	`match_squad_rank=7` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 2 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=2` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.0658` `goal_rate=0.0` `scorer_weight=0.0545`	-
`integer_assist_projection`	Team scoreline allocation: 2 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=2` `assists_allocated=0` `assist_probability=0.0663` `assist_rate=0.0` `provider_weight=0.0549`	-

Pau Navarro D 85 0 0.01 0.00 319% 0 0.0 6.8 44.3 35 FBref Prior D_service

Step-by-step (9 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost D_service raw output over 7 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0031, quality 0.575)	`market_value_eur=14600000` `quality_factor=0.5748` `boost_factor=1.0031` `reference_quality=0.505` `history_matches=29` `history_shrink=0.1471`	-
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` `duel_context={'opponent_faced': 'Martín Satriano', 'matchup_zone': 'Striker vs Two Centre Backs — 1v2 Aerial/Physical Battle'}`	-
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 2 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=2`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 2.14 (squad xG sum 1.0517, xG scale 2.0376, xA scale 1.7540, team xA 1.71 = 0.80 x lambda)	`scoreline_reset=true` `team_lambda=2.143` `squad_xg_sum=1.0517` `squad_xa_sum=0.9774` `team_expected_assists=1.7144` `team_xa_to_xg_ratio=0.8`	expected_assists: 0.0079 -> 0.0139
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 13 by projected rating, fifa_overall 74.0)	`match_squad_rank=13` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 2 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=2` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.0061` `goal_rate=0.0` `scorer_weight=0.0058`	-
`integer_assist_projection`	Team scoreline allocation: 2 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=2` `assists_allocated=0` `assist_probability=0.0138` `assist_rate=0.0` `provider_weight=0.012`	-

Sergi Cardona D 82 0 0.07 0.01 406% 1 0.4 6.8 48.3 35 FBref Prior D_service

Step-by-step (9 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost D_service raw output over 7 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0000, quality 0.506)	`market_value_eur=8600000` `quality_factor=0.5059` `boost_factor=1.0` `reference_quality=0.505` `history_matches=38` `history_shrink=0.1163`	-
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` `duel_context={'opponent_faced': 'Ramón Terrats', 'matchup_zone': 'Left Mid vs Right Back — Wide Channel'}`	-
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 2 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=2`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 2.14 (squad xG sum 1.0517, xG scale 2.0376, xA scale 1.7540, team xA 1.71 = 0.80 x lambda)	`scoreline_reset=true` `team_lambda=2.143` `squad_xg_sum=1.0517` `squad_xa_sum=0.9774` `team_expected_assists=1.7144` `team_xa_to_xg_ratio=0.8`	expected_goals: 0.0367 -> 0.0748 expected_assists: 0.1071 -> 0.1879
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 12 by projected rating, fifa_overall 79.0)	`match_squad_rank=12` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 2 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=2` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.0721` `goal_rate=0.0405` `scorer_weight=0.0781`	-
`integer_assist_projection`	Team scoreline allocation: 2 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=2` `assists_allocated=0` `assist_probability=0.1713` `assist_rate=0.0081` `provider_weight=0.1503`	-

Tajon Buchanan M 74 1 0.38 0.09 330% 2 1.1 6.9 58.4 34 FBref Prior M_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost M_service raw output over 8 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0016, quality 0.557)	`market_value_eur=13000000` `quality_factor=0.557` `boost_factor=1.0016` `reference_quality=0.505` `history_matches=43` `history_shrink=0.1042`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0063)	`blend_weight=0.0063` `history_matches=43` `season_overlap_discount=1.0` `per90_to_match_factor=0.8183` `prior_reliability=0.0933` `competition=La Liga`	totalCross: 2.7078 -> 2.7239 wasFouled: 1.6067 -> 1.6187 fouls: 0.9082 -> 0.9135
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` `duel_context={'opponent_faced': 'Djené', 'matchup_zone': 'LM vs CB', 'duel_win_rate': 0.501, 'pressing_faced': 'LOW', 'numerical_situation': '1v1', 'xg_multiplier': 1.1191, 'skill_edge': 0.061, 'height_edge': 0.02, 'attacker_skill': 0.5731, 'defender_skill': 0.4374, 'attacker_role': 'LM', 'attacker_mv_price_index': 0.9, 'defender_adjusted_parts': [{'player': 'Djené', 'market_value': 1500000, 'role': 'CB', 'price_index': 0.75, 'adjusted_quality': 0.4374}], 'xg_override': 0.1852}`	aerialWon: 0.5398 -> 1.1 totalShots: 2.0298 -> 2.2715 shotOffTarget: 1.3395 -> 1.499 onTargetScoringAttempt: 0.6903 -> 0.7725 goals: 0.1989 -> 0.2226 expectedGoals: 0.1655 -> 0.1852 bigChanceMissed: 0.1241 -> 0.1389 expectedGoalsOnTarget: 0.0834 -> 0.0933
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 2 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=2`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 2.14 (squad xG sum 1.0517, xG scale 2.0376, xA scale 1.7540, team xA 1.71 = 0.80 x lambda)	`scoreline_reset=true` `team_lambda=2.143` `squad_xg_sum=1.0517` `squad_xa_sum=0.9774` `team_expected_assists=1.7144` `team_xa_to_xg_ratio=0.8`	expected_goals: 0.1852 -> 0.3774 expected_assists: 0.1523 -> 0.2671
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 8 by projected rating, fifa_overall 76.0)	`match_squad_rank=8` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 2 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=2` `allocation_applied=true` `goals_allocated=1` `goal_probability=0.3144` `goal_rate=0.2226` `scorer_weight=0.4374`	-
`integer_assist_projection`	Team scoreline allocation: 2 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=2` `assists_allocated=0` `assist_probability=0.2344` `assist_rate=0.1231` `provider_weight=0.2286`	-

Santi Comesaña M 73 0 0.11 0.07 59% 1 0.4 6.9 63.2 31 FBref Prior M_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost M_service raw output over 8 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0000, quality 0.503)	`market_value_eur=8400000` `quality_factor=0.5035` `boost_factor=1.0` `reference_quality=0.505` `history_matches=44` `history_shrink=0.102`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0097)	`blend_weight=0.0097` `history_matches=44` `season_overlap_discount=1.0` `per90_to_match_factor=0.8117` `prior_reliability=0.1467` `competition=La Liga`	wasFouled: 1.2021 -> 1.194 wonTackle: 0.6616 -> 0.6695 totalShots: 0.6804 -> 0.6738 totalCross: 0.5438 -> 0.5385
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` `duel_context={'opponent_faced': 'Djené', 'matchup_zone': 'RM vs CB', 'duel_win_rate': 0.491, 'pressing_faced': 'LOW', 'numerical_situation': '1v1', 'xg_multiplier': 1.104, 'skill_edge': 0.0347, 'height_edge': 0.036, 'attacker_skill': 0.5146, 'defender_skill': 0.4374, 'attacker_role': 'RM', 'attacker_mv_price_index': 0.9, 'defender_adjusted_parts': [{'player': 'Djené', 'market_value': 1500000, 'role': 'CB', 'price_index': 0.75, 'adjusted_quality': 0.4374}], 'xg_override': 0.0562}`	aerialWon: 0.4888 -> 1.2 totalShots: 0.6787 -> 0.7493 shotOffTarget: 0.51 -> 0.563 onTargetScoringAttempt: 0.1687 -> 0.1862 bigChanceMissed: 0.1141 -> 0.126 expectedGoalsOnTarget: 0.0741 -> 0.0818 goals: 0.052 -> 0.0574 expectedGoals: 0.0509 -> 0.0562
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 2 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=2`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 2.14 (squad xG sum 1.0517, xG scale 2.0376, xA scale 1.7540, team xA 1.71 = 0.80 x lambda)	`scoreline_reset=true` `team_lambda=2.143` `squad_xg_sum=1.0517` `squad_xa_sum=0.9774` `team_expected_assists=1.7144` `team_xa_to_xg_ratio=0.8`	expected_goals: 0.0562 -> 0.1145 expected_assists: 0.0782 -> 0.1372
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 9 by projected rating, fifa_overall 79.0)	`match_squad_rank=9` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 2 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=2` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.1082` `goal_rate=0.0574` `scorer_weight=0.1158`	-
`integer_assist_projection`	Team scoreline allocation: 2 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=2` `assists_allocated=0` `assist_probability=0.1282` `assist_rate=0.0654` `provider_weight=0.1235`	-

Pape Gueye M 79 0 0.40 0.15 171% 3 1.4 7.1 62.2 31 FBref Prior M_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost M_service raw output over 8 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0090, quality 0.775)	`market_value_eur=39000000` `quality_factor=0.7754` `boost_factor=1.009` `reference_quality=0.505` `history_matches=40` `history_shrink=0.1111`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0144)	`blend_weight=0.0144` `history_matches=40` `season_overlap_discount=1.0` `per90_to_match_factor=0.8823` `prior_reliability=0.2` `competition=La Liga`	totalShots: 1.7229 -> 1.7108
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` duel_context={'opponent_faced': 'Mario Martín', 'matchup_zone': 'Central Mid vs Central Mid — Box-to-Box Midfield Contest', 'duel_win_rate': 0.574, 'pressing_faced': 'LOW', 'numerical_situation': '1v1', 'xg_multiplier': 1.226, 'skill_edge': 0.1138, 'height_edge': 0.04, 'attacker_skill': 0.7972, 'defender_skill': 0.5444, 'attacker_role': 'CM', 'attacker_mv_price_index': 0.9, 'defender_adjusted_parts': [{'player': 'Mario Martín', 'market_value': 10700000, 'role': 'CM', 'price_index': 0.9, 'adjusted_quality': 0.5444}], 'xg_override': 0.197}	aerialWon: 0.3346 -> 1.2 totalShots: 1.7201 -> 2.1089 shotOffTarget: 1.2406 -> 1.521 onTargetScoringAttempt: 0.4795 -> 0.5879 expectedGoals: 0.1607 -> 0.197 goals: 0.1556 -> 0.1908 expectedGoalsOnTarget: 0.0716 -> 0.0878 bigChanceMissed: 0.0471 -> 0.0577
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 2 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=2`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 2.14 (squad xG sum 1.0517, xG scale 2.0376, xA scale 1.7540, team xA 1.71 = 0.80 x lambda)	`scoreline_reset=true` `team_lambda=2.143` `squad_xg_sum=1.0517` `squad_xa_sum=0.9774` `team_expected_assists=1.7144` `team_xa_to_xg_ratio=0.8`	expected_goals: 0.197 -> 0.4014 expected_assists: 0.1078 -> 0.1891
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 2 by projected rating, fifa_overall 80.0)	`match_squad_rank=2` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 2 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=2` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.3306` `goal_rate=0.1908` `scorer_weight=0.3913`	-
`integer_assist_projection`	Team scoreline allocation: 2 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=2` `assists_allocated=0` `assist_probability=0.1723` `assist_rate=0.0916` `provider_weight=0.1697`	-

Carlos Maciá M 84 0 0.10 0.01 1250% 1 0.1 6.9 32.2 19 FBref Prior M_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost M_service raw output over 8 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0061, quality 0.574)	`market_value_eur=14500000` `quality_factor=0.5737` `boost_factor=1.0061` `reference_quality=0.505` `history_matches=12` `history_shrink=0.2941`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0064)	`blend_weight=0.0064` `history_matches=12` `season_overlap_discount=1.0` `per90_to_match_factor=0.9373` `prior_reliability=0.0333` `competition=La Liga`	wonTackle: 2.2948 -> 2.316
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` duel_context={'opponent_faced': 'Mario Martín', 'matchup_zone': 'Central Mid vs Central Mid — Box-to-Box Midfield Contest', 'duel_win_rate': 0.501, 'pressing_faced': 'LOW', 'numerical_situation': '1v1', 'xg_multiplier': 1.1191, 'skill_edge': 0.021, 'height_edge': 0.06, 'attacker_skill': 0.5912, 'defender_skill': 0.5444, 'attacker_role': 'CM', 'attacker_mv_price_index': 0.9, 'defender_adjusted_parts': [{'player': 'Mario Martín', 'market_value': 10700000, 'role': 'CM', 'price_index': 0.9, 'adjusted_quality': 0.5444}], 'xg_override': 0.0496}	aerialWon: 1.056 -> 1.4 totalShots: 0.5626 -> 0.6296 shotOffTarget: 0.4307 -> 0.482 onTargetScoringAttempt: 0.1319 -> 0.1476 goals: 0.0555 -> 0.0621 expectedGoals: 0.0443 -> 0.0496
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 2 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=2`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 2.14 (squad xG sum 1.0517, xG scale 2.0376, xA scale 1.7540, team xA 1.71 = 0.80 x lambda)	`scoreline_reset=true` `team_lambda=2.143` `squad_xg_sum=1.0517` `squad_xa_sum=0.9774` `team_expected_assists=1.7144` `team_xa_to_xg_ratio=0.8`	expected_goals: 0.0496 -> 0.1011 expected_assists: 0.034 -> 0.0596
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 10 by projected rating, fifa_overall 65.0)	`match_squad_rank=10` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 2 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=2` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.0962` `goal_rate=0.0621` `scorer_weight=0.1138`	-
`integer_assist_projection`	Team scoreline allocation: 2 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=2` `assists_allocated=0` `assist_probability=0.0579` `assist_rate=0.0499` `provider_weight=0.0709`	-

Ilias Akhomach F 72 0 0.23 0.07 238% 2 1.1 7.0 54.6 29 FBref Prior F_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost F_service raw output over 7 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0013, quality 0.535)	`market_value_eur=11100000` `quality_factor=0.5353` `boost_factor=1.0013` `reference_quality=0.505` `history_matches=29` `history_shrink=0.1471`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0032)	`blend_weight=0.0032` `history_matches=29` `season_overlap_discount=1.0` `per90_to_match_factor=0.7984` `prior_reliability=0.0333` `competition=La Liga`	totalCross: 3.4019 -> 3.4317 totalShots: 1.77 -> 1.7644 wasFouled: 1.5983 -> 1.6034
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` duel_context={'opponent_faced': 'Djené, Nemanja Gudelj', 'matchup_zone': 'Striker vs Two Centre Backs — 1v2 Aerial/Physical Battle', 'duel_win_rate': 0.214, 'pressing_faced': 'LOW', 'numerical_situation': '1v2', 'xg_multiplier': 1.0206, 'skill_edge': 0.044, 'height_edge': -0.03, 'attacker_skill': 0.5353, 'defender_skill': 0.4374, 'attacker_role': 'ST', 'attacker_mv_price_index': 1.0, 'defender_adjusted_parts': [{'player': 'Djené', 'market_value': 1500000, 'role': 'CB', 'price_index': 0.75, 'adjusted_quality': 0.4374}, {'player': 'Nemanja Gudelj', 'market_value': 1900000, 'role': 'CB', 'price_index': 0.75, 'adjusted_quality': 0.4374}], 'xg_override': 0.1125}	totalShots: 1.7682 -> 1.8046 shotOffTarget: 1.206 -> 1.2308 onTargetScoringAttempt: 0.5622 -> 0.5738
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 2 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=2`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 2.14 (squad xG sum 1.0517, xG scale 2.0376, xA scale 1.7540, team xA 1.71 = 0.80 x lambda)	`scoreline_reset=true` `team_lambda=2.143` `squad_xg_sum=1.0517` `squad_xa_sum=0.9774` `team_expected_assists=1.7144` `team_xa_to_xg_ratio=0.8`	expected_goals: 0.1125 -> 0.2292 expected_assists: 0.1352 -> 0.2371
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 6 by projected rating, fifa_overall 75.0)	`match_squad_rank=6` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 2 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=2` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.2048` `goal_rate=0.1578` `scorer_weight=0.2865`	-
`integer_assist_projection`	Team scoreline allocation: 2 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=2` `assists_allocated=1` `assist_probability=0.2111` `assist_rate=0.1895` `provider_weight=0.2964`	-

Alberto Moleiro F 77 1 0.60 0.16 279% 3 1.1 7.1 61.8 45 FBref Prior F_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost F_service raw output over 7 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0094, quality 0.837)	`market_value_eur=54000000` `quality_factor=0.837` `boost_factor=1.0094` `reference_quality=0.505` `history_matches=48` `history_shrink=0.0943`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0065)	`blend_weight=0.0065` `history_matches=48` `season_overlap_discount=1.0` `per90_to_match_factor=0.8587` `prior_reliability=0.1067` `competition=La Liga`	totalShots: 2.1428 -> 2.1288 wonTackle: 0.5379 -> 0.5449 onTargetScoringAttempt: 0.9432 -> 0.937
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` duel_context={'opponent_faced': 'Djené, Nemanja Gudelj', 'matchup_zone': 'Striker vs Two Centre Backs — 1v2 Aerial/Physical Battle', 'duel_win_rate': 0.338, 'pressing_faced': 'LOW', 'numerical_situation': '1v2', 'xg_multiplier': 1.2026, 'skill_edge': 0.1798, 'height_edge': -0.042, 'attacker_skill': 0.837, 'defender_skill': 0.4374, 'attacker_role': 'ST', 'attacker_mv_price_index': 1.0, 'defender_adjusted_parts': [{'player': 'Djené', 'market_value': 1500000, 'role': 'CB', 'price_index': 0.75, 'adjusted_quality': 0.4374}, {'player': 'Nemanja Gudelj', 'market_value': 1900000, 'role': 'CB', 'price_index': 0.75, 'adjusted_quality': 0.4374}], 'xg_override': 0.2922}	totalShots: 2.1366 -> 2.5695 shotOffTarget: 1.1996 -> 1.4426 onTargetScoringAttempt: 0.937 -> 1.1268 bigChanceMissed: 0.4984 -> 0.5994 expectedGoalsOnTarget: 0.3206 -> 0.3856 goals: 0.2865 -> 0.3445 expectedGoals: 0.243 -> 0.2922
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 2 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=2`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 2.14 (squad xG sum 1.0517, xG scale 2.0376, xA scale 1.7540, team xA 1.71 = 0.80 x lambda)	`scoreline_reset=true` `team_lambda=2.143` `squad_xg_sum=1.0517` `squad_xa_sum=0.9774` `team_expected_assists=1.7144` `team_xa_to_xg_ratio=0.8`	expected_goals: 0.2922 -> 0.5953 expected_assists: 0.183 -> 0.321
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 1 by projected rating, fifa_overall 79.0)	`match_squad_rank=1` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv_filled` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 2 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=2` `allocation_applied=true` `goals_allocated=1` `goal_probability=0.4486` `goal_rate=0.3445` `scorer_weight=0.6863`	-
`integer_assist_projection`	Team scoreline allocation: 2 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=2` `assists_allocated=1` `assist_probability=0.2746` `assist_rate=0.2791` `provider_weight=0.4012`	-

Team totals 2 2.14 16 7.0

Bench Candidates (7 modeled substitutes with empirical playing time)

Player	Pos	Exp Min	App Prob	Substitute Modeling Details
Péter Gulácsi	G	4.4m	10.9%	Market-value substitute prior
Carlos Romero	D	14.5m	23.6%	Market-value substitute prior
Santiago Mouriño	D	14.3m	23.2%	Market-value substitute prior
Nathan-Dylan Saliba	M	22.5m	52.2%	Market-value substitute prior
Alassane Diatta	M	22.1m	50.3%	Market-value substitute prior
Georges Mikautadze	F	26.7m	36.3%	Market-value substitute prior

Getafe — projected line vs season baseline (5-4-1)

Player Pos Min Proj G Proj xG Season xG/g Proj Sh Season Sh/g Proj Rtg Season Rtg Rec Model / FBref Prior

David Soria G 84 0 0.00 - n/a 0.0 6.9 66.0 34 FBref Prior G_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost G_service raw output over 3 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 0.9981, quality 0.437)	`market_value_eur=2700000` `quality_factor=0.4374` `boost_factor=0.9981` `reference_quality=0.505` `history_matches=48` `history_shrink=0.0943`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0123)	`blend_weight=0.0123` `history_matches=48` `season_overlap_discount=1.0` `per90_to_match_factor=0.9328` `prior_reliability=0.2` `competition=La Liga`	saves: 2.9518 -> 2.9461 wasFouled: 0.2123 -> 0.2173
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` `duel_context={'opponent_faced': 'Alberto Moleiro, Pape Gueye', 'matchup_zone': 'Goalkeeper shot-stopping context', 'opponent_lambda': 1.6044}`	-
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 0 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=0`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 0.84 (squad xG sum 0.7867, xG scale 1.0678, xA scale 1.0678, team xA 0.60 = 0.71 x lambda)	`scoreline_reset=true` `team_lambda=0.84` `squad_xg_sum=0.7867` `squad_xa_sum=0.5598` `team_expected_assists=0.5978` `team_xa_to_xg_ratio=0.7116`	shots_faced: 15.2 -> 20.4 expected_saves: 3.72 -> 5.0 high_pressure_saves: 1.28 -> 1.71
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 1 by projected rating, fifa_overall 80.0)	`match_squad_rank=1` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation skipped: 0 projected goals	`team_goals=0` `allocation_applied=false` `goals_allocated=0` `goal_probability=0.0` `scorer_weight=0.0` `squad_weight_total=0.8513`	-
`integer_assist_projection`	Team scoreline allocation skipped: 0 projected assists	`allocation_applied=false` `team_assists=0` `assists_allocated=0` `assist_probability=0.0` `provider_weight=0.0` `squad_weight_total=0.6041`	-

Zaid Romero D 80 0 0.02 0.04 37% 1 0.5 6.8 63.4 24 FBref Prior D_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost D_service raw output over 7 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 0.9975, quality 0.460)	`market_value_eur=4800000` `quality_factor=0.4597` `boost_factor=0.9975` `reference_quality=0.505` `history_matches=22` `history_shrink=0.1852`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0241)	`blend_weight=0.0241` `history_matches=22` `season_overlap_discount=1.0` `per90_to_match_factor=0.8944` `prior_reliability=0.2` `competition=La Liga`	totalCross: 0.3339 -> 0.3546 totalShots: 0.5349 -> 0.522 fouls: 0.7487 -> 0.7379
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` `duel_context={'opponent_faced': 'Ilias Akhomach', 'matchup_zone': 'Shape fallback'}`	-
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 0 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=0`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 0.84 (squad xG sum 0.7867, xG scale 1.0678, xA scale 1.0678, team xA 0.60 = 0.71 x lambda)	`scoreline_reset=true` `team_lambda=0.84` `squad_xg_sum=0.7867` `squad_xa_sum=0.5598` `team_expected_assists=0.5978` `team_xa_to_xg_ratio=0.7116`	-
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 6 by projected rating, fifa_overall 77.0)	`match_squad_rank=6` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation skipped: 0 projected goals	`team_goals=0` `allocation_applied=false` `goals_allocated=0` `goal_probability=0.0246` `goal_rate=0.0361` `scorer_weight=0.0292`	-
`integer_assist_projection`	Team scoreline allocation skipped: 0 projected assists	`allocation_applied=false` `team_assists=0` `assists_allocated=0` `assist_probability=0.0112` `assist_rate=0.0` `provider_weight=0.0093`	-

Djené D 75 0 0.00 - n/a 0 0.0 6.6 58.1 29 FBref Prior D_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost D_service raw output over 7 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 0.9979, quality 0.437)	`market_value_eur=1500000` `quality_factor=0.4374` `boost_factor=0.9979` `reference_quality=0.505` `history_matches=43` `history_shrink=0.1042`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0131)	`blend_weight=0.0131` `history_matches=43` `season_overlap_discount=1.0` `per90_to_match_factor=0.8328` `prior_reliability=0.1933` `competition=La Liga`	fouls: 0.9704 -> 0.9765
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` `duel_context={'opponent_faced': 'Tajon Buchanan', 'matchup_zone': 'LM vs CB'}`	-
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 0 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=0`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 0.84 (squad xG sum 0.7867, xG scale 1.0678, xA scale 1.0678, team xA 0.60 = 0.71 x lambda)	`scoreline_reset=true` `team_lambda=0.84` `squad_xg_sum=0.7867` `squad_xa_sum=0.5598` `team_expected_assists=0.5978` `team_xa_to_xg_ratio=0.7116`	-
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 17 by projected rating, fifa_overall 79.0)	`match_squad_rank=17` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv_filled` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation skipped: 0 projected goals	`team_goals=0` `allocation_applied=false` `goals_allocated=0` `goal_probability=0.0003` `goal_rate=0.0` `scorer_weight=0.001`	-
`integer_assist_projection`	Team scoreline allocation skipped: 0 projected assists	`allocation_applied=false` `team_assists=0` `assists_allocated=0` `assist_probability=0.01` `assist_rate=0.0` `provider_weight=0.0085`	-

Kiko Femenía D 62 0 0.02 0.02 ok 0 0.2 6.5 56.1 25 FBref Prior D_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost D_service raw output over 7 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 0.9974, quality 0.437)	`market_value_eur=525000` `quality_factor=0.4374` `boost_factor=0.9974` `reference_quality=0.505` `history_matches=34` `history_shrink=0.1282`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0083)	`blend_weight=0.0083` `history_matches=34` `season_overlap_discount=1.0` `per90_to_match_factor=0.689` `prior_reliability=0.1` `competition=La Liga`	interceptionWon: 0.7846 -> 0.7781
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` `duel_context={'opponent_faced': 'Alberto Moleiro', 'matchup_zone': 'Shape fallback'}`	-
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 0 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=0`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 0.84 (squad xG sum 0.7867, xG scale 1.0678, xA scale 1.0678, team xA 0.60 = 0.71 x lambda)	`scoreline_reset=true` `team_lambda=0.84` `squad_xg_sum=0.7867` `squad_xa_sum=0.5598` `team_expected_assists=0.5978` `team_xa_to_xg_ratio=0.7116`	-
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 18 by projected rating, fifa_overall 74.0)	`match_squad_rank=18` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation skipped: 0 projected goals	`team_goals=0` `allocation_applied=false` `goals_allocated=0` `goal_probability=0.0189` `goal_rate=0.0266` `scorer_weight=0.0215`	-
`integer_assist_projection`	Team scoreline allocation skipped: 0 projected assists	`allocation_applied=false` `team_assists=0` `assists_allocated=0` `assist_probability=0.0532` `assist_rate=0.0036` `provider_weight=0.0438`	-

Johan Mojica D 80 0 0.02 0.01 44% 1 0.4 6.7 59.9 31 FBref Prior D_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost D_service raw output over 7 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 0.9979, quality 0.437)	`market_value_eur=1400000` `quality_factor=0.4374` `boost_factor=0.9979` `reference_quality=0.505` `history_matches=44` `history_shrink=0.102`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0088)	`blend_weight=0.0088` `history_matches=44` `season_overlap_discount=1.0` `per90_to_match_factor=0.8896` `prior_reliability=0.1333` `competition=La Liga`	totalCross: 6.4211 -> 6.443 wasFouled: 1.174 -> 1.1872
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` `duel_context={'opponent_faced': 'Alberto Moleiro', 'matchup_zone': 'Shape fallback'}`	-
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 0 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=0`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 0.84 (squad xG sum 0.7867, xG scale 1.0678, xA scale 1.0678, team xA 0.60 = 0.71 x lambda)	`scoreline_reset=true` `team_lambda=0.84` `squad_xg_sum=0.7867` `squad_xa_sum=0.5598` `team_expected_assists=0.5978` `team_xa_to_xg_ratio=0.7116`	expected_assists: 0.0925 -> 0.0988
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 9 by projected rating, fifa_overall 76.0)	`match_squad_rank=9` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation skipped: 0 projected goals	`team_goals=0` `allocation_applied=false` `goals_allocated=0` `goal_probability=0.0212` `goal_rate=0.0` `scorer_weight=0.0171`	-
`integer_assist_projection`	Team scoreline allocation skipped: 0 projected assists	`allocation_applied=false` `team_assists=0` `assists_allocated=0` `assist_probability=0.0941` `assist_rate=0.0315` `provider_weight=0.079`	-

Nemanja Gudelj D 81 0 0.04 0.03 40% 1 0.8 6.7 56.3 42 FBref Prior D_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost D_service raw output over 7 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 0.9978, quality 0.437)	`market_value_eur=1900000` `quality_factor=0.4374` `boost_factor=0.9978` `reference_quality=0.505` `history_matches=42` `history_shrink=0.1064`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0009)	`blend_weight=0.0009` `history_matches=42` `season_overlap_discount=1.0` `per90_to_match_factor=0.9048` `prior_reliability=0.0133` `competition=La Liga`	-
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` `duel_context={'opponent_faced': 'Ilias Akhomach', 'matchup_zone': 'Striker vs Two Centre Backs — 1v2 Aerial/Physical Battle'}`	-
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 0 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=0`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 0.84 (squad xG sum 0.7867, xG scale 1.0678, xA scale 1.0678, team xA 0.60 = 0.71 x lambda)	`scoreline_reset=true` `team_lambda=0.84` `squad_xg_sum=0.7867` `squad_xa_sum=0.5598` `team_expected_assists=0.5978` `team_xa_to_xg_ratio=0.7116`	-
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 11 by projected rating, fifa_overall 78.0)	`match_squad_rank=11` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv_filled` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation skipped: 0 projected goals	`team_goals=0` `allocation_applied=false` `goals_allocated=0` `goal_probability=0.0406` `goal_rate=0.027` `scorer_weight=0.0359`	-
`integer_assist_projection`	Team scoreline allocation skipped: 0 projected assists	`allocation_applied=false` `team_assists=0` `assists_allocated=0` `assist_probability=0.0231` `assist_rate=0.0` `provider_weight=0.0187`	-

Mario Martín M 72 0 0.08 0.08 ok 1 0.7 6.7 59.9 28 FBref Prior M_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost M_service raw output over 8 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0008, quality 0.531)	`market_value_eur=10700000` `quality_factor=0.5306` `boost_factor=1.0008` `reference_quality=0.505` `history_matches=42` `history_shrink=0.1064`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0032)	`blend_weight=0.0032` `history_matches=42` `season_overlap_discount=1.0` `per90_to_match_factor=0.804` `prior_reliability=0.0467` `competition=La Liga`	wasFouled: 1.2657 -> 1.2764
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` duel_context={'opponent_faced': 'Pape Gueye', 'matchup_zone': 'Central Mid vs Central Mid — Box-to-Box Midfield Contest', 'duel_win_rate': 0.266, 'pressing_faced': 'HIGH', 'numerical_situation': '1v1', 'xg_multiplier': 0.774, 'skill_edge': -0.1138, 'height_edge': -0.04, 'attacker_skill': 0.5444, 'defender_skill': 0.7972, 'attacker_role': 'CM', 'attacker_mv_price_index': 0.9, 'defender_adjusted_parts': [{'player': 'Pape Gueye', 'market_value': 39000000, 'role': 'CM', 'price_index': 0.9, 'adjusted_quality': 0.7972}], 'xg_override': 0.0774}	aerialWon: 1.4382 -> 1.0 totalShots: 1.1238 -> 0.8698 shotOffTarget: 0.735 -> 0.5689 onTargetScoringAttempt: 0.3888 -> 0.3009 bigChanceMissed: 0.1342 -> 0.1039 expectedGoals: 0.1 -> 0.0774 goals: 0.0896 -> 0.0694 expectedGoalsOnTarget: 0.0249 -> 0.0193
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 0 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=0`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 0.84 (squad xG sum 0.7867, xG scale 1.0678, xA scale 1.0678, team xA 0.60 = 0.71 x lambda)	`scoreline_reset=true` `team_lambda=0.84` `squad_xg_sum=0.7867` `squad_xa_sum=0.5598` `team_expected_assists=0.5978` `team_xa_to_xg_ratio=0.7116`	expected_goals: 0.0774 -> 0.0826
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 12 by projected rating, fifa_overall 74.0)	`match_squad_rank=12` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation skipped: 0 projected goals	`team_goals=0` `allocation_applied=false` `goals_allocated=0` `goal_probability=0.0793` `goal_rate=0.0694` `scorer_weight=0.0774`	-
`integer_assist_projection`	Team scoreline allocation skipped: 0 projected assists	`allocation_applied=false` `team_assists=0` `assists_allocated=0` `assist_probability=0.0487` `assist_rate=0.0645` `provider_weight=0.0591`	-

Ramón Terrats M 71 0 0.15 0.07 103% 1 1.0 6.8 59.2 32 FBref Prior M_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost M_service raw output over 8 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 0.9977, quality 0.440)	`market_value_eur=3200000` `quality_factor=0.4399` `boost_factor=0.9977` `reference_quality=0.505` `history_matches=38` `history_shrink=0.1163`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0151)	`blend_weight=0.0151` `history_matches=38` `season_overlap_discount=1.0` `per90_to_match_factor=0.7922` `prior_reliability=0.2` `competition=La Liga`	totalCross: 1.4756 -> 1.5052 totalShots: 1.6682 -> 1.6831 wasFouled: 2.0806 -> 2.0891 fouls: 0.8907 -> 0.8972
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` duel_context={'opponent_faced': 'Sergi Cardona', 'matchup_zone': 'Left Mid vs Right Back — Wide Channel', 'duel_win_rate': 0.385, 'pressing_faced': 'LOW', 'numerical_situation': '1v1', 'xg_multiplier': 0.8897, 'skill_edge': -0.0471, 'height_edge': -0.028, 'attacker_skill': 0.4443, 'defender_skill': 0.5489, 'attacker_role': 'LM', 'attacker_mv_price_index': 0.9, 'defender_adjusted_parts': [{'player': 'Sergi Cardona', 'market_value': 8600000, 'role': 'RB', 'price_index': 0.7, 'adjusted_quality': 0.5489}], 'xg_override': 0.1394}	aerialWon: 0.7146 -> 1.0 totalShots: 1.6831 -> 1.4975 shotOffTarget: 1.1314 -> 1.0066 onTargetScoringAttempt: 0.5367 -> 0.4775 bigChanceMissed: 0.1694 -> 0.1507 expectedGoals: 0.1567 -> 0.1394 goals: 0.1482 -> 0.1319 expectedGoalsOnTarget: 0.141 -> 0.1254
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 0 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=0`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 0.84 (squad xG sum 0.7867, xG scale 1.0678, xA scale 1.0678, team xA 0.60 = 0.71 x lambda)	`scoreline_reset=true` `team_lambda=0.84` `squad_xg_sum=0.7867` `squad_xa_sum=0.5598` `team_expected_assists=0.5978` `team_xa_to_xg_ratio=0.7116`	expected_goals: 0.1394 -> 0.1488 expected_assists: 0.1046 -> 0.1116
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 3 by projected rating, fifa_overall 82.0)	`match_squad_rank=3` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv_filled` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation skipped: 0 projected goals	`team_goals=0` `allocation_applied=false` `goals_allocated=0` `goal_probability=0.1383` `goal_rate=0.1319` `scorer_weight=0.1429`	-
`integer_assist_projection`	Team scoreline allocation skipped: 0 projected assists	`allocation_applied=false` `team_assists=0` `assists_allocated=0` `assist_probability=0.1056` `assist_rate=0.1106` `provider_weight=0.1159`	-

Francho Serrano M 79 0 0.06 0.01 395% 1 0.3 6.8 44.7 12 FBref Prior M_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost M_service raw output over 8 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 0.9923, quality 0.439)	`market_value_eur=3100000` `quality_factor=0.4387` `boost_factor=0.9923` `reference_quality=0.505` `history_matches=8` `history_shrink=0.3846`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0300)	`blend_weight=0.03` `history_matches=8` `season_overlap_discount=1.0` `per90_to_match_factor=0.8789` `prior_reliability=0.12` `competition=La Liga`	totalCross: 1.3016 -> 1.3358 totalShots: 0.8835 -> 0.857 interceptionWon: 0.6562 -> 0.6805 wonTackle: 0.1061 -> 0.1176 onTargetScoringAttempt: 0.2622 -> 0.2543
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` duel_context={'opponent_faced': 'Juan Foyth', 'matchup_zone': 'Right Mid vs Left Back — Wide Channel', 'duel_win_rate': 0.372, 'pressing_faced': 'LOW', 'numerical_situation': '1v1', 'xg_multiplier': 0.8711, 'skill_edge': -0.0477, 'height_edge': -0.04, 'attacker_skill': 0.4429, 'defender_skill': 0.5489, 'attacker_role': 'RM', 'attacker_mv_price_index': 0.9, 'defender_adjusted_parts': [{'player': 'Juan Foyth', 'market_value': 8600000, 'role': 'LB', 'price_index': 0.7, 'adjusted_quality': 0.5489}], 'xg_override': 0.0524}	aerialWon: 0.1727 -> 1.0 totalShots: 0.8756 -> 0.7627 shotOffTarget: 0.6213 -> 0.5412 onTargetScoringAttempt: 0.2543 -> 0.2215 goals: 0.0641 -> 0.0558 expectedGoals: 0.0602 -> 0.0524
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 0 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=0`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 0.84 (squad xG sum 0.7867, xG scale 1.0678, xA scale 1.0678, team xA 0.60 = 0.71 x lambda)	`scoreline_reset=true` `team_lambda=0.84` `squad_xg_sum=0.7867` `squad_xa_sum=0.5598` `team_expected_assists=0.5978` `team_xa_to_xg_ratio=0.7116`	-
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 5 by projected rating, fifa_overall 79.0)	`match_squad_rank=5` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv_filled` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation skipped: 0 projected goals	`team_goals=0` `allocation_applied=false` `goals_allocated=0` `goal_probability=0.0545` `goal_rate=0.0558` `scorer_weight=0.0579`	-
`integer_assist_projection`	Team scoreline allocation skipped: 0 projected assists	`allocation_applied=false` `team_assists=0` `assists_allocated=0` `assist_probability=0.0732` `assist_rate=0.0712` `provider_weight=0.076`	-

Orel Mangala M 87 0 0.04 0.02 124% 1 0.3 6.8 46.0 10 FBref Prior M_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost M_service raw output over 8 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0001, quality 0.506)	`market_value_eur=8600000` `quality_factor=0.5059` `boost_factor=1.0001` `reference_quality=0.505` `history_matches=7` `history_shrink=0.4167`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0542)	`blend_weight=0.0542` `history_matches=7` `season_overlap_discount=1.0` `per90_to_match_factor=0.9683` `prior_reliability=0.2` `competition=La Liga`	interceptionWon: 2.4551 -> 2.4095 wasFouled: 0.9361 -> 0.9728 totalCross: 0.4455 -> 0.4214 wonTackle: 0.2326 -> 0.255 totalShots: 0.6792 -> 0.6949 onTargetScoringAttempt: 0.18 -> 0.1702 fouls: 2.091 -> 2.1001
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` duel_context={'opponent_faced': 'Pape Gueye', 'matchup_zone': 'Central Mid vs Central Mid — Box-to-Box Midfield Contest', 'duel_win_rate': 0.258, 'pressing_faced': 'HIGH', 'numerical_situation': '1v1', 'xg_multiplier': 0.7619, 'skill_edge': -0.126, 'height_edge': -0.036, 'attacker_skill': 0.5172, 'defender_skill': 0.7972, 'attacker_role': 'CM', 'attacker_mv_price_index': 0.9, 'defender_adjusted_parts': [{'player': 'Pape Gueye', 'market_value': 39000000, 'role': 'CM', 'price_index': 0.9, 'adjusted_quality': 0.7972}], 'xg_override': 0.0347}	aerialWon: 0.3877 -> 1.0 totalShots: 0.6949 -> 0.5294 shotOffTarget: 0.4992 -> 0.3803 onTargetScoringAttempt: 0.1702 -> 0.1297 expectedGoals: 0.0456 -> 0.0347 goals: 0.044 -> 0.0335
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 0 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=0`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 0.84 (squad xG sum 0.7867, xG scale 1.0678, xA scale 1.0678, team xA 0.60 = 0.71 x lambda)	`scoreline_reset=true` `team_lambda=0.84` `squad_xg_sum=0.7867` `squad_xa_sum=0.5598` `team_expected_assists=0.5978` `team_xa_to_xg_ratio=0.7116`	-
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 7 by projected rating, fifa_overall 75.0)	`match_squad_rank=7` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation skipped: 0 projected goals	`team_goals=0` `allocation_applied=false` `goals_allocated=0` `goal_probability=0.0364` `goal_rate=0.0335` `scorer_weight=0.0366`	-
`integer_assist_projection`	Team scoreline allocation skipped: 0 projected assists	`allocation_applied=false` `team_assists=0` `assists_allocated=0` `assist_probability=0.0445` `assist_rate=0.0585` `provider_weight=0.0533`	-

Martín Satriano F 77 0 0.20 0.16 ok 2 1.7 6.8 61.8 34 FBref Prior F_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost F_service raw output over 7 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0011, quality 0.529)	`market_value_eur=10600000` `quality_factor=0.5295` `boost_factor=1.0011` `reference_quality=0.505` `history_matches=28` `history_shrink=0.1515`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0177)	`blend_weight=0.0177` `history_matches=28` `season_overlap_discount=1.0` `per90_to_match_factor=0.8584` `prior_reliability=0.18` `competition=La Liga`	wasFouled: 1.9834 -> 1.9933 interceptionWon: 0.5072 -> 0.4982 onTargetScoringAttempt: 0.6281 -> 0.6226
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` duel_context={'opponent_faced': 'Renato Veiga, Pau Navarro', 'matchup_zone': 'Striker vs Two Centre Backs — 1v2 Aerial/Physical Battle', 'duel_win_rate': 0.18, 'pressing_faced': 'MEDIUM', 'numerical_situation': '1v2', 'xg_multiplier': 0.8845, 'skill_edge': -0.0766, 'height_edge': -0.002, 'attacker_skill': 0.5295, 'defender_skill': 0.6997, 'attacker_role': 'ST', 'attacker_mv_price_index': 1.0, 'defender_adjusted_parts': [{'player': 'Renato Veiga', 'market_value': 29000000, 'role': 'CB', 'price_index': 0.75, 'adjusted_quality': 0.7736}, {'player': 'Pau Navarro', 'market_value': 14600000, 'role': 'CB', 'price_index': 0.75, 'adjusted_quality': 0.6258}], 'xg_override': 0.1888}	totalShots: 1.869 -> 1.6531 shotOffTarget: 1.2405 -> 1.0972 onTargetScoringAttempt: 0.6226 -> 0.5507 expectedGoalsOnTarget: 0.2644 -> 0.2339 goals: 0.2254 -> 0.1994 expectedGoals: 0.2135 -> 0.1888 bigChanceMissed: 0.0938 -> 0.083
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 0 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=0`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 0.84 (squad xG sum 0.7867, xG scale 1.0678, xA scale 1.0678, team xA 0.60 = 0.71 x lambda)	`scoreline_reset=true` `team_lambda=0.84` `squad_xg_sum=0.7867` `squad_xa_sum=0.5598` `team_expected_assists=0.5978` `team_xa_to_xg_ratio=0.7116`	expected_goals: 0.1888 -> 0.2016
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 2 by projected rating, fifa_overall 77.0)	`match_squad_rank=2` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation skipped: 0 projected goals	`team_goals=0` `allocation_applied=false` `goals_allocated=0` `goal_probability=0.1826` `goal_rate=0.1994` `scorer_weight=0.2106`	-
`integer_assist_projection`	Team scoreline allocation skipped: 0 projected assists	`allocation_applied=false` `team_assists=0` `assists_allocated=0` `assist_probability=0.0577` `assist_rate=0.119` `provider_weight=0.0742`	-

Team totals 0 0.84 11 6.7

Bench Candidates (7 modeled substitutes with empirical playing time)

Player	Pos	Exp Min	App Prob	Substitute Modeling Details
Jiří Letáček	G	4.4m	10.9%	Market-value substitute prior
Andrés García	D	13.8m	22.2%	Market-value substitute prior
Dani Carvajal	D	13.6m	30.0%	Market-value substitute prior
Alberto Risco	M	19.2m	69.2%	Market-value substitute prior
Oscar Lopez	M	22.0m	49.9%	Market-value substitute prior
Christantus Uche	F	25.6m	57.4%	Market-value substitute prior

Antagonist / tactical summary

Midfield is evenly contested — Pape Gueye vs Mario Martín will be the pivotal duel. Total antagonist xG — Home: 0.89, Away: 0.49. Home danger zone: Striker vs Two Centre Backs — 1v2 Aerial/Physical Battle. Away danger zone: Striker vs Two Centre Backs — 1v2 Aerial/Physical Battle.

Midfield — possession 51.8% / 48.2%, dominance ratio 0.56, key duel: Pape Gueye vs Mario Martín

Tajon Buchanan xG 0.19Pape Gueye xG 0.20Carlos Maciá xG 0.05Martín Satriano xG 0.19Ramón Terrats xG 0.14Francho Serrano xG 0.05

Summary

Antagonist matchup analysis for Villarreal vs Getafe. Most likely scoreline: 2-0. Lambdas derived from 6 home and 5 away individual player duels. Head-to-head: Villarreal 5W-4D-1L vs Getafe over 10 meetings.

Tactical parity between 4-4-2 and 5-4-1. Match outcome driven primarily by individual player quality. Midfield is evenly contested — Pape Gueye vs Mario Martín will be the pivotal duel. Total antagonist xG — Home: 0.89, Away: 0.49. Home danger zone: Striker vs Two Centre Backs — 1v2 Aerial/Physical Battle. Away danger zone: Striker vs Two Centre Backs — 1v2 Aerial/Physical Battle.

Celta vs Levante

LaLiga · 20261108 · 3-4-3 vs 4-4-2

Predicted 2-1Winner: CeltaWinning H 2-1 (9.6%)Winning A 1-2 (6.6%)Confidence 0.68λ home/away 1.76 / 1.22Most-likely P 9.6%

Actual vs predicted: result not available yet (upcoming fixture).

XGBoost Reasoning & FBref Dataset Tracing

Antagonist + XGBoost match analysis for Celta vs Levante. Player statistics regressed using position-specific XGBoost models (G, D, M, F) with 2026-2027 FBref priors. Micro-duels resolved across 3-4-3 vs 4-4-2 (duel lambdas: Home 1.366, Away 0.838). Final calibrated Poisson lambdas: Home 1.764, Away 1.220 (most likely score 2-1).

Home Starters 11/11 XGBoost (10 FBref) Away Starters 11/11 XGBoost (0 FBref) Bench Modeled H: 7 / A: 7 Duel λ (H/A) 1.37 / 0.84 Final λ (H/A) 1.76 / 1.22

FBref Team Priors: Celta Vigo (Gls/90: 0.25, GA/90: 0.25 → E[λ]: 0.309) vs Levante UD (Gls/90: 1.6666666666666667, GA/90: 1.67 → E[λ]: 0.309) | Blend Weight: 0.228

Modeling Pipeline Stages (5 layers)

1. XGBoost player regressions for volume & event rates with market-value quality scaling
2. Multi-source prior blending (FBref 2026-2027 tables + Sofascore season history)
3. Antagonist formation duel matrix resolving 1v1 and zonal matchups
4. Team-level prior calibration (FBref possession and attack/defense efficiencies)
5. Bivariate Poisson matrix with Dixon-Coles low-score correction

Standing & form context

Metric	Celta	Levante
League position	9.00	17.00
Points per game	1.19	1.07
Attack index	0.85	0.94
Defense (concede) index	1.08	0.85
Form multiplier	1.01	0.99
Standing λ multiplier	0.97	0.91

Lambda decomposition (how the goal rates were built)

Stage	Celta λ	Levante λ
FBref Squad Prior E[λ] (weight 0.228)	0.309	0.309
Pre-competitive-floor	1.325	0.922
Pre-calibration (Micro-duels)	1.366	0.838
Final λ	1.764	1.220

competition calibration scale 1.328competition observed total goals 2.984quality gap applied Truemismatch score pressure applied Falsecompetitive underdog floor applied Falsestage discount 1.0

Scoreline distribution & markets

Top 5 scorelines	P
1-1	11.0%
2-1	9.6%
1-0	8.8%
2-0	7.9%
1-2	6.6%

Home win	50.1%
Draw	23.5%
Away win	26.4%
Over 2.5	57.3%
BTTS yes	58.5%

Pregame fair line — favorite Celta (50.1%), fair total 2.98, fair spread 0.54

Projected players vs season baseline

Proj = projected for this match · Season = this season's per-game average from match data · deviation badge compares projected xG to the player's season xG/game.

Celta — projected line vs season baseline (3-4-3)

Player Pos Min Proj G Proj xG Season xG/g Proj Sh Season Sh/g Proj Rtg Season Rtg Rec Model / FBref Prior

Ionuț Radu G 90 0 0.00 - n/a 0.0 6.9 65.9 34 FBref Prior G_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost G_service raw output over 3 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0004, quality 0.519)	`market_value_eur=9700000` `quality_factor=0.5189` `boost_factor=1.0004` `reference_quality=0.505` `history_matches=49` `history_shrink=0.0926`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0161)	`blend_weight=0.0161` `history_matches=49` `season_overlap_discount=1.0` `per90_to_match_factor=1.0` `prior_reliability=0.2667` `competition=La Liga`	saves: 3.0356 -> 3.0431
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` `duel_context={'opponent_faced': 'Roger Brugué, Iván Romero', 'matchup_zone': 'Goalkeeper shot-stopping context', 'opponent_lambda': 0.838}`	-
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 2 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=2`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.76 (squad xG sum 0.9840, xG scale 1.7928, xA scale 1.7928, team xA 1.31 = 0.74 x lambda)	`scoreline_reset=true` `team_lambda=1.764` `squad_xg_sum=0.984` `squad_xa_sum=0.7314` `team_expected_assists=1.3112` `team_xa_to_xg_ratio=0.7433`	shots_faced: 8.0 -> 11.6 expected_saves: 1.96 -> 2.84 high_pressure_saves: 0.67 -> 0.98
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 4 by projected rating, fifa_overall 78.0)	`match_squad_rank=4` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 2 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=2` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.0` `scorer_weight=0.0` `squad_weight_total=1.8097`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.0` `provider_weight=0.0` `squad_weight_total=1.4371`	-

Yoel Lago D 90 0 0.03 0.01 363% 0 0.1 6.9 44.9 29 FBref Prior D_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost D_service raw output over 7 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 0.9990, quality 0.487)	`market_value_eur=7000000` `quality_factor=0.4866` `boost_factor=0.999` `reference_quality=0.505` `history_matches=23` `history_shrink=0.1786`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0271)	`blend_weight=0.0271` `history_matches=23` `season_overlap_discount=1.0` `per90_to_match_factor=0.9947` `prior_reliability=0.2333` `competition=La Liga`	totalShots: 0.3835 -> 0.3731 wasFouled: 1.4375 -> 1.4294 totalCross: 0.0 -> 0.0077 interceptionWon: 1.1603 -> 1.1674 fouls: 1.4484 -> 1.4554 wonTackle: 0.8274 -> 0.8204
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` `duel_context={'opponent_faced': 'Jon Ander Olasagasti', 'matchup_zone': 'LM vs CB'}`	-
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 2 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=2`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.76 (squad xG sum 0.9840, xG scale 1.7928, xA scale 1.7928, team xA 1.31 = 0.74 x lambda)	`scoreline_reset=true` `team_lambda=1.764` `squad_xg_sum=0.984` `squad_xa_sum=0.7314` `team_expected_assists=1.3112` `team_xa_to_xg_ratio=0.7433`	expected_goals: 0.0151 -> 0.0271 expected_assists: 0.0111 -> 0.0199
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 5 by projected rating, fifa_overall 71.0)	`match_squad_rank=5` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 2 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=2` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.0267` `goal_rate=0.0` `scorer_weight=0.0217`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.0197` `assist_rate=0.0` `provider_weight=0.0163`	-

Álvaro Núñez D 81 0 0.03 0.01 332% 1 0.3 6.9 33.0 27 FBref Prior D_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost D_service raw output over 7 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 0.9975, quality 0.449)	`market_value_eur=3900000` `quality_factor=0.4486` `boost_factor=0.9975` `reference_quality=0.505` `history_matches=29` `history_shrink=0.1471`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0121)	`blend_weight=0.0121` `history_matches=29` `season_overlap_discount=1.0` `per90_to_match_factor=0.9022` `prior_reliability=0.1267` `competition=La Liga`	interceptionWon: 0.8183 -> 0.8256 totalCross: 1.49 -> 1.4835 wonTackle: 0.9029 -> 0.8977
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` `duel_context={'opponent_faced': 'Roger Brugué', 'matchup_zone': 'Striker vs Two Centre Backs — 1v2 Aerial/Physical Battle'}`	-
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 2 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=2`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.76 (squad xG sum 0.9840, xG scale 1.7928, xA scale 1.7928, team xA 1.31 = 0.74 x lambda)	`scoreline_reset=true` `team_lambda=1.764` `squad_xg_sum=0.984` `squad_xa_sum=0.7314` `team_expected_assists=1.3112` `team_xa_to_xg_ratio=0.7433`	expected_goals: 0.014 -> 0.0251 expected_assists: 0.0588 -> 0.1054
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 7 by projected rating, fifa_overall 76.0)	`match_squad_rank=7` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 2 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=2` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.0248` `goal_rate=0.0` `scorer_weight=0.0201`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.1` `assist_rate=0.0204` `provider_weight=0.0843`	-

Javi Rodríguez D 85 0 0.03 0.02 79% 0 0.3 7.0 54.2 33 FBref Prior D_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost D_service raw output over 7 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0034, quality 0.606)	`market_value_eur=17500000` `quality_factor=0.6058` `boost_factor=1.0034` `reference_quality=0.505` `history_matches=40` `history_shrink=0.1111`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0125)	`blend_weight=0.0125` `history_matches=40` `season_overlap_discount=1.0` `per90_to_match_factor=0.9456` `prior_reliability=0.1733` `competition=La Liga`	wasFouled: 1.0938 -> 1.0847
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` `duel_context={'opponent_faced': 'Iván Romero', 'matchup_zone': 'Shape fallback'}`	-
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 2 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=2`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.76 (squad xG sum 0.9840, xG scale 1.7928, xA scale 1.7928, team xA 1.31 = 0.74 x lambda)	`scoreline_reset=true` `team_lambda=1.764` `squad_xg_sum=0.984` `squad_xa_sum=0.7314` `team_expected_assists=1.3112` `team_xa_to_xg_ratio=0.7433`	expected_goals: 0.0168 -> 0.0301 expected_assists: 0.0817 -> 0.1465
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 3 by projected rating, fifa_overall 76.0)	`match_squad_rank=3` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 2 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=2` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.0297` `goal_rate=0.1162` `scorer_weight=0.0376`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.1363` `assist_rate=0.1231` `provider_weight=0.1831`	-

Ilaix Moriba M 82 0 0.07 0.03 130% 1 0.4 6.9 63.9 32 FBref Prior M_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost M_service raw output over 8 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0017, quality 0.565)	`market_value_eur=13700000` `quality_factor=0.5649` `boost_factor=1.0017` `reference_quality=0.505` `history_matches=47` `history_shrink=0.0962`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0112)	`blend_weight=0.0112` `history_matches=47` `season_overlap_discount=1.0` `per90_to_match_factor=0.913` `prior_reliability=0.18` `competition=La Liga`	fouls: 1.8036 -> 1.7833 interceptionWon: 1.4066 -> 1.3984
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` duel_context={'opponent_faced': 'Daniel Requena', 'matchup_zone': 'Central Mid vs Central Mid — Box-to-Box Midfield Contest', 'duel_win_rate': 0.473, 'pressing_faced': 'LOW', 'numerical_situation': '1v1', 'xg_multiplier': 1.0777, 'skill_edge': 0.0649, 'height_edge': -0.012, 'attacker_skill': 0.5816, 'defender_skill': 0.4374, 'attacker_role': 'CM', 'attacker_mv_price_index': 0.9, 'defender_adjusted_parts': [{'player': 'Daniel Requena', 'market_value': 1500000, 'role': 'CM', 'price_index': 0.9, 'adjusted_quality': 0.4374}], 'xg_override': 0.04}	aerialWon: 0.1643 -> 1.1 totalShots: 0.5356 -> 0.5772 shotOffTarget: 0.3895 -> 0.4198 onTargetScoringAttempt: 0.1428 -> 0.1539 expectedGoalsOnTarget: 0.0729 -> 0.0786
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 2 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=2`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.76 (squad xG sum 0.9840, xG scale 1.7928, xA scale 1.7928, team xA 1.31 = 0.74 x lambda)	`scoreline_reset=true` `team_lambda=1.764` `squad_xg_sum=0.984` `squad_xa_sum=0.7314` `team_expected_assists=1.3112` `team_xa_to_xg_ratio=0.7433`	expected_goals: 0.04 -> 0.0717 expected_assists: 0.0409 -> 0.0733
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 6 by projected rating, fifa_overall 78.0)	`match_squad_rank=6` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 2 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=2` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.0692` `goal_rate=0.0464` `scorer_weight=0.0768`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.0707` `assist_rate=0.0491` `provider_weight=0.0799`	-

Miguel Román M 88 0 0.11 0.01 980% 1 0.4 7.1 64.0 19 FBref Prior M_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost M_service raw output over 8 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0033, quality 0.573)	`market_value_eur=14400000` `quality_factor=0.5726` `boost_factor=1.0033` `reference_quality=0.505` `history_matches=26` `history_shrink=0.1613`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0245)	`blend_weight=0.0245` `history_matches=26` `season_overlap_discount=1.0` `per90_to_match_factor=0.9833` `prior_reliability=0.2333` `competition=La Liga`	totalCross: 1.6349 -> 1.6842 totalShots: 0.979 -> 0.9551 wonTackle: 0.9445 -> 0.9214 wasFouled: 0.5078 -> 0.516 onTargetScoringAttempt: 0.3311 -> 0.323
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=true`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` duel_context={'opponent_faced': 'Daniel Requena', 'matchup_zone': 'Central Mid vs Central Mid — Box-to-Box Midfield Contest', 'duel_win_rate': 0.457, 'pressing_faced': 'LOW', 'numerical_situation': '1v1', 'xg_multiplier': 1.0539, 'skill_edge': 0.0686, 'height_edge': -0.032, 'attacker_skill': 0.59, 'defender_skill': 0.4374, 'attacker_role': 'CM', 'attacker_mv_price_index': 0.9, 'defender_adjusted_parts': [{'player': 'Daniel Requena', 'market_value': 1500000, 'role': 'CM', 'price_index': 0.9, 'adjusted_quality': 0.4374}], 'xg_override': 0.0624}	aerialWon: 0.6671 -> 1.0 totalShots: 0.9709 -> 1.0232 shotOffTarget: 0.6479 -> 0.6828 onTargetScoringAttempt: 0.323 -> 0.3404
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 2 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=2`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.76 (squad xG sum 0.9840, xG scale 1.7928, xA scale 1.7928, team xA 1.31 = 0.74 x lambda)	`scoreline_reset=true` `team_lambda=1.764` `squad_xg_sum=0.984` `squad_xa_sum=0.7314` `team_expected_assists=1.3112` `team_xa_to_xg_ratio=0.7433`	expected_goals: 0.0624 -> 0.1119 expected_assists: 0.1156 -> 0.2073
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 1 by projected rating, fifa_overall 74.0)	`match_squad_rank=1` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 2 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=2` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.1059` `goal_rate=0.0799` `scorer_weight=0.1293`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.1872` `assist_rate=0.1168` `provider_weight=0.2088`	-

Aleix Febas M 85 0 0.07 0.03 117% 1 0.4 6.9 60.6 29 FBref Prior M_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost M_service raw output over 8 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 0.9987, quality 0.465)	`market_value_eur=5200000` `quality_factor=0.4646` `boost_factor=0.9987` `reference_quality=0.505` `history_matches=42` `history_shrink=0.1064`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0083)	`blend_weight=0.0083` `history_matches=42` `season_overlap_discount=1.0` `per90_to_match_factor=0.9444` `prior_reliability=0.12` `competition=La Liga`	fouls: 1.5384 -> 1.5474 wasFouled: 2.8447 -> 2.8516
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` duel_context={'opponent_faced': 'Adrián de la Fuente', 'matchup_zone': 'Left Wingback vs Right Back — Overlapping Run', 'duel_win_rate': 0.36, 'pressing_faced': 'LOW', 'numerical_situation': '1v1', 'xg_multiplier': 0.9412, 'skill_edge': 0.0, 'height_edge': -0.04, 'attacker_skill': 0.4918, 'defender_skill': 0.4918, 'attacker_role': 'LWB', 'attacker_mv_price_index': 0.7, 'defender_adjusted_parts': [{'player': 'Adrián de la Fuente', 'market_value': 5200000, 'role': 'RB', 'price_index': 0.7, 'adjusted_quality': 0.4918}], 'xg_override': 0.0392}	aerialWon: 0.0787 -> 0.8 totalShots: 0.5903 -> 0.5556 shotOffTarget: 0.3897 -> 0.3668 onTargetScoringAttempt: 0.2006 -> 0.1888
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 2 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=2`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.76 (squad xG sum 0.9840, xG scale 1.7928, xA scale 1.7928, team xA 1.31 = 0.74 x lambda)	`scoreline_reset=true` `team_lambda=1.764` `squad_xg_sum=0.984` `squad_xa_sum=0.7314` `team_expected_assists=1.3112` `team_xa_to_xg_ratio=0.7433`	expected_goals: 0.0392 -> 0.0703 expected_assists: 0.0437 -> 0.0783
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 8 by projected rating, fifa_overall 79.0)	`match_squad_rank=8` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 2 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=2` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.0679` `goal_rate=0.0492` `scorer_weight=0.0782`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.0753` `assist_rate=0.0605` `provider_weight=0.0923`	-

Javier Rueda M 76 0 0.15 0.03 450% 1 0.3 6.8 52.0 36 XGBoost M_service

Step-by-step (9 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost M_service raw output over 8 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 0.9990, quality 0.477)	`market_value_eur=6200000` `quality_factor=0.4769` `boost_factor=0.999` `reference_quality=0.505` `history_matches=37` `history_shrink=0.119`	-
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` duel_context={'opponent_faced': 'Manuel Sánchez', 'matchup_zone': 'Right Wingback vs Left Back — Overlapping Run', 'duel_win_rate': 0.386, 'pressing_faced': 'LOW', 'numerical_situation': '1v1', 'xg_multiplier': 0.9801, 'skill_edge': -0.0015, 'height_edge': -0.012, 'attacker_skill': 0.5089, 'defender_skill': 0.5123, 'attacker_role': 'RWB', 'attacker_mv_price_index': 0.7, 'defender_adjusted_parts': [{'player': 'Manuel Sánchez', 'market_value': 6400000, 'role': 'LB', 'price_index': 0.7, 'adjusted_quality': 0.5123}], 'xg_override': 0.0849}	aerialWon: 0.6499 -> 0.9 totalShots: 0.93 -> 0.9115 shotOffTarget: 0.6663 -> 0.653 onTargetScoringAttempt: 0.2636 -> 0.2584
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 2 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=2`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.76 (squad xG sum 0.9840, xG scale 1.7928, xA scale 1.7928, team xA 1.31 = 0.74 x lambda)	`scoreline_reset=true` `team_lambda=1.764` `squad_xg_sum=0.984` `squad_xa_sum=0.7314` `team_expected_assists=1.3112` `team_xa_to_xg_ratio=0.7433`	expected_goals: 0.0849 -> 0.1522 expected_assists: 0.111 -> 0.199
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 9 by projected rating, fifa_overall 81.0)	`match_squad_rank=9` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv_filled` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 2 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=2` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.1412` `goal_rate=0.0738` `scorer_weight=0.1397`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.1805` `assist_rate=0.1096` `provider_weight=0.1973`	-

Hugo Álvarez F 73 0 0.25 0.09 193% 1 0.4 6.8 53.2 42 FBref Prior F_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost F_service raw output over 7 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 0.9997, quality 0.496)	`market_value_eur=7800000` `quality_factor=0.4963` `boost_factor=0.9997` `reference_quality=0.505` `history_matches=37` `history_shrink=0.119`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0067)	`blend_weight=0.0067` `history_matches=37` `season_overlap_discount=1.0` `per90_to_match_factor=0.8111` `prior_reliability=0.0867` `competition=La Liga`	totalCross: 1.2226 -> 1.2395 totalShots: 1.256 -> 1.2476
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` duel_context={'opponent_faced': 'Aïssa Mandi', 'matchup_zone': 'Shadow Striker vs Centre Back — Half-Space Threat', 'duel_win_rate': 0.436, 'pressing_faced': 'LOW', 'numerical_situation': '1v1', 'xg_multiplier': 0.9801, 'skill_edge': 0.0265, 'height_edge': -0.04, 'attacker_skill': 0.4963, 'defender_skill': 0.4374, 'attacker_role': 'SS', 'attacker_mv_price_index': 1.0, 'defender_adjusted_parts': [{'player': 'Aïssa Mandi', 'market_value': 1100000, 'role': 'CB', 'price_index': 0.75, 'adjusted_quality': 0.4374}], 'xg_override': 0.1397}	totalShots: 1.2527 -> 1.2278 shotOffTarget: 0.7641 -> 0.7489 onTargetScoringAttempt: 0.4886 -> 0.4789
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 2 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=2`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.76 (squad xG sum 0.9840, xG scale 1.7928, xA scale 1.7928, team xA 1.31 = 0.74 x lambda)	`scoreline_reset=true` `team_lambda=1.764` `squad_xg_sum=0.984` `squad_xa_sum=0.7314` `team_expected_assists=1.3112` `team_xa_to_xg_ratio=0.7433`	expected_goals: 0.1397 -> 0.2504 expected_assists: 0.0925 -> 0.1658
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 11 by projected rating, fifa_overall 75.0)	`match_squad_rank=11` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 2 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=2` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.2215` `goal_rate=0.1676` `scorer_weight=0.2872`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.1528` `assist_rate=0.2173` `provider_weight=0.2072`	-

Hugo González F 61 1 0.40 0.07 457% 2 0.6 6.8 24.0 20 FBref Prior F_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost F_service raw output over 7 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 0.9917, quality 0.439)	`market_value_eur=3100000` `quality_factor=0.4387` `boost_factor=0.9917` `reference_quality=0.505` `history_matches=7` `history_shrink=0.4167`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0090)	`blend_weight=0.009` `history_matches=7` `season_overlap_discount=1.0` `per90_to_match_factor=0.6761` `prior_reliability=0.0333` `competition=La Liga`	totalShots: 1.9412 -> 1.9237 fouls: 0.3773 -> 0.3861 wasFouled: 0.48 -> 0.4879 onTargetScoringAttempt: 0.7047 -> 0.6983
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` duel_context={'opponent_faced': 'Aïssa Mandi, Jeremy Toljan', 'matchup_zone': 'Striker vs Two Centre Backs — 1v2 Aerial/Physical Battle', 'duel_win_rate': 0.189, 'pressing_faced': 'LOW', 'numerical_situation': '1v2', 'xg_multiplier': 0.9832, 'skill_edge': 0.0006, 'height_edge': -0.012, 'attacker_skill': 0.4387, 'defender_skill': 0.4374, 'attacker_role': 'ST', 'attacker_mv_price_index': 1.0, 'defender_adjusted_parts': [{'player': 'Aïssa Mandi', 'market_value': 1100000, 'role': 'CB', 'price_index': 0.75, 'adjusted_quality': 0.4374}, {'player': 'Jeremy Toljan', 'market_value': 1900000, 'role': 'CB', 'price_index': 0.75, 'adjusted_quality': 0.4374}], 'xg_override': 0.2207}	totalShots: 1.9348 -> 1.9023 shotOffTarget: 1.2365 -> 1.2157 onTargetScoringAttempt: 0.6983 -> 0.6866 expectedGoalsOnTarget: 0.3637 -> 0.3576
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 2 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=2`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.76 (squad xG sum 0.9840, xG scale 1.7928, xA scale 1.7928, team xA 1.31 = 0.74 x lambda)	`scoreline_reset=true` `team_lambda=1.764` `squad_xg_sum=0.984` `squad_xa_sum=0.7314` `team_expected_assists=1.3112` `team_xa_to_xg_ratio=0.7433`	expected_goals: 0.2207 -> 0.3957 expected_assists: 0.1001 -> 0.1795
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 10 by projected rating, fifa_overall 70.0)	`match_squad_rank=10` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 2 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=2` `allocation_applied=true` `goals_allocated=1` `goal_probability=0.3268` `goal_rate=0.2303` `scorer_weight=0.4097`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=1` `assist_probability=0.1643` `assist_rate=0.2689` `provider_weight=0.2244`	-

Ferrán Jutglà F 73 1 0.50 0.18 174% 2 1.2 6.7 60.1 31 FBref Prior F_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost F_service raw output over 7 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0008, quality 0.531)	`market_value_eur=10700000` `quality_factor=0.5306` `boost_factor=1.0008` `reference_quality=0.505` `history_matches=41` `history_shrink=0.1087`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0122)	`blend_weight=0.0122` `history_matches=41` `season_overlap_discount=1.0` `per90_to_match_factor=0.8148` `prior_reliability=0.1733` `competition=La Liga`	totalShots: 1.8121 -> 1.7938 onTargetScoringAttempt: 0.6918 -> 0.6833 wasFouled: 1.5249 -> 1.5331
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` duel_context={'opponent_faced': 'Aïssa Mandi', 'matchup_zone': 'Shadow Striker vs Centre Back — Half-Space Threat', 'duel_win_rate': 0.448, 'pressing_faced': 'LOW', 'numerical_situation': '1v1', 'xg_multiplier': 0.997, 'skill_edge': 0.0419, 'height_edge': -0.044, 'attacker_skill': 0.5306, 'defender_skill': 0.4374, 'attacker_role': 'SS', 'attacker_mv_price_index': 1.0, 'defender_adjusted_parts': [{'player': 'Aïssa Mandi', 'market_value': 1100000, 'role': 'CB', 'price_index': 0.75, 'adjusted_quality': 0.4374}], 'xg_override': 0.2787}	totalShots: 1.8036 -> 1.7982
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 2 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=2`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.76 (squad xG sum 0.9840, xG scale 1.7928, xA scale 1.7928, team xA 1.31 = 0.74 x lambda)	`scoreline_reset=true` `team_lambda=1.764` `squad_xg_sum=0.984` `squad_xa_sum=0.7314` `team_expected_assists=1.3112` `team_xa_to_xg_ratio=0.7433`	expected_goals: 0.2787 -> 0.4995 expected_assists: 0.046 -> 0.0825
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 14 by projected rating, fifa_overall 75.0)	`match_squad_rank=14` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 2 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=2` `allocation_applied=true` `goals_allocated=1` `goal_probability=0.3932` `goal_rate=0.2701` `scorer_weight=0.4864`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.0792` `assist_rate=0.0524` `provider_weight=0.088`	-

Team totals 2 1.76 12 6.9

Bench Candidates (7 modeled substitutes with empirical playing time)

Player	Pos	Exp Min	App Prob	Substitute Modeling Details
Altay Bayındır	G	4.4m	11.0%	Market-value substitute prior
Abdoulaye Faye	D	13.6m	21.8%	Market-value substitute prior
Sebastián Cáceres	D	13.6m	30.1%	Market-value substitute prior
Sergio Carreira	M	24.4m	41.1%	Market-value substitute prior
Andrés Antañón	M	22.0m	30.7%	Market-value substitute prior
Williot Swedberg	F	30.4m	44.5%	Market-value substitute prior

Levante — projected line vs season baseline (4-4-2)

Player Pos Min Proj G Proj xG Season xG/g Proj Sh Season Sh/g Proj Rtg Season Rtg Rec Model / FBref Prior

Mathew Ryan G 86 0 0.00 0.00 100% 0.1 6.9 67.8 35 XGBoost G_service

Step-by-step (9 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost G_service raw output over 3 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 0.9981, quality 0.437)	`market_value_eur=1700000` `quality_factor=0.4374` `boost_factor=0.9981` `reference_quality=0.505` `history_matches=47` `history_shrink=0.0962`	-
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` `duel_context={'opponent_faced': 'Ferrán Jutglà, Hugo González', 'matchup_zone': 'Goalkeeper shot-stopping context', 'opponent_lambda': 1.3662}`	-
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 1 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=1`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.22 (squad xG sum 1.1283, xG scale 1.0813, xA scale 1.1158, team xA 0.79 = 0.65 x lambda)	`scoreline_reset=true` `team_lambda=1.22` `squad_xg_sum=1.1283` `squad_xa_sum=0.7107` `team_expected_assists=0.793` `team_xa_to_xg_ratio=0.65`	shots_faced: 13.0 -> 16.8 expected_saves: 3.18 -> 4.12 high_pressure_saves: 1.09 -> 1.41
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 4 by projected rating, fifa_overall 79.0)	`match_squad_rank=4` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 1 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=1` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.0` `scorer_weight=0.0` `squad_weight_total=1.2934`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.0` `provider_weight=0.0` `squad_weight_total=0.798`	-

Manuel Sánchez D 84 0 0.01 0.01 ok 0 0.2 6.8 54.9 34 XGBoost D_service

Step-by-step (9 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost D_service raw output over 7 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 0.9992, quality 0.479)	`market_value_eur=6400000` `quality_factor=0.4793` `boost_factor=0.9992` `reference_quality=0.505` `history_matches=43` `history_shrink=0.1042`	-
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` `duel_context={'opponent_faced': 'Javier Rueda', 'matchup_zone': 'Right Wingback vs Left Back — Overlapping Run'}`	-
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 1 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=1`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.22 (squad xG sum 1.1283, xG scale 1.0813, xA scale 1.1158, team xA 0.79 = 0.65 x lambda)	`scoreline_reset=true` `team_lambda=1.22` `squad_xg_sum=1.1283` `squad_xa_sum=0.7107` `team_expected_assists=0.793` `team_xa_to_xg_ratio=0.65`	expected_assists: 0.0501 -> 0.0559
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 12 by projected rating, fifa_overall 78.0)	`match_squad_rank=12` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv_filled` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 1 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=1` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.0084` `goal_rate=0.0015` `scorer_weight=0.0075`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.0544` `assist_rate=0.0154` `provider_weight=0.0447`	-

Aïssa Mandi D 87 0 0.00 0.00 ok 0 0.1 6.8 47.4 13 XGBoost D_service

Step-by-step (9 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost D_service raw output over 7 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 0.9928, quality 0.437)	`market_value_eur=1100000` `quality_factor=0.4374` `boost_factor=0.9928` `reference_quality=0.505` `history_matches=9` `history_shrink=0.3571`	-
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` `duel_context={'opponent_faced': 'Hugo Álvarez', 'matchup_zone': 'Shadow Striker vs Centre Back — Half-Space Threat'}`	-
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 1 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=1`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.22 (squad xG sum 1.1283, xG scale 1.0813, xA scale 1.1158, team xA 0.79 = 0.65 x lambda)	`scoreline_reset=true` `team_lambda=1.22` `squad_xg_sum=1.1283` `squad_xa_sum=0.7107` `team_expected_assists=0.793` `team_xa_to_xg_ratio=0.65`	-
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 6 by projected rating, fifa_overall 79.0)	`match_squad_rank=6` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 1 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=1` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.0027` `goal_rate=0.0` `scorer_weight=0.0026`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.0133` `assist_rate=0.0` `provider_weight=0.0108`	-

Adrián de la Fuente D 82 0 0.14 0.13 ok 2 1.5 6.9 66.9 31 XGBoost D_service

Step-by-step (9 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost D_service raw output over 7 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 0.9987, quality 0.465)	`market_value_eur=5200000` `quality_factor=0.4646` `boost_factor=0.9987` `reference_quality=0.505` `history_matches=42` `history_shrink=0.1064`	-
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` `duel_context={'opponent_faced': 'Aleix Febas', 'matchup_zone': 'Left Wingback vs Right Back — Overlapping Run'}`	-
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 1 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=1`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.22 (squad xG sum 1.1283, xG scale 1.0813, xA scale 1.1158, team xA 0.79 = 0.65 x lambda)	`scoreline_reset=true` `team_lambda=1.22` `squad_xg_sum=1.1283` `squad_xa_sum=0.7107` `team_expected_assists=0.793` `team_xa_to_xg_ratio=0.65`	expected_goals: 0.1291 -> 0.1396 expected_assists: 0.0945 -> 0.1054
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 1 by projected rating, fifa_overall 86.0)	`match_squad_rank=1` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv_filled` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 1 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=1` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.1303` `goal_rate=0.3312` `scorer_weight=0.1745`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.1` `assist_rate=0.0026` `provider_weight=0.0843`	-

Jeremy Toljan D 83 0 0.03 0.01 139% 0 0.2 6.6 57.5 34 XGBoost D_service

Step-by-step (9 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost D_service raw output over 7 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 0.9980, quality 0.437)	`market_value_eur=1900000` `quality_factor=0.4374` `boost_factor=0.998` `reference_quality=0.505` `history_matches=46` `history_shrink=0.098`	-
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` `duel_context={'opponent_faced': 'Hugo González', 'matchup_zone': 'Striker vs Two Centre Backs — 1v2 Aerial/Physical Battle'}`	-
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 1 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=1`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.22 (squad xG sum 1.1283, xG scale 1.0813, xA scale 1.1158, team xA 0.79 = 0.65 x lambda)	`scoreline_reset=true` `team_lambda=1.22` `squad_xg_sum=1.1283` `squad_xa_sum=0.7107` `team_expected_assists=0.793` `team_xa_to_xg_ratio=0.65`	expected_assists: 0.0962 -> 0.1073
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 17 by projected rating, fifa_overall 74.0)	`match_squad_rank=17` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 1 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=1` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.0313` `goal_rate=0.0002` `scorer_weight=0.0254`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=1` `assist_probability=0.1017` `assist_rate=0.1449` `provider_weight=0.1341`	-

Jon Ander Olasagasti M 86 0 0.08 0.09 ok 1 1.0 6.9 62.8 34 XGBoost M_service

Step-by-step (9 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost M_service raw output over 8 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 0.9979, quality 0.437)	`market_value_eur=2700000` `quality_factor=0.4374` `boost_factor=0.9979` `reference_quality=0.505` `history_matches=43` `history_shrink=0.1042`	-
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` `duel_context={'opponent_faced': 'Yoel Lago', 'matchup_zone': 'LM vs CB', 'duel_win_rate': 0.349, 'pressing_faced': 'LOW', 'numerical_situation': '1v1', 'xg_multiplier': 0.896, 'skill_edge': -0.0347, 'height_edge': -0.036, 'attacker_skill': 0.4374, 'defender_skill': 0.5146, 'attacker_role': 'LM', 'attacker_mv_price_index': 0.9, 'defender_adjusted_parts': [{'player': 'Yoel Lago', 'market_value': 7000000, 'role': 'CB', 'price_index': 0.75, 'adjusted_quality': 0.5146}], 'xg_override': 0.0777}`	aerialWon: 0.5268 -> 0.9 totalShots: 1.1332 -> 1.0154 shotOffTarget: 0.7453 -> 0.6678 onTargetScoringAttempt: 0.3879 -> 0.3476 expectedGoalsOnTarget: 0.1005 -> 0.09 goals: 0.0977 -> 0.0875 expectedGoals: 0.0867 -> 0.0777 bigChanceMissed: 0.0476 -> 0.0426
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 1 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=1`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.22 (squad xG sum 1.1283, xG scale 1.0813, xA scale 1.1158, team xA 0.79 = 0.65 x lambda)	`scoreline_reset=true` `team_lambda=1.22` `squad_xg_sum=1.1283` `squad_xa_sum=0.7107` `team_expected_assists=0.793` `team_xa_to_xg_ratio=0.65`	expected_goals: 0.0777 -> 0.084 expected_assists: 0.1316 -> 0.147
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 2 by projected rating, fifa_overall 84.0)	`match_squad_rank=2` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv_filled` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 1 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=1` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.0806` `goal_rate=0.0875` `scorer_weight=0.0904`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.1367` `assist_rate=0.0995` `provider_weight=0.121`	-

Daniel Requena M 78 0 0.08 0.05 79% 1 0.5 6.8 46.1 13 XGBoost M_service

Step-by-step (9 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost M_service raw output over 8 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 0.9922, quality 0.437)	`market_value_eur=1500000` `quality_factor=0.4374` `boost_factor=0.9922` `reference_quality=0.505` `history_matches=8` `history_shrink=0.3846`	-
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` duel_context={'opponent_faced': 'Miguel Román', 'matchup_zone': 'Central Mid vs Central Mid — Box-to-Box Midfield Contest', 'duel_win_rate': 0.383, 'pressing_faced': 'LOW', 'numerical_situation': '1v1', 'xg_multiplier': 0.9461, 'skill_edge': -0.0686, 'height_edge': 0.032, 'attacker_skill': 0.4374, 'defender_skill': 0.59, 'attacker_role': 'CM', 'attacker_mv_price_index': 0.9, 'defender_adjusted_parts': [{'player': 'Miguel Román', 'market_value': 14400000, 'role': 'CM', 'price_index': 0.9, 'adjusted_quality': 0.59}], 'xg_override': 0.0764}	aerialWon: 0.6252 -> 1.2 totalShots: 1.145 -> 1.0833 shotOffTarget: 0.8356 -> 0.7906 onTargetScoringAttempt: 0.3093 -> 0.2926 bigChanceMissed: 0.2317 -> 0.2192 expectedGoalsOnTarget: 0.226 -> 0.2138
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 1 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=1`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.22 (squad xG sum 1.1283, xG scale 1.0813, xA scale 1.1158, team xA 0.79 = 0.65 x lambda)	`scoreline_reset=true` `team_lambda=1.22` `squad_xg_sum=1.1283` `squad_xa_sum=0.7107` `team_expected_assists=0.793` `team_xa_to_xg_ratio=0.65`	expected_goals: 0.0764 -> 0.0826
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 9 by projected rating, fifa_overall 84.0)	`match_squad_rank=9` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv_filled` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 1 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=1` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.0793` `goal_rate=0.0739` `scorer_weight=0.081`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.0376` `assist_rate=0.0479` `provider_weight=0.0445`	-

Enzo Bardeli M 73 0 0.08 0.02 265% 1 0.3 6.7 46.5 10 XGBoost M_service

Step-by-step (9 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost M_service raw output over 8 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 0.9959, quality 0.472)	`market_value_eur=5800000` `quality_factor=0.472` `boost_factor=0.9959` `reference_quality=0.505` `history_matches=7` `history_shrink=0.4167`	-
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` `duel_context={'opponent_faced': 'Yoel Lago', 'matchup_zone': 'RM vs CB', 'duel_win_rate': 0.352, 'pressing_faced': 'LOW', 'numerical_situation': '1v1', 'xg_multiplier': 0.9006, 'skill_edge': -0.0156, 'height_edge': -0.052, 'attacker_skill': 0.4799, 'defender_skill': 0.5146, 'attacker_role': 'RM', 'attacker_mv_price_index': 0.9, 'defender_adjusted_parts': [{'player': 'Yoel Lago', 'market_value': 7000000, 'role': 'CB', 'price_index': 0.75, 'adjusted_quality': 0.5146}], 'xg_override': 0.0745}`	aerialWon: 0.0 -> 0.8 totalShots: 0.8747 -> 0.7878 shotOffTarget: 0.6243 -> 0.5622 onTargetScoringAttempt: 0.2504 -> 0.2255 expectedGoals: 0.0827 -> 0.0745 goals: 0.0689 -> 0.0621
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 1 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=1`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.22 (squad xG sum 1.1283, xG scale 1.0813, xA scale 1.1158, team xA 0.79 = 0.65 x lambda)	`scoreline_reset=true` `team_lambda=1.22` `squad_xg_sum=1.1283` `squad_xa_sum=0.7107` `team_expected_assists=0.793` `team_xa_to_xg_ratio=0.65`	expected_goals: 0.0745 -> 0.0806 expected_assists: 0.0951 -> 0.1061
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 14 by projected rating, fifa_overall 73.0)	`match_squad_rank=14` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 1 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=1` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.0774` `goal_rate=0.0621` `scorer_weight=0.0726`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.1007` `assist_rate=0.1185` `provider_weight=0.1232`	-

Oriol Rey M 78 0 0.04 0.01 442% 0 0.2 6.9 37.7 32 XGBoost M_service

Step-by-step (9 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost M_service raw output over 8 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 0.9971, quality 0.437)	`market_value_eur=1700000` `quality_factor=0.4374` `boost_factor=0.9971` `reference_quality=0.505` `history_matches=30` `history_shrink=0.1429`	-
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` duel_context={'opponent_faced': 'Miguel Román', 'matchup_zone': 'Central Mid vs Central Mid — Box-to-Box Midfield Contest', 'duel_win_rate': 0.339, 'pressing_faced': 'LOW', 'numerical_situation': '1v1', 'xg_multiplier': 0.8815, 'skill_edge': -0.0686, 'height_edge': -0.012, 'attacker_skill': 0.4374, 'defender_skill': 0.59, 'attacker_role': 'CM', 'attacker_mv_price_index': 0.9, 'defender_adjusted_parts': [{'player': 'Miguel Román', 'market_value': 14400000, 'role': 'CM', 'price_index': 0.9, 'adjusted_quality': 0.59}], 'xg_override': 0.0353}	totalShots: 0.5627 -> 0.496 aerialWon: 0.8459 -> 0.9 shotOffTarget: 0.4057 -> 0.3576 onTargetScoringAttempt: 0.1569 -> 0.1383 goals: 0.0544 -> 0.048
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 1 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=1`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.22 (squad xG sum 1.1283, xG scale 1.0813, xA scale 1.1158, team xA 0.79 = 0.65 x lambda)	`scoreline_reset=true` `team_lambda=1.22` `squad_xg_sum=1.1283` `squad_xa_sum=0.7107` `team_expected_assists=0.793` `team_xa_to_xg_ratio=0.65`	-
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 5 by projected rating, fifa_overall 73.0)	`match_squad_rank=5` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 1 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=1` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.0375` `goal_rate=0.048` `scorer_weight=0.0439`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.0397` `assist_rate=0.0518` `provider_weight=0.0478`	-

Roger Brugué F 75 1 0.31 0.33 ok 2 1.9 6.8 52.9 25 XGBoost F_service

Step-by-step (9 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost F_service raw output over 7 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 0.9970, quality 0.437)	`market_value_eur=2800000` `quality_factor=0.4374` `boost_factor=0.997` `reference_quality=0.505` `history_matches=29` `history_shrink=0.1471`	-
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` duel_context={'opponent_faced': 'Yoel Lago, Álvaro Núñez', 'matchup_zone': 'Striker vs Two Centre Backs — 1v2 Aerial/Physical Battle', 'duel_win_rate': 0.18, 'pressing_faced': 'LOW', 'numerical_situation': '1v2', 'xg_multiplier': 0.9273, 'skill_edge': -0.0235, 'height_edge': -0.026, 'attacker_skill': 0.4374, 'defender_skill': 0.4896, 'attacker_role': 'ST', 'attacker_mv_price_index': 1.0, 'defender_adjusted_parts': [{'player': 'Yoel Lago', 'market_value': 7000000, 'role': 'CB', 'price_index': 0.75, 'adjusted_quality': 0.5146}, {'player': 'Álvaro Núñez', 'market_value': 3900000, 'role': 'CB', 'price_index': 0.75, 'adjusted_quality': 0.4646}], 'xg_override': 0.2865}	totalShots: 2.2304 -> 2.0682 shotOffTarget: 1.3327 -> 1.2358 onTargetScoringAttempt: 0.8977 -> 0.8324 expectedGoalsOnTarget: 0.353 -> 0.3273 goals: 0.3209 -> 0.2976 expectedGoals: 0.309 -> 0.2865 bigChanceMissed: 0.1675 -> 0.1553
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 1 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=1`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.22 (squad xG sum 1.1283, xG scale 1.0813, xA scale 1.1158, team xA 0.79 = 0.65 x lambda)	`scoreline_reset=true` `team_lambda=1.22` `squad_xg_sum=1.1283` `squad_xa_sum=0.7107` `team_expected_assists=0.793` `team_xa_to_xg_ratio=0.65`	expected_goals: 0.2865 -> 0.3098 expected_assists: 0.0546 -> 0.0609
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 8 by projected rating, fifa_overall 88.0)	`match_squad_rank=8` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv_filled` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 1 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=1` `allocation_applied=true` `goals_allocated=1` `goal_probability=0.2664` `goal_rate=0.2976` `scorer_weight=0.32`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.0591` `assist_rate=0.1049` `provider_weight=0.0761`	-

Iván Romero F 73 0 0.26 0.23 ok 2 1.6 6.8 61.1 34 XGBoost F_service

Step-by-step (9 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost F_service raw output over 7 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 0.9996, quality 0.491)	`market_value_eur=7400000` `quality_factor=0.4915` `boost_factor=0.9996` `reference_quality=0.505` `history_matches=43` `history_shrink=0.1042`	-
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` duel_context={'opponent_faced': 'Yoel Lago, Álvaro Núñez', 'matchup_zone': 'Striker vs Two Centre Backs — 1v2 Aerial/Physical Battle', 'duel_win_rate': 0.18, 'pressing_faced': 'LOW', 'numerical_situation': '1v2', 'xg_multiplier': 0.9571, 'skill_edge': 0.0008, 'height_edge': -0.03, 'attacker_skill': 0.4915, 'defender_skill': 0.4896, 'attacker_role': 'ST', 'attacker_mv_price_index': 1.0, 'defender_adjusted_parts': [{'player': 'Yoel Lago', 'market_value': 7000000, 'role': 'CB', 'price_index': 0.75, 'adjusted_quality': 0.5146}, {'player': 'Álvaro Núñez', 'market_value': 3900000, 'role': 'CB', 'price_index': 0.75, 'adjusted_quality': 0.4646}], 'xg_override': 0.2363}	totalShots: 1.9325 -> 1.8496 shotOffTarget: 1.1748 -> 1.1244 onTargetScoringAttempt: 0.7577 -> 0.7252 goals: 0.2916 -> 0.2791 expectedGoals: 0.2469 -> 0.2363 bigChanceMissed: 0.24 -> 0.2297 expectedGoalsOnTarget: 0.2253 -> 0.2156
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 1 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=1`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.22 (squad xG sum 1.1283, xG scale 1.0813, xA scale 1.1158, team xA 0.79 = 0.65 x lambda)	`scoreline_reset=true` `team_lambda=1.22` `squad_xg_sum=1.1283` `squad_xa_sum=0.7107` `team_expected_assists=0.793` `team_xa_to_xg_ratio=0.65`	expected_goals: 0.2363 -> 0.2555 expected_assists: 0.0624 -> 0.0696
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 7 by projected rating, fifa_overall 75.0)	`match_squad_rank=7` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 1 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=1` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.2255` `goal_rate=0.2791` `scorer_weight=0.2937`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.0672` `assist_rate=0.0028` `provider_weight=0.0557`	-

Team totals 1 1.22 11 6.8

Bench Candidates (7 modeled substitutes with empirical playing time)

Player	Pos	Exp Min	App Prob	Substitute Modeling Details
Pablo Cuñat	G	4.4m	10.9%	Market-value substitute prior
Axel Tape	D	13.8m	31.1%	Market-value substitute prior
Ifeanyi Ndukwe	D	13.6m	29.9%	Market-value substitute prior
Hugo Sotelo	M	22.1m	50.3%	Market-value substitute prior
Manel Usedo Domingo	M	22.0m	49.9%	Market-value substitute prior
Etta Eyong	F	25.6m	57.3%	Market-value substitute prior

Antagonist / tactical summary

Midfield is evenly contested — Miguel Román vs Enzo Bardeli will be the pivotal duel. Total antagonist xG — Home: 0.87, Away: 0.79. Home danger zone: Shadow Striker vs Centre Back — Half-Space Threat. Away danger zone: Striker vs Two Centre Backs — 1v2 Aerial/Physical Battle.

Midfield — possession 50.8% / 49.2%, dominance ratio 0.54, key duel: Miguel Román vs Enzo Bardeli

Aleix Febas xG 0.04Miguel Román xG 0.06Ilaix Moriba xG 0.04Jon Ander Olasagasti xG 0.08Daniel Requena xG 0.08Oriol Rey xG 0.04

Summary

Antagonist matchup analysis for Celta vs Levante. Most likely scoreline: 2-1. Lambdas derived from 7 home and 6 away individual player duels. Head-to-head: Celta 4W-2D-4L vs Levante over 10 meetings.

3-4-3 creates structural superiority against 4-4-2. Home team should exploit wider spaces and numerical advantages in transition. Midfield is evenly contested — Miguel Román vs Enzo Bardeli will be the pivotal duel. Total antagonist xG — Home: 0.87, Away: 0.79. Home danger zone: Shadow Striker vs Centre Back — Half-Space Threat. Away danger zone: Striker vs Two Centre Backs — 1v2 Aerial/Physical Battle.

Crystal Palace vs Liverpool

Premier League · 20261108 · 3-4-2-1 vs 4-2-3-1

Predicted 1-1Winner: DrawWinning H 1-0 (8.3%)Winning A 0-1 (8.7%)Confidence 0.65λ home/away 1.36 / 1.42Most-likely P 12.1%

Actual vs predicted: result not available yet (upcoming fixture).

XGBoost Reasoning & FBref Dataset Tracing

Antagonist + XGBoost match analysis for Crystal Palace vs Liverpool. Player statistics regressed using position-specific XGBoost models (G, D, M, F) with 2026-2027 FBref priors. Micro-duels resolved across 3-4-2-1 vs 4-2-3-1 (duel lambdas: Home 1.209, Away 1.646). Final calibrated Poisson lambdas: Home 1.359, Away 1.422 (most likely score 1-1).

Home Starters 11/11 XGBoost (11 FBref) Away Starters 11/11 XGBoost (0 FBref) Bench Modeled H: 7 / A: 7 Duel λ (H/A) 1.21 / 1.65 Final λ (H/A) 1.36 / 1.42

Modeling Pipeline Stages (5 layers)

1. XGBoost player regressions for volume & event rates with market-value quality scaling
2. Multi-source prior blending (FBref 2026-2027 tables + Sofascore season history)
3. Antagonist formation duel matrix resolving 1v1 and zonal matchups
4. Team-level prior calibration (FBref possession and attack/defense efficiencies)
5. Bivariate Poisson matrix with Dixon-Coles low-score correction

Standing & form context

Metric	Crystal Palace	Liverpool
League position	18.00	5.00
Points per game	1.01	1.65
Attack index	0.93	1.00
Defense (concede) index	0.89	1.37
Form multiplier	0.97	1.03
Standing λ multiplier	0.92	1.14

Lambda decomposition (how the goal rates were built)

Stage	Crystal Palace λ	Liverpool λ
Pre-competitive-floor	1.461	1.667
Pre-calibration (Micro-duels)	1.209	1.646
Final λ	1.359	1.422

competition calibration scale 0.889competition observed total goals 2.781quality gap applied Truemismatch score pressure applied Falsecompetitive underdog floor applied Falsestage discount 1.0

Scoreline distribution & markets

Top 5 scorelines	P
1-1	12.1%
0-1	8.7%
1-2	8.5%
1-0	8.3%
2-1	8.1%

Home win	35.8%
Draw	25.6%
Away win	38.6%
Over 2.5	52.6%
BTTS yes	56.5%

Pregame fair line — favorite Liverpool (38.6%), fair total 2.78, fair spread -0.06

Projected players vs season baseline

Proj = projected for this match · Season = this season's per-game average from match data · deviation badge compares projected xG to the player's season xG/game.

Crystal Palace — projected line vs season baseline (3-4-2-1)

Player Pos Min Proj G Proj xG Season xG/g Proj Sh Season Sh/g Proj Rtg Season Rtg Rec Model / FBref Prior

Dean Henderson G 90 0 0.00 - n/a 0.0 6.8 56.6 28 FBref Prior G_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost G_service raw output over 3 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0067, quality 0.710)	`market_value_eur=29000000` `quality_factor=0.71` `boost_factor=1.0067` `reference_quality=0.505` `history_matches=41` `history_shrink=0.1087`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0094)	`blend_weight=0.0094` `history_matches=41` `season_overlap_discount=1.0` `per90_to_match_factor=1.0` `prior_reliability=0.1333` `competition=Premier League`	-
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` `duel_context={'opponent_faced': 'Alexander Isak, Dominik Szoboszlai', 'matchup_zone': 'Goalkeeper shot-stopping context', 'opponent_lambda': 1.646}`	-
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 1 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=1`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.36 (squad xG sum 1.1097, xG scale 1.2247, xA scale 1.3679, team xA 0.88 = 0.65 x lambda)	`scoreline_reset=true` `team_lambda=1.359` `squad_xg_sum=1.1097` `squad_xa_sum=0.6458` `team_expected_assists=0.8833` `team_xa_to_xg_ratio=0.65`	shots_faced: 15.6 -> 13.5 expected_saves: 3.81 -> 3.3 high_pressure_saves: 1.32 -> 1.14
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 5 by projected rating, fifa_overall 82.0)	`match_squad_rank=5` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 1 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=1` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.0` `scorer_weight=0.0` `squad_weight_total=1.3646`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.0` `provider_weight=0.0` `squad_weight_total=0.8034`	-

Jaydee Canvot D 90 0 0.03 0.05 34% 0 0.2 6.7 48.2 32 FBref Prior D_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost D_service raw output over 7 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0103, quality 0.732)	`market_value_eur=32000000` `quality_factor=0.732` `boost_factor=1.0103` `reference_quality=0.505` `history_matches=28` `history_shrink=0.1515`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0131)	`blend_weight=0.0131` `history_matches=28` `season_overlap_discount=1.0` `per90_to_match_factor=0.9969` `prior_reliability=0.1333` `competition=Premier League`	wonTackle: 1.069 -> 1.0811 fouls: 0.8068 -> 0.7962 interceptionWon: 1.2283 -> 1.2187
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` `duel_context={'opponent_faced': 'Dominik Szoboszlai', 'matchup_zone': 'LW vs CB'}`	-
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 1 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=1`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.36 (squad xG sum 1.1097, xG scale 1.2247, xA scale 1.3679, team xA 0.88 = 0.65 x lambda)	`scoreline_reset=true` `team_lambda=1.359` `squad_xg_sum=1.1097` `squad_xa_sum=0.6458` `team_expected_assists=0.8833` `team_xa_to_xg_ratio=0.65`	expected_goals: 0.0252 -> 0.0309 expected_assists: 0.0203 -> 0.0278
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 12 by projected rating, fifa_overall 76.0)	`match_squad_rank=12` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 1 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=1` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.0304` `goal_rate=0.0` `scorer_weight=0.0247`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.0274` `assist_rate=0.0` `provider_weight=0.0222`	-

Chris Richards D 90 0 0.06 0.07 ok 1 0.5 6.8 54.7 31 FBref Prior D_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost D_service raw output over 7 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0055, quality 0.677)	`market_value_eur=25000000` `quality_factor=0.6773` `boost_factor=1.0055` `reference_quality=0.505` `history_matches=42` `history_shrink=0.1064`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0092)	`blend_weight=0.0092` `history_matches=42` `season_overlap_discount=1.0` `per90_to_match_factor=0.9999` `prior_reliability=0.1333` `competition=Premier League`	interceptionWon: 1.1996 -> 1.207 fouls: 0.5849 -> 0.5795
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` `duel_context={'opponent_faced': 'Alexander Isak', 'matchup_zone': 'Striker vs Two Centre Backs — 1v2 Aerial/Physical Battle'}`	-
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 1 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=1`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.36 (squad xG sum 1.1097, xG scale 1.2247, xA scale 1.3679, team xA 0.88 = 0.65 x lambda)	`scoreline_reset=true` `team_lambda=1.359` `squad_xg_sum=1.1097` `squad_xa_sum=0.6458` `team_expected_assists=0.8833` `team_xa_to_xg_ratio=0.65`	expected_goals: 0.0526 -> 0.0644 expected_assists: 0.0162 -> 0.0222
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 7 by projected rating, fifa_overall 80.0)	`match_squad_rank=7` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 1 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=1` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.0624` `goal_rate=0.0055` `scorer_weight=0.0515`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.022` `assist_rate=0.0` `provider_weight=0.0178`	-

Axel Disasi D 84 0 0.07 0.06 35% 0 0.4 6.7 53.1 24 FBref Prior D_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost D_service raw output over 7 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0050, quality 0.585)	`market_value_eur=15500000` `quality_factor=0.5846` `boost_factor=1.005` `reference_quality=0.505` `history_matches=19` `history_shrink=0.2083`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0018)	`blend_weight=0.0018` `history_matches=19` `season_overlap_discount=1.0` `per90_to_match_factor=0.9329` `prior_reliability=0.0133` `competition=Premier League`	fouls: 0.7435 -> 0.7506
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` `duel_context={'opponent_faced': 'Alexander Isak', 'matchup_zone': 'Shape fallback'}`	-
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 1 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=1`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.36 (squad xG sum 1.1097, xG scale 1.2247, xA scale 1.3679, team xA 0.88 = 0.65 x lambda)	`scoreline_reset=true` `team_lambda=1.359` `squad_xg_sum=1.1097` `squad_xa_sum=0.6458` `team_expected_assists=0.8833` `team_xa_to_xg_ratio=0.65`	expected_goals: 0.0612 -> 0.075
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 10 by projected rating, fifa_overall 77.0)	`match_squad_rank=10` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 1 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=1` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.0723` `goal_rate=0.0` `scorer_weight=0.06`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.0098` `assist_rate=0.0` `provider_weight=0.0086`	-

Daichi Kamada M 85 0 0.08 0.07 ok 1 0.5 6.9 57.4 31 FBref Prior M_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost M_service raw output over 8 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0007, quality 0.526)	`market_value_eur=10300000` `quality_factor=0.526` `boost_factor=1.0007` `reference_quality=0.505` `history_matches=37` `history_shrink=0.119`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0103)	`blend_weight=0.0103` `history_matches=37` `season_overlap_discount=1.0` `per90_to_match_factor=0.9441` `prior_reliability=0.1333` `competition=Premier League`	wonTackle: 1.2953 -> 1.316 totalCross: 0.699 -> 0.7113 interceptionWon: 0.3306 -> 0.3369
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` duel_context={'opponent_faced': 'Miloš Kerkez', 'matchup_zone': 'Left Wingback vs Right Back — Overlapping Run', 'duel_win_rate': 0.302, 'pressing_faced': 'HIGH', 'numerical_situation': '1v1', 'xg_multiplier': 0.8562, 'skill_edge': -0.1138, 'height_edge': 0.016, 'attacker_skill': 0.5761, 'defender_skill': 0.829, 'attacker_role': 'LWB', 'attacker_mv_price_index': 0.7, 'defender_adjusted_parts': [{'player': 'Miloš Kerkez', 'market_value': 36000000, 'role': 'RB', 'price_index': 0.7, 'adjusted_quality': 0.829}], 'xg_override': 0.0658}	aerialWon: 0.2922 -> 1.1 totalShots: 0.9386 -> 0.8036 shotOffTarget: 0.5749 -> 0.4922 onTargetScoringAttempt: 0.3637 -> 0.3114 bigChanceMissed: 0.0915 -> 0.0783 expectedGoalsOnTarget: 0.0898 -> 0.0769 goals: 0.0888 -> 0.076 expectedGoals: 0.0769 -> 0.0658
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 1 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=1`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.36 (squad xG sum 1.1097, xG scale 1.2247, xA scale 1.3679, team xA 0.88 = 0.65 x lambda)	`scoreline_reset=true` `team_lambda=1.359` `squad_xg_sum=1.1097` `squad_xa_sum=0.6458` `team_expected_assists=0.8833` `team_xa_to_xg_ratio=0.65`	expected_goals: 0.0658 -> 0.0806 expected_assists: 0.0704 -> 0.0963
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 3 by projected rating, fifa_overall 79.0)	`match_squad_rank=3` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 1 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=1` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.0774` `goal_rate=0.076` `scorer_weight=0.0877`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.0918` `assist_rate=0.0825` `provider_weight=0.106`	-

Adam Wharton M 88 0 0.06 0.03 135% 1 0.7 6.9 58.1 32 FBref Prior M_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost M_service raw output over 8 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0118, quality 0.881)	`market_value_eur=79000000` `quality_factor=0.8811` `boost_factor=1.0118` `reference_quality=0.505` `history_matches=43` `history_shrink=0.1042`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0090)	`blend_weight=0.009` `history_matches=43` `season_overlap_discount=1.0` `per90_to_match_factor=0.9748` `prior_reliability=0.1333` `competition=Premier League`	totalCross: 2.7116 -> 2.6915 wasFouled: 1.0977 -> 1.1054 interceptionWon: 1.1201 -> 1.1276 totalShots: 0.7797 -> 0.7727 fouls: 0.5997 -> 0.5943
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=true`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` duel_context={'opponent_faced': 'Alexis Mac Allister', 'matchup_zone': 'Central Mid vs Defensive Mid — Pressing/Creation Duel', 'duel_win_rate': 0.473, 'pressing_faced': 'ULTRA_HIGH', 'numerical_situation': '1v1', 'xg_multiplier': 1.0342, 'skill_edge': -0.0007, 'height_edge': 0.024, 'attacker_skill': 0.8877, 'defender_skill': 0.8893, 'attacker_role': 'CM', 'attacker_mv_price_index': 0.9, 'defender_adjusted_parts': [{'player': 'Alexis Mac Allister', 'market_value': 77000000, 'role': 'CDM', 'price_index': 0.85, 'adjusted_quality': 0.8893}], 'xg_override': 0.0502}	totalShots: 0.7777 -> 0.8043 shotOffTarget: 0.5654 -> 0.5847 aerialWon: 1.0851 -> 1.1 onTargetScoringAttempt: 0.2123 -> 0.2196
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 1 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=1`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.36 (squad xG sum 1.1097, xG scale 1.2247, xA scale 1.3679, team xA 0.88 = 0.65 x lambda)	`scoreline_reset=true` `team_lambda=1.359` `squad_xg_sum=1.1097` `squad_xa_sum=0.6458` `team_expected_assists=0.8833` `team_xa_to_xg_ratio=0.65`	expected_goals: 0.0502 -> 0.0615 expected_assists: 0.1074 -> 0.1468
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 1 by projected rating, fifa_overall 82.0)	`match_squad_rank=1` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 1 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=1` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.0596` `goal_rate=0.0426` `scorer_weight=0.0568`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=1` `assist_probability=0.1365` `assist_rate=0.0884` `provider_weight=0.1291`	-

Tyrick Mitchell M 87 0 0.08 0.06 40% 1 0.6 6.8 58.8 34 FBref Prior M_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost M_service raw output over 8 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0045, quality 0.660)	`market_value_eur=23000000` `quality_factor=0.6595` `boost_factor=1.0045` `reference_quality=0.505` `history_matches=47` `history_shrink=0.0962`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0083)	`blend_weight=0.0083` `history_matches=47` `season_overlap_discount=1.0` `per90_to_match_factor=0.9677` `prior_reliability=0.1333` `competition=Premier League`	wonTackle: 1.9337 -> 1.9257 wasFouled: 0.7116 -> 0.7057
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` duel_context={'opponent_faced': 'Conor Bradley', 'matchup_zone': 'Right Wingback vs Left Back — Overlapping Run', 'duel_win_rate': 0.376, 'pressing_faced': 'MEDIUM', 'numerical_situation': '1v1', 'xg_multiplier': 0.9654, 'skill_edge': 0.0045, 'height_edge': -0.028, 'attacker_skill': 0.7379, 'defender_skill': 0.728, 'attacker_role': 'RWB', 'attacker_mv_price_index': 0.7, 'defender_adjusted_parts': [{'player': 'Conor Bradley', 'market_value': 22000000, 'role': 'LB', 'price_index': 0.7, 'adjusted_quality': 0.728}], 'xg_override': 0.0658}	aerialWon: 0.7792 -> 0.9 totalShots: 0.8957 -> 0.8647 shotOffTarget: 0.6754 -> 0.652 onTargetScoringAttempt: 0.2178 -> 0.2103
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 1 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=1`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.36 (squad xG sum 1.1097, xG scale 1.2247, xA scale 1.3679, team xA 0.88 = 0.65 x lambda)	`scoreline_reset=true` `team_lambda=1.359` `squad_xg_sum=1.1097` `squad_xa_sum=0.6458` `team_expected_assists=0.8833` `team_xa_to_xg_ratio=0.65`	expected_goals: 0.0658 -> 0.0806 expected_assists: 0.0597 -> 0.0817
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 6 by projected rating, fifa_overall 79.0)	`match_squad_rank=6` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 1 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=1` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.0774` `goal_rate=0.0572` `scorer_weight=0.0746`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.0785` `assist_rate=0.0678` `provider_weight=0.0877`	-

Anan Khalaili M 88 0 0.10 0.08 ok 1 0.7 6.8 43.9 16 FBref Prior M_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost M_service raw output over 8 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0175, quality 0.669)	`market_value_eur=24000000` `quality_factor=0.6685` `boost_factor=1.0175` `reference_quality=0.505` `history_matches=9` `history_shrink=0.3571`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0186)	`blend_weight=0.0186` `history_matches=9` `season_overlap_discount=1.0` `per90_to_match_factor=0.9793` `prior_reliability=0.08` `competition=Premier League`	wonTackle: 1.5706 -> 1.6778 totalCross: 3.2523 -> 3.298 interceptionWon: 1.5339 -> 1.5509 wasFouled: 0.5003 -> 0.5062 fouls: 1.3334 -> 1.3389 onTargetScoringAttempt: 0.2956 -> 0.2901
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` duel_context={'opponent_faced': 'Alexis Mac Allister', 'matchup_zone': 'Central Mid vs Defensive Mid — Pressing/Creation Duel', 'duel_win_rate': 0.389, 'pressing_faced': 'ULTRA_HIGH', 'numerical_situation': '1v1', 'xg_multiplier': 0.9102, 'skill_edge': -0.0891, 'height_edge': 0.028, 'attacker_skill': 0.6914, 'defender_skill': 0.8893, 'attacker_role': 'CM', 'attacker_mv_price_index': 0.9, 'defender_adjusted_parts': [{'player': 'Alexis Mac Allister', 'market_value': 77000000, 'role': 'CDM', 'price_index': 0.85, 'adjusted_quality': 0.8893}], 'xg_override': 0.0805}	aerialWon: 1.2411 -> 1.1 totalShots: 0.9885 -> 0.8997 shotOffTarget: 0.6964 -> 0.6339 onTargetScoringAttempt: 0.2901 -> 0.264 bigChanceMissed: 0.2169 -> 0.1974 goals: 0.1242 -> 0.113 expectedGoals: 0.0884 -> 0.0805
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 1 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=1`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.36 (squad xG sum 1.1097, xG scale 1.2247, xA scale 1.3679, team xA 0.88 = 0.65 x lambda)	`scoreline_reset=true` `team_lambda=1.359` `squad_xg_sum=1.1097` `squad_xa_sum=0.6458` `team_expected_assists=0.8833` `team_xa_to_xg_ratio=0.65`	expected_goals: 0.0805 -> 0.0986 expected_assists: 0.0817 -> 0.1118
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 4 by projected rating, fifa_overall 78.0)	`match_squad_rank=4` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 1 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=1` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.0939` `goal_rate=0.113` `scorer_weight=0.1232`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.1058` `assist_rate=0.0673` `provider_weight=0.0996`	-

Yéremy Pino F 70 0 0.14 0.08 89% 1 0.9 6.9 54.4 35 FBref Prior F_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost F_service raw output over 7 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0057, quality 0.694)	`market_value_eur=27000000` `quality_factor=0.6941` `boost_factor=1.0057` `reference_quality=0.505` `history_matches=45` `history_shrink=0.1`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0048)	`blend_weight=0.0048` `history_matches=45` `season_overlap_discount=1.0` `per90_to_match_factor=0.7821` `prior_reliability=0.0733` `competition=Premier League`	totalCross: 2.6757 -> 2.6934 fouls: 0.5386 -> 0.5462 totalOffside: 0.2506 -> 0.2562 wasFouled: 1.0128 -> 1.0181
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` duel_context={'opponent_faced': 'Alexis Mac Allister, Virgil van Dijk', 'matchup_zone': 'Shadow Striker vs CDM + CB — Zone 14 Battle', 'duel_win_rate': 0.185, 'pressing_faced': 'HIGH', 'numerical_situation': '1v2', 'xg_multiplier': 0.8748, 'skill_edge': -0.0312, 'height_edge': -0.054, 'attacker_skill': 0.6941, 'defender_skill': 0.7634, 'attacker_role': 'SS', 'attacker_mv_price_index': 1.0, 'defender_adjusted_parts': [{'player': 'Alexis Mac Allister', 'market_value': 77000000, 'role': 'CDM', 'price_index': 0.85, 'adjusted_quality': 0.8893}, {'player': 'Virgil van Dijk', 'market_value': 15500000, 'role': 'CB', 'price_index': 0.75, 'adjusted_quality': 0.6376}], 'xg_override': 0.1177}	totalShots: 1.2307 -> 1.0766 shotOffTarget: 0.7651 -> 0.6693 onTargetScoringAttempt: 0.4594 -> 0.4019 goals: 0.1825 -> 0.1597 expectedGoalsOnTarget: 0.1627 -> 0.1423 expectedGoals: 0.1345 -> 0.1177 bigChanceMissed: 0.0687 -> 0.0601
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 1 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=1`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.36 (squad xG sum 1.1097, xG scale 1.2247, xA scale 1.3679, team xA 0.88 = 0.65 x lambda)	`scoreline_reset=true` `team_lambda=1.359` `squad_xg_sum=1.1097` `squad_xa_sum=0.6458` `team_expected_assists=0.8833` `team_xa_to_xg_ratio=0.65`	expected_goals: 0.1177 -> 0.1441 expected_assists: 0.0874 -> 0.1196
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 2 by projected rating, fifa_overall 79.0)	`match_squad_rank=2` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 1 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=1` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.1342` `goal_rate=0.1597` `scorer_weight=0.1801`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.1127` `assist_rate=0.0042` `provider_weight=0.0957`	-

Jørgen Strand Larsen F 73 1 0.41 0.13 208% 2 1.2 6.7 54.9 31 FBref Prior F_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost F_service raw output over 7 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0088, quality 0.786)	`market_value_eur=41000000` `quality_factor=0.7859` `boost_factor=1.0088` `reference_quality=0.505` `history_matches=43` `history_shrink=0.1042`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0045)	`blend_weight=0.0045` `history_matches=43` `season_overlap_discount=1.0` `per90_to_match_factor=0.8119` `prior_reliability=0.0667` `competition=Premier League`	fouls: 1.0912 -> 1.0973
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` duel_context={'opponent_faced': 'Virgil van Dijk, Jérémy Jacquet', 'matchup_zone': 'Striker vs Two Centre Backs — 1v2 Aerial/Physical Battle', 'duel_win_rate': 0.21, 'pressing_faced': 'HIGH', 'numerical_situation': '1v2', 'xg_multiplier': 1.0151, 'skill_edge': 0.0103, 'height_edge': 0.0, 'attacker_skill': 0.7859, 'defender_skill': 0.7631, 'attacker_role': 'ST', 'attacker_mv_price_index': 1.0, 'defender_adjusted_parts': [{'player': 'Virgil van Dijk', 'market_value': 15500000, 'role': 'CB', 'price_index': 0.75, 'adjusted_quality': 0.6376}, {'player': 'Jérémy Jacquet', 'market_value': 67000000, 'role': 'CB', 'price_index': 0.75, 'adjusted_quality': 0.8886}], 'xg_override': 0.3325}	totalShots: 1.788 -> 1.815 shotOffTarget: 1.0725 -> 1.0887 onTargetScoringAttempt: 0.7155 -> 0.7263
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 1 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=1`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.36 (squad xG sum 1.1097, xG scale 1.2247, xA scale 1.3679, team xA 0.88 = 0.65 x lambda)	`scoreline_reset=true` `team_lambda=1.359` `squad_xg_sum=1.1097` `squad_xa_sum=0.6458` `team_expected_assists=0.8833` `team_xa_to_xg_ratio=0.65`	expected_goals: 0.3325 -> 0.4072 expected_assists: 0.0524 -> 0.0717
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 13 by projected rating, fifa_overall 77.0)	`match_squad_rank=13` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 1 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=1` `allocation_applied=true` `goals_allocated=1` `goal_probability=0.3345` `goal_rate=0.2823` `scorer_weight=0.3538`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.0692` `assist_rate=0.0` `provider_weight=0.0574`	-

Dwight McNeil F 74 0 0.16 0.03 376% 1 0.6 6.7 32.5 35 FBref Prior F_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost F_service raw output over 7 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0070, quality 0.660)	`market_value_eur=23000000` `quality_factor=0.6595` `boost_factor=1.007` `reference_quality=0.505` `history_matches=28` `history_shrink=0.1515`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0059)	`blend_weight=0.0059` `history_matches=28` `season_overlap_discount=1.0` `per90_to_match_factor=0.8173` `prior_reliability=0.06` `competition=Premier League`	totalCross: 1.5333 -> 1.5242 wonTackle: 1.1651 -> 1.1582
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` duel_context={'opponent_faced': 'Alexis Mac Allister, Virgil van Dijk', 'matchup_zone': 'Shadow Striker vs CDM + CB — Zone 14 Battle', 'duel_win_rate': 0.213, 'pressing_faced': 'HIGH', 'numerical_situation': '1v2', 'xg_multiplier': 0.9165, 'skill_edge': -0.0468, 'height_edge': -0.01, 'attacker_skill': 0.6595, 'defender_skill': 0.7634, 'attacker_role': 'SS', 'attacker_mv_price_index': 1.0, 'defender_adjusted_parts': [{'player': 'Alexis Mac Allister', 'market_value': 77000000, 'role': 'CDM', 'price_index': 0.85, 'adjusted_quality': 0.8893}, {'player': 'Virgil van Dijk', 'market_value': 15500000, 'role': 'CB', 'price_index': 0.75, 'adjusted_quality': 0.6376}], 'xg_override': 0.1339}	totalShots: 1.2442 -> 1.1403 shotOffTarget: 0.8498 -> 0.7788 onTargetScoringAttempt: 0.3943 -> 0.3614 goals: 0.1522 -> 0.1395 expectedGoals: 0.1461 -> 0.1339
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 1 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=1`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.36 (squad xG sum 1.1097, xG scale 1.2247, xA scale 1.3679, team xA 0.88 = 0.65 x lambda)	`scoreline_reset=true` `team_lambda=1.359` `squad_xg_sum=1.1097` `squad_xa_sum=0.6458` `team_expected_assists=0.8833` `team_xa_to_xg_ratio=0.65`	expected_goals: 0.1339 -> 0.164 expected_assists: 0.0776 -> 0.1061
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 15 by projected rating, fifa_overall 76.0)	`match_squad_rank=15` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 1 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=1` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.1513` `goal_rate=0.1395` `scorer_weight=0.169`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.1007` `assist_rate=0.0165` `provider_weight=0.0849`	-

Team totals 1 1.36 11 6.8

Bench Candidates (7 modeled substitutes with empirical playing time)

Player	Pos	Exp Min	App Prob	Substitute Modeling Details
Walter Benítez	G	4.4m	11.0%	Market-value substitute prior
Honest Ahanor	D	14.9m	36.1%	Market-value substitute prior
Chadi Riad	D	14.1m	22.8%	Market-value substitute prior
Quinten Timber	M	22.9m	54.0%	Market-value substitute prior
Óscar Mingueza	M	18.9m	70.4%	Market-value substitute prior
Ismaïla Sarr	F	26.7m	62.3%	Market-value substitute prior

Liverpool — projected line vs season baseline (4-2-3-1)

Player Pos Min Proj G Proj xG Season xG/g Proj Sh Season Sh/g Proj Rtg Season Rtg Rec Model / FBref Prior

Giorgi Mamardashvili G 86 0 0.00 - n/a 0.0 7.0 11.2 31 XGBoost G_service

Step-by-step (9 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost G_service raw output over 3 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0205, quality 0.710)	`market_value_eur=29000000` `quality_factor=0.71` `boost_factor=1.0205` `reference_quality=0.505` `history_matches=10` `history_shrink=0.3333`	-
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` `duel_context={'opponent_faced': 'Jørgen Strand Larsen, Dwight McNeil', 'matchup_zone': 'Goalkeeper shot-stopping context', 'opponent_lambda': 1.2087}`	-
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 1 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=1`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.42 (squad xG sum 1.4199, xG scale 1.0015, xA scale 1.1218, team xA 0.92 = 0.65 x lambda)	`scoreline_reset=true` `team_lambda=1.422` `squad_xg_sum=1.4199` `squad_xa_sum=0.8239` `team_expected_assists=0.9243` `team_xa_to_xg_ratio=0.65`	shots_faced: 11.5 -> 12.9 expected_saves: 2.82 -> 3.16 high_pressure_saves: 0.97 -> 1.09
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 7 by projected rating, fifa_overall 83.0)	`match_squad_rank=7` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 1 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=1` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.0` `scorer_weight=0.0` `squad_weight_total=1.5473`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.0` `provider_weight=0.0` `squad_weight_total=0.8798`	-

Virgil van Dijk D 90 0 0.06 0.10 34% 1 0.7 7.1 59.5 34 XGBoost D_service

Step-by-step (9 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost D_service raw output over 7 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0023, quality 0.585)	`market_value_eur=15500000` `quality_factor=0.5846` `boost_factor=1.0023` `reference_quality=0.505` `history_matches=46` `history_shrink=0.098`	-
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` `duel_context={'opponent_faced': 'Jørgen Strand Larsen', 'matchup_zone': 'Striker vs Two Centre Backs — 1v2 Aerial/Physical Battle'}`	-
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 1 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=1`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.42 (squad xG sum 1.4199, xG scale 1.0015, xA scale 1.1218, team xA 0.92 = 0.65 x lambda)	`scoreline_reset=true` `team_lambda=1.422` `squad_xg_sum=1.4199` `squad_xa_sum=0.8239` `team_expected_assists=0.9243` `team_xa_to_xg_ratio=0.65`	-
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 1 by projected rating, fifa_overall 88.0)	`match_squad_rank=1` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 1 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=1` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.0619` `goal_rate=0.1149` `scorer_weight=0.0799`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.0254` `assist_rate=0.0` `provider_weight=0.0206`	-

Jérémy Jacquet D 88 0 0.05 0.02 121% 1 0.1 7.0 43.3 14 XGBoost D_service

Step-by-step (9 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost D_service raw output over 7 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0417, quality 0.866)	`market_value_eur=67000000` `quality_factor=0.8661` `boost_factor=1.0417` `reference_quality=0.505` `history_matches=8` `history_shrink=0.3846`	-
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` `duel_context={'opponent_faced': 'Jørgen Strand Larsen', 'matchup_zone': 'Striker vs Two Centre Backs — 1v2 Aerial/Physical Battle'}`	-
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 1 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=1`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.42 (squad xG sum 1.4199, xG scale 1.0015, xA scale 1.1218, team xA 0.92 = 0.65 x lambda)	`scoreline_reset=true` `team_lambda=1.422` `squad_xg_sum=1.4199` `squad_xa_sum=0.8239` `team_expected_assists=0.9243` `team_xa_to_xg_ratio=0.65`	-
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 6 by projected rating, fifa_overall 77.0)	`match_squad_rank=6` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 1 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=1` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.0465` `goal_rate=0.0` `scorer_weight=0.0381`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.0252` `assist_rate=0.0` `provider_weight=0.0204`	-

Conor Bradley D 77 0 0.02 0.02 ok 1 0.4 6.8 24.6 8 XGBoost D_service

Step-by-step (9 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost D_service raw output over 7 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0109, quality 0.650)	`market_value_eur=22000000` `quality_factor=0.6503` `boost_factor=1.0109` `reference_quality=0.505` `history_matches=15` `history_shrink=0.25`	-
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` `duel_context={'opponent_faced': 'Tyrick Mitchell', 'matchup_zone': 'Right Wingback vs Left Back — Overlapping Run'}`	-
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 1 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=1`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.42 (squad xG sum 1.4199, xG scale 1.0015, xA scale 1.1218, team xA 0.92 = 0.65 x lambda)	`scoreline_reset=true` `team_lambda=1.422` `squad_xg_sum=1.4199` `squad_xa_sum=0.8239` `team_expected_assists=0.9243` `team_xa_to_xg_ratio=0.65`	expected_assists: 0.0769 -> 0.0863
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 11 by projected rating, fifa_overall 79.0)	`match_squad_rank=11` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 1 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=1` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.024` `goal_rate=0.0` `scorer_weight=0.0194`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.0827` `assist_rate=0.1502` `provider_weight=0.1079`	-

Miloš Kerkez D 86 0 0.03 0.02 44% 0 0.2 6.7 48.5 34 XGBoost D_service

Step-by-step (9 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost D_service raw output over 7 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0081, quality 0.758)	`market_value_eur=36000000` `quality_factor=0.7581` `boost_factor=1.0081` `reference_quality=0.505` `history_matches=42` `history_shrink=0.1064`	-
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` `duel_context={'opponent_faced': 'Daichi Kamada', 'matchup_zone': 'Left Wingback vs Right Back — Overlapping Run'}`	-
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 1 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=1`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.42 (squad xG sum 1.4199, xG scale 1.0015, xA scale 1.1218, team xA 0.92 = 0.65 x lambda)	`scoreline_reset=true` `team_lambda=1.422` `squad_xg_sum=1.4199` `squad_xa_sum=0.8239` `team_expected_assists=0.9243` `team_xa_to_xg_ratio=0.65`	-
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 16 by projected rating, fifa_overall 81.0)	`match_squad_rank=16` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 1 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=1` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.0263` `goal_rate=0.0381` `scorer_weight=0.0307`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.0199` `assist_rate=0.0156` `provider_weight=0.0194`	-

Dominik Szoboszlai M 80 0 0.19 0.14 34% 2 1.8 7.1 60.0 30 XGBoost M_service

Step-by-step (9 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost M_service raw output over 8 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0123, quality 0.890)	`market_value_eur=91000000` `quality_factor=0.8895` `boost_factor=1.0123` `reference_quality=0.505` `history_matches=42` `history_shrink=0.1064`	-
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` duel_context={'opponent_faced': 'Jaydee Canvot', 'matchup_zone': 'LW vs CB', 'duel_win_rate': 0.459, 'pressing_faced': 'HIGH', 'numerical_situation': '1v1', 'xg_multiplier': 1.0573, 'skill_edge': 0.0429, 'height_edge': -0.004, 'attacker_skill': 0.8895, 'defender_skill': 0.7941, 'attacker_role': 'LW', 'attacker_mv_price_index': 1.0, 'defender_adjusted_parts': [{'player': 'Jaydee Canvot', 'market_value': 32000000, 'role': 'CB', 'price_index': 0.75, 'adjusted_quality': 0.7941}], 'xg_override': 0.1859}	aerialWon: 0.6485 -> 1.2 totalShots: 1.8512 -> 1.9573 shotOffTarget: 1.232 -> 1.3026 onTargetScoringAttempt: 0.6192 -> 0.6547 expectedGoalsOnTarget: 0.2487 -> 0.263 goals: 0.1883 -> 0.1991 expectedGoals: 0.1758 -> 0.1859
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 1 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=1`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.42 (squad xG sum 1.4199, xG scale 1.0015, xA scale 1.1218, team xA 0.92 = 0.65 x lambda)	`scoreline_reset=true` `team_lambda=1.422` `squad_xg_sum=1.4199` `squad_xa_sum=0.8239` `team_expected_assists=0.9243` `team_xa_to_xg_ratio=0.65`	expected_assists: 0.1899 -> 0.213
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 2 by projected rating, fifa_overall 86.0)	`match_squad_rank=2` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 1 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=1` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.1699` `goal_rate=0.1991` `scorer_weight=0.1966`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.1918` `assist_rate=0.1244` `provider_weight=0.1704`	-

Florian Wirtz M 79 0 0.16 0.17 ok 2 1.5 7.0 54.8 30 XGBoost M_service

Step-by-step (9 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost M_service raw output over 8 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0123, quality 0.883)	`market_value_eur=81000000` `quality_factor=0.8829` `boost_factor=1.0123` `reference_quality=0.505` `history_matches=41` `history_shrink=0.1087`	-
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` duel_context={'opponent_faced': 'Adam Wharton', 'matchup_zone': 'CAM vs CM', 'duel_win_rate': 0.391, 'pressing_faced': 'ULTRA_HIGH', 'numerical_situation': '1v1', 'xg_multiplier': 0.9578, 'skill_edge': -0.0007, 'height_edge': -0.028, 'attacker_skill': 0.8861, 'defender_skill': 0.8877, 'attacker_role': 'CAM', 'attacker_mv_price_index': 0.95, 'defender_adjusted_parts': [{'player': 'Adam Wharton', 'market_value': 79000000, 'role': 'CM', 'price_index': 0.9, 'adjusted_quality': 0.8877}], 'xg_override': 0.1585}	aerialWon: 0.3131 -> 0.9 totalShots: 1.7368 -> 1.6635 shotOffTarget: 1.1296 -> 1.0819 onTargetScoringAttempt: 0.6071 -> 0.5815 expectedGoals: 0.1655 -> 0.1585 goals: 0.1634 -> 0.1565
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 1 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=1`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.42 (squad xG sum 1.4199, xG scale 1.0015, xA scale 1.1218, team xA 0.92 = 0.65 x lambda)	`scoreline_reset=true` `team_lambda=1.422` `squad_xg_sum=1.4199` `squad_xa_sum=0.8239` `team_expected_assists=0.9243` `team_xa_to_xg_ratio=0.65`	expected_assists: 0.1688 -> 0.1894
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 5 by projected rating, fifa_overall 86.0)	`match_squad_rank=5` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 1 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=1` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.1467` `goal_rate=0.1565` `scorer_weight=0.1572`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=1` `assist_probability=0.1725` `assist_rate=0.1493` `provider_weight=0.1725`	-

Alexis Mac Allister M 88 0 0.14 0.08 65% 2 1.2 7.1 56.4 34 XGBoost M_service

Step-by-step (9 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost M_service raw output over 8 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0112, quality 0.879)	`market_value_eur=77000000` `quality_factor=0.8792` `boost_factor=1.0112` `reference_quality=0.505` `history_matches=45` `history_shrink=0.1`	-
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` `duel_context={'opponent_faced': 'Yéremy Pino, Dwight McNeil, Adam Wharton', 'matchup_zone': 'Shadow Striker vs CDM + CB — Zone 14 Battle'}`	-
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 1 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=1`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.42 (squad xG sum 1.4199, xG scale 1.0015, xA scale 1.1218, team xA 0.92 = 0.65 x lambda)	`scoreline_reset=true` `team_lambda=1.422` `squad_xg_sum=1.4199` `squad_xa_sum=0.8239` `team_expected_assists=0.9243` `team_xa_to_xg_ratio=0.65`	expected_assists: 0.1153 -> 0.1293
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 3 by projected rating, fifa_overall 84.0)	`match_squad_rank=3` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 1 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=1` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.1285` `goal_rate=0.1418` `scorer_weight=0.1408`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.1213` `assist_rate=0.0876` `provider_weight=0.1076`	-

Ryan Gravenberch M 79 0 0.07 0.03 102% 1 0.8 6.9 56.2 33 XGBoost M_service

Step-by-step (9 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost M_service raw output over 8 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0118, quality 0.883)	`market_value_eur=81000000` `quality_factor=0.8829` `boost_factor=1.0118` `reference_quality=0.505` `history_matches=43` `history_shrink=0.1042`	-
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=true`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` `duel_context={'opponent_faced': 'Daichi Kamada', 'matchup_zone': 'Midfield lane'}`	-
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 1 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=1`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.42 (squad xG sum 1.4199, xG scale 1.0015, xA scale 1.1218, team xA 0.92 = 0.65 x lambda)	`scoreline_reset=true` `team_lambda=1.422` `squad_xg_sum=1.4199` `squad_xa_sum=0.8239` `team_expected_assists=0.9243` `team_xa_to_xg_ratio=0.65`	expected_assists: 0.0445 -> 0.0499
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 10 by projected rating, fifa_overall 85.0)	`match_squad_rank=10` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 1 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=1` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.0659` `goal_rate=0.0742` `scorer_weight=0.0717`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.0487` `assist_rate=0.0757` `provider_weight=0.0624`	-

Wataru Endo M 80 0 0.03 0.00 1355% 0 0.1 6.8 11.4 18 XGBoost M_service

Step-by-step (9 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost M_service raw output over 8 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 0.9922, quality 0.437)	`market_value_eur=2700000` `quality_factor=0.4374` `boost_factor=0.9922` `reference_quality=0.505` `history_matches=8` `history_shrink=0.3846`	-
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=true`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` duel_context={'opponent_faced': 'Jaydee Canvot', 'matchup_zone': 'RW vs CB', 'duel_win_rate': 0.219, 'pressing_faced': 'HIGH', 'numerical_situation': '1v1', 'xg_multiplier': 0.7053, 'skill_edge': -0.1605, 'height_edge': -0.04, 'attacker_skill': 0.4374, 'defender_skill': 0.7941, 'attacker_role': 'RW', 'attacker_mv_price_index': 1.0, 'defender_adjusted_parts': [{'player': 'Jaydee Canvot', 'market_value': 32000000, 'role': 'CB', 'price_index': 0.75, 'adjusted_quality': 0.7941}], 'xg_override': 0.0312}	aerialWon: 1.2863 -> 1.0 totalShots: 0.6426 -> 0.4532 shotOffTarget: 0.491 -> 0.3463 onTargetScoringAttempt: 0.1516 -> 0.1069 goals: 0.0701 -> 0.0494 expectedGoals: 0.0442 -> 0.0312
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 1 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=1`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.42 (squad xG sum 1.4199, xG scale 1.0015, xA scale 1.1218, team xA 0.92 = 0.65 x lambda)	`scoreline_reset=true` `team_lambda=1.422` `squad_xg_sum=1.4199` `squad_xa_sum=0.8239` `team_expected_assists=0.9243` `team_xa_to_xg_ratio=0.65`	-
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 14 by projected rating, fifa_overall 78.0)	`match_squad_rank=14` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 1 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=1` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.0307` `goal_rate=0.0494` `scorer_weight=0.0382`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.0371` `assist_rate=0.0604` `provider_weight=0.0473`	-

Alexander Isak F 79 1 0.50 0.27 87% 3 1.2 6.9 43.4 18 XGBoost F_service

Step-by-step (9 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost F_service raw output over 7 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0222, quality 0.891)	`market_value_eur=93000000` `quality_factor=0.8905` `boost_factor=1.0222` `reference_quality=0.505` `history_matches=21` `history_shrink=0.1923`	-
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` duel_context={'opponent_faced': 'Jaydee Canvot, Chris Richards', 'matchup_zone': 'Striker vs Two Centre Backs — 1v2 Aerial/Physical Battle', 'duel_win_rate': 0.275, 'pressing_faced': 'MEDIUM', 'numerical_situation': '1v2', 'xg_multiplier': 1.1107, 'skill_edge': 0.0553, 'height_edge': 0.02, 'attacker_skill': 0.8905, 'defender_skill': 0.7676, 'attacker_role': 'ST', 'attacker_mv_price_index': 1.0, 'defender_adjusted_parts': [{'player': 'Jaydee Canvot', 'market_value': 32000000, 'role': 'CB', 'price_index': 0.75, 'adjusted_quality': 0.7941}, {'player': 'Chris Richards', 'market_value': 25000000, 'role': 'CB', 'price_index': 0.75, 'adjusted_quality': 0.7411}], 'xg_override': 0.4985}	totalShots: 2.7489 -> 3.0532 shotOffTarget: 1.622 -> 1.8016 onTargetScoringAttempt: 1.1269 -> 1.2516 expectedGoalsOnTarget: 0.6991 -> 0.7765 goals: 0.54 -> 0.5998 expectedGoals: 0.4488 -> 0.4985 bigChanceMissed: 0.3738 -> 0.4152
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 1 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=1`	goals: 1.0 -> 0
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.42 (squad xG sum 1.4199, xG scale 1.0015, xA scale 1.1218, team xA 0.92 = 0.65 x lambda)	`scoreline_reset=true` `team_lambda=1.422` `squad_xg_sum=1.4199` `squad_xa_sum=0.8239` `team_expected_assists=0.9243` `team_xa_to_xg_ratio=0.65`	-
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 9 by projected rating, fifa_overall 86.0)	`match_squad_rank=9` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 1 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=1` `allocation_applied=true` `goals_allocated=1` `goal_probability=0.3931` `goal_rate=0.5998` `scorer_weight=0.5916`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.0426` `assist_rate=0.0203` `provider_weight=0.0348`	-

Team totals 1 1.42 15 6.9

Bench Candidates (7 modeled substitutes with empirical playing time)

Player	Pos	Exp Min	App Prob	Substitute Modeling Details
Alisson	G	5.0m	13.7%	Market-value substitute prior
Jeremie Frimpong	D	14.7m	35.0%	Market-value substitute prior
Ronald Araújo	D	14.6m	34.5%	Market-value substitute prior
Trey Nyoni	M	22.7m	53.2%	Market-value substitute prior
James McConnell	M	22.0m	49.9%	Market-value substitute prior
Bradley Barcola	F	27.1m	64.1%	Market-value substitute prior

Antagonist / tactical summary

Midfield is evenly contested — Adam Wharton vs Dominik Szoboszlai will be the pivotal duel. Total antagonist xG — Home: 0.85, Away: 0.87. Home danger zone: Striker vs Two Centre Backs — 1v2 Aerial/Physical Battle. Away danger zone: Striker vs Two Centre Backs — 1v2 Aerial/Physical Battle.

Midfield — possession 49.5% / 50.5%, dominance ratio 0.46, key duel: Adam Wharton vs Dominik Szoboszlai

Jørgen Strand Larsen xG 0.33Yéremy Pino xG 0.12Dwight McNeil xG 0.13Dominik Szoboszlai xG 0.19Florian Wirtz xG 0.16Wataru Endo xG 0.03

Summary

Antagonist matchup analysis for Crystal Palace vs Liverpool. Most likely scoreline: 1-1. Lambdas derived from 7 home and 4 away individual player duels. Head-to-head: Crystal Palace 4W-3D-3L vs Liverpool over 10 meetings.

Tactical parity between 3-4-2-1 and 4-2-3-1. Match outcome driven primarily by individual player quality. Midfield is evenly contested — Adam Wharton vs Dominik Szoboszlai will be the pivotal duel. Total antagonist xG — Home: 0.85, Away: 0.87. Home danger zone: Striker vs Two Centre Backs — 1v2 Aerial/Physical Battle. Away danger zone: Striker vs Two Centre Backs — 1v2 Aerial/Physical Battle.

Real Sociedad vs Rayo Vallecano

LaLiga · 20261108 · 4-4-2 vs 4-3-3

Predicted 2-1Winner: Real SociedadWinning H 2-1 (9.9%)Winning A 1-2 (5.5%)Confidence 0.7λ home/away 1.91 / 1.07Most-likely P 9.9%

Actual vs predicted: result not available yet (upcoming fixture).

XGBoost Reasoning & FBref Dataset Tracing

Antagonist + XGBoost match analysis for Real Sociedad vs Rayo Vallecano. Player statistics regressed using position-specific XGBoost models (G, D, M, F) with 2026-2027 FBref priors. Micro-duels resolved across 4-4-2 vs 4-3-3 (duel lambdas: Home 1.874, Away 1.003). Final calibrated Poisson lambdas: Home 1.913, Away 1.071 (most likely score 2-1).

Home Starters 11/11 XGBoost (9 FBref) Away Starters 11/11 XGBoost (8 FBref) Bench Modeled H: 7 / A: 7 Duel λ (H/A) 1.87 / 1.00 Final λ (H/A) 1.91 / 1.07

FBref Team Priors: Real Sociedad (Gls/90: 0.75, GA/90: 0.75 → E[λ]: 0.739) vs Rayo Vallecano (Gls/90: 1.3333333333333333, GA/90: 1.33 → E[λ]: 0.741) | Blend Weight: 0.228

Modeling Pipeline Stages (5 layers)

1. XGBoost player regressions for volume & event rates with market-value quality scaling
2. Multi-source prior blending (FBref 2026-2027 tables + Sofascore season history)
3. Antagonist formation duel matrix resolving 1v1 and zonal matchups
4. Team-level prior calibration (FBref possession and attack/defense efficiencies)
5. Bivariate Poisson matrix with Dixon-Coles low-score correction

Standing & form context

Metric	Real Sociedad	Rayo Vallecano
League position	5.00	11.00
Points per game	1.70	1.26
Attack index	1.12	1.00
Defense (concede) index	1.15	1.09
Form multiplier	1.02	0.97
Standing λ multiplier	1.11	1.03

Lambda decomposition (how the goal rates were built)

Stage	Real Sociedad λ	Rayo Vallecano λ
FBref Squad Prior E[λ] (weight 0.228)	0.739	0.741
Pre-competitive-floor	1.887	0.996
Pre-calibration (Micro-duels)	1.874	1.003
Final λ	1.913	1.071

competition calibration scale 1.035competition observed total goals 2.984quality gap applied Truemismatch score pressure applied Falsecompetitive underdog floor applied Falsestage discount 1.0

Scoreline distribution & markets

Top 5 scorelines	P
1-1	10.4%
2-1	9.9%
1-0	9.6%
2-0	9.3%
3-1	6.3%

Home win	56.9%
Draw	22.2%
Away win	20.8%
Over 2.5	57.3%
BTTS yes	56.1%

Pregame fair line — favorite Real Sociedad (56.9%), fair total 2.98, fair spread 0.84

Projected players vs season baseline

Proj = projected for this match · Season = this season's per-game average from match data · deviation badge compares projected xG to the player's season xG/game.

Real Sociedad — projected line vs season baseline (4-4-2)

Player Pos Min Proj G Proj xG Season xG/g Proj Sh Season Sh/g Proj Rtg Season Rtg Rec Model / FBref Prior

Álex Remiro G 90 0 0.00 - n/a 0.0 7.1 62.2 33 FBref Prior G_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost G_service raw output over 3 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0004, quality 0.518)	`market_value_eur=9600000` `quality_factor=0.5177` `boost_factor=1.0004` `reference_quality=0.505` `history_matches=46` `history_shrink=0.098`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0170)	`blend_weight=0.017` `history_matches=46` `season_overlap_discount=1.0` `per90_to_match_factor=1.0` `prior_reliability=0.2667` `competition=La Liga`	saves: 2.978 -> 2.9996
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` `duel_context={'opponent_faced': 'Sergio Camello, Álvaro García', 'matchup_zone': 'Goalkeeper shot-stopping context', 'opponent_lambda': 1.0034}`	-
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 2 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=2`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.91 (squad xG sum 1.3935, xG scale 1.3728, xA scale 1.7141, team xA 1.24 = 0.65 x lambda)	`scoreline_reset=true` `team_lambda=1.913` `squad_xg_sum=1.3935` `squad_xa_sum=0.7254` `team_expected_assists=1.2435` `team_xa_to_xg_ratio=0.65`	shots_faced: 9.5 -> 10.2 expected_saves: 2.32 -> 2.5 high_pressure_saves: 0.8 -> 0.86
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 2 by projected rating, fifa_overall 81.0)	`match_squad_rank=2` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 2 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=2` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.0` `scorer_weight=0.0` `squad_weight_total=2.1341`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.0` `provider_weight=0.0` `squad_weight_total=1.2255`	-

Jon Martin D 90 0 0.10 0.06 68% 1 0.4 7.0 53.2 32 FBref Prior D_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost D_service raw output over 7 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0063, quality 0.669)	`market_value_eur=24000000` `quality_factor=0.6685` `boost_factor=1.0063` `reference_quality=0.505` `history_matches=34` `history_shrink=0.1282`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0222)	`blend_weight=0.0222` `history_matches=34` `season_overlap_discount=1.0` `per90_to_match_factor=1.0` `prior_reliability=0.2667` `competition=La Liga`	interceptionWon: 1.2874 -> 1.2643 wonTackle: 0.8433 -> 0.8579 fouls: 1.411 -> 1.4019
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` `duel_context={'opponent_faced': 'Sergio Camello', 'matchup_zone': 'Striker vs Two Centre Backs — 1v2 Aerial/Physical Battle'}`	-
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 2 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=2`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.91 (squad xG sum 1.3935, xG scale 1.3728, xA scale 1.7141, team xA 1.24 = 0.65 x lambda)	`scoreline_reset=true` `team_lambda=1.913` `squad_xg_sum=1.3935` `squad_xa_sum=0.7254` `team_expected_assists=1.2435` `team_xa_to_xg_ratio=0.65`	expected_goals: 0.0734 -> 0.1008 expected_assists: 0.0088 -> 0.0151
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 5 by projected rating, fifa_overall 78.0)	`match_squad_rank=5` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 2 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=2` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.0959` `goal_rate=0.1736` `scorer_weight=0.126`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.015` `assist_rate=0.126` `provider_weight=0.0189`	-

Sergio Gómez D 86 0 0.02 0.02 ok 1 0.4 6.8 59.3 32 FBref Prior D_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost D_service raw output over 7 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0021, quality 0.575)	`market_value_eur=14600000` `quality_factor=0.5748` `boost_factor=1.0021` `reference_quality=0.505` `history_matches=45` `history_shrink=0.1`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0165)	`blend_weight=0.0165` `history_matches=45` `season_overlap_discount=1.0` `per90_to_match_factor=0.9584` `prior_reliability=0.2533` `competition=La Liga`	totalCross: 4.2214 -> 4.2474 wasFouled: 0.339 -> 0.3334 fouls: 0.4172 -> 0.4228
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` `duel_context={'opponent_faced': 'Jorge de Frutos', 'matchup_zone': 'Right Wing vs Left Back — Flank Battle'}`	-
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 2 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=2`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.91 (squad xG sum 1.3935, xG scale 1.3728, xA scale 1.7141, team xA 1.24 = 0.65 x lambda)	`scoreline_reset=true` `team_lambda=1.913` `squad_xg_sum=1.3935` `squad_xa_sum=0.7254` `team_expected_assists=1.2435` `team_xa_to_xg_ratio=0.65`	expected_assists: 0.1029 -> 0.1764
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 10 by projected rating, fifa_overall 78.0)	`match_squad_rank=10` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 2 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=2` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.0182` `goal_rate=0.0` `scorer_weight=0.0147`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.1617` `assist_rate=0.116` `provider_weight=0.1915`	-

Jon Aramburu D 86 0 0.02 0.04 47% 1 0.4 6.7 59.9 31 FBref Prior D_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost D_service raw output over 7 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0019, quality 0.567)	`market_value_eur=13900000` `quality_factor=0.5671` `boost_factor=1.0019` `reference_quality=0.505` `history_matches=43` `history_shrink=0.1042`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0163)	`blend_weight=0.0163` `history_matches=43` `season_overlap_discount=1.0` `per90_to_match_factor=0.9573` `prior_reliability=0.24` `competition=La Liga`	fouls: 2.793 -> 2.8254 interceptionWon: 1.2068 -> 1.239 wonTackle: 1.6201 -> 1.6283 wasFouled: 1.6968 -> 1.7038
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` `duel_context={'opponent_faced': 'Álvaro García', 'matchup_zone': 'Left Wing vs Right Back — Flank Battle'}`	-
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 2 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=2`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.91 (squad xG sum 1.3935, xG scale 1.3728, xA scale 1.7141, team xA 1.24 = 0.65 x lambda)	`scoreline_reset=true` `team_lambda=1.913` `squad_xg_sum=1.3935` `squad_xa_sum=0.7254` `team_expected_assists=1.2435` `team_xa_to_xg_ratio=0.65`	expected_goals: 0.0143 -> 0.0196 expected_assists: 0.0243 -> 0.0417
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 14 by projected rating, fifa_overall 77.0)	`match_squad_rank=14` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 2 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=2` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.0194` `goal_rate=0.0045` `scorer_weight=0.0166`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.0408` `assist_rate=0.0` `provider_weight=0.0334`	-

Mamadou Sarr D 87 0 0.09 0.03 269% 0 0.1 6.8 25.1 21 FBref Prior D_service

Step-by-step (9 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost D_service raw output over 7 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0189, quality 0.669)	`market_value_eur=24000000` `quality_factor=0.6685` `boost_factor=1.0189` `reference_quality=0.505` `history_matches=8` `history_shrink=0.3846`	-
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` `duel_context={'opponent_faced': 'Sergio Camello', 'matchup_zone': 'Striker vs Two Centre Backs — 1v2 Aerial/Physical Battle'}`	-
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 2 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=2`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.91 (squad xG sum 1.3935, xG scale 1.3728, xA scale 1.7141, team xA 1.24 = 0.65 x lambda)	`scoreline_reset=true` `team_lambda=1.913` `squad_xg_sum=1.3935` `squad_xa_sum=0.7254` `team_expected_assists=1.2435` `team_xa_to_xg_ratio=0.65`	expected_goals: 0.0688 -> 0.0945 expected_assists: 0.008 -> 0.0137
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 9 by projected rating, fifa_overall 77.0)	`match_squad_rank=9` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 2 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=2` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.0902` `goal_rate=0.0` `scorer_weight=0.0756`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.0136` `assist_rate=0.0` `provider_weight=0.0118`	-

Luka Sučić M 81 0 0.24 0.09 156% 2 1.2 6.9 55.4 34 FBref Prior M_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost M_service raw output over 8 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0037, quality 0.593)	`market_value_eur=16300000` `quality_factor=0.5932` `boost_factor=1.0037` `reference_quality=0.505` `history_matches=31` `history_shrink=0.1389`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0187)	`blend_weight=0.0187` `history_matches=31` `season_overlap_discount=1.0` `per90_to_match_factor=0.9029` `prior_reliability=0.2067` `competition=La Liga`	totalCross: 1.0135 -> 1.0272 interceptionWon: 0.6075 -> 0.5962 totalShots: 1.671 -> 1.6774
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` duel_context={'opponent_faced': 'Adrià Pedrosa', 'matchup_zone': 'Left Mid vs Right Back — Wide Channel', 'duel_win_rate': 0.57, 'pressing_faced': 'LOW', 'numerical_situation': '1v1', 'xg_multiplier': 1.1614, 'skill_edge': 0.0738, 'height_edge': 0.036, 'attacker_skill': 0.6121, 'defender_skill': 0.4481, 'attacker_role': 'LM', 'attacker_mv_price_index': 0.9, 'defender_adjusted_parts': [{'player': 'Adrià Pedrosa', 'market_value': 2700000, 'role': 'RB', 'price_index': 0.7, 'adjusted_quality': 0.4481}], 'xg_override': 0.1715}	aerialWon: 0.4427 -> 1.2 totalShots: 1.6774 -> 1.9481 shotOffTarget: 1.1072 -> 1.2859 onTargetScoringAttempt: 0.5641 -> 0.6551 expectedGoalsOnTarget: 0.2826 -> 0.3282 bigChanceMissed: 0.1874 -> 0.2176 goals: 0.1523 -> 0.1769 expectedGoals: 0.1477 -> 0.1715
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 2 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=2`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.91 (squad xG sum 1.3935, xG scale 1.3728, xA scale 1.7141, team xA 1.24 = 0.65 x lambda)	`scoreline_reset=true` `team_lambda=1.913` `squad_xg_sum=1.3935` `squad_xa_sum=0.7254` `team_expected_assists=1.2435` `team_xa_to_xg_ratio=0.65`	expected_goals: 0.1715 -> 0.2354 expected_assists: 0.1243 -> 0.2131
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 8 by projected rating, fifa_overall 76.0)	`match_squad_rank=8` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 2 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=2` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.2097` `goal_rate=0.1769` `scorer_weight=0.2411`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.1919` `assist_rate=0.1376` `provider_weight=0.2294`	-

Carlos Soler M 84 0 0.15 0.09 78% 1 0.8 7.1 59.6 33 FBref Prior M_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost M_service raw output over 8 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 0.9993, quality 0.482)	`market_value_eur=6600000` `quality_factor=0.4818` `boost_factor=0.9993` `reference_quality=0.505` `history_matches=42` `history_shrink=0.1064`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0106)	`blend_weight=0.0106` `history_matches=42` `season_overlap_discount=1.0` `per90_to_match_factor=0.9369` `prior_reliability=0.1533` `competition=La Liga`	totalShots: 1.4607 -> 1.4495 wonTackle: 0.2835 -> 0.2934
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` duel_context={'opponent_faced': 'Andrei Rațiu', 'matchup_zone': 'Right Mid vs Left Back — Wide Channel', 'duel_win_rate': 0.341, 'pressing_faced': 'MEDIUM', 'numerical_situation': '1v1', 'xg_multiplier': 0.8254, 'skill_edge': -0.1068, 'height_edge': -0.012, 'attacker_skill': 0.4907, 'defender_skill': 0.728, 'attacker_role': 'RM', 'attacker_mv_price_index': 0.9, 'defender_adjusted_parts': [{'player': 'Andrei Rațiu', 'market_value': 22000000, 'role': 'LB', 'price_index': 0.7, 'adjusted_quality': 0.728}], 'xg_override': 0.1109}	aerialWon: 0.1221 -> 1.1 totalShots: 1.456 -> 1.2018 shotOffTarget: 1.0179 -> 0.8402 onTargetScoringAttempt: 0.4381 -> 0.3616 expectedGoalsOnTarget: 0.1844 -> 0.1522 goals: 0.1438 -> 0.1187 expectedGoals: 0.1343 -> 0.1109 bigChanceMissed: 0.0728 -> 0.0601
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 2 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=2`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.91 (squad xG sum 1.3935, xG scale 1.3728, xA scale 1.7141, team xA 1.24 = 0.65 x lambda)	`scoreline_reset=true` `team_lambda=1.913` `squad_xg_sum=1.3935` `squad_xa_sum=0.7254` `team_expected_assists=1.2435` `team_xa_to_xg_ratio=0.65`	expected_goals: 0.1109 -> 0.1522 expected_assists: 0.1665 -> 0.2853
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 3 by projected rating, fifa_overall 79.0)	`match_squad_rank=3` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 2 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=2` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.1412` `goal_rate=0.1187` `scorer_weight=0.1596`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=1` `assist_probability=0.2482` `assist_rate=0.1262` `provider_weight=0.2322`	-

Beñat Turrientes M 78 0 0.12 0.03 355% 1 0.6 6.9 49.7 33 FBref Prior M_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost M_service raw output over 8 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0010, quality 0.529)	`market_value_eur=10600000` `quality_factor=0.5295` `boost_factor=1.001` `reference_quality=0.505` `history_matches=32` `history_shrink=0.1351`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0170)	`blend_weight=0.017` `history_matches=32` `season_overlap_discount=1.0` `per90_to_match_factor=0.8618` `prior_reliability=0.1933` `competition=La Liga`	fouls: 0.6634 -> 0.6521 wasFouled: 0.9162 -> 0.9057 onTargetScoringAttempt: 0.3734 -> 0.3824 interceptionWon: 0.435 -> 0.4276 wonTackle: 0.8128 -> 0.8192
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=true`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` duel_context={'opponent_faced': 'Pathé Ismaël Ciss, Unai López', 'matchup_zone': 'CM vs CDM + CM', 'duel_win_rate': 0.36, 'pressing_faced': 'LOW', 'numerical_situation': '1v2', 'xg_multiplier': 1.0875, 'skill_edge': 0.0476, 'height_edge': 0.012, 'attacker_skill': 0.5431, 'defender_skill': 0.4374, 'attacker_role': 'CM', 'attacker_mv_price_index': 0.9, 'defender_adjusted_parts': [{'player': 'Pathé Ismaël Ciss', 'market_value': 1600000, 'role': 'CDM', 'price_index': 0.85, 'adjusted_quality': 0.4374}, {'player': 'Unai López', 'market_value': 2600000, 'role': 'CM', 'price_index': 0.9, 'adjusted_quality': 0.4374}], 'xg_override': 0.0838}	aerialWon: 0.2815 -> 1.0 totalShots: 1.0695 -> 1.1631 shotOffTarget: 0.6871 -> 0.7472 onTargetScoringAttempt: 0.3824 -> 0.4159 expectedGoalsOnTarget: 0.1996 -> 0.2171 goals: 0.0873 -> 0.0949 expectedGoals: 0.0771 -> 0.0838
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 2 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=2`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.91 (squad xG sum 1.3935, xG scale 1.3728, xA scale 1.7141, team xA 1.24 = 0.65 x lambda)	`scoreline_reset=true` `team_lambda=1.913` `squad_xg_sum=1.3935` `squad_xa_sum=0.7254` `team_expected_assists=1.2435` `team_xa_to_xg_ratio=0.65`	expected_goals: 0.0838 -> 0.115 expected_assists: 0.0756 -> 0.1296
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 6 by projected rating, fifa_overall 81.0)	`match_squad_rank=6` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv_filled` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 2 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=2` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.1086` `goal_rate=0.0949` `scorer_weight=0.1245`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.1216` `assist_rate=0.0667` `provider_weight=0.1204`	-

Jon Gorrotxategi M 78 0 0.05 0.01 705% 1 0.1 6.9 50.0 27 FBref Prior M_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost M_service raw output over 8 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0068, quality 0.694)	`market_value_eur=27000000` `quality_factor=0.6941` `boost_factor=1.0068` `reference_quality=0.505` `history_matches=37` `history_shrink=0.119`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0015)	`blend_weight=0.0015` `history_matches=37` `season_overlap_discount=1.0` `per90_to_match_factor=0.8701` `prior_reliability=0.02` `competition=La Liga`	-
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` duel_context={'opponent_faced': 'Pathé Ismaël Ciss, Unai López', 'matchup_zone': 'CM vs CDM + CM', 'duel_win_rate': 0.43, 'pressing_faced': 'LOW', 'numerical_situation': '1v2', 'xg_multiplier': 1.1912, 'skill_edge': 0.1261, 'height_edge': 0.004, 'attacker_skill': 0.7176, 'defender_skill': 0.4374, 'attacker_role': 'CM', 'attacker_mv_price_index': 0.9, 'defender_adjusted_parts': [{'player': 'Pathé Ismaël Ciss', 'market_value': 1600000, 'role': 'CDM', 'price_index': 0.85, 'adjusted_quality': 0.4374}, {'player': 'Unai López', 'market_value': 2600000, 'role': 'CM', 'price_index': 0.9, 'adjusted_quality': 0.4374}], 'xg_override': 0.0385}	totalShots: 0.4635 -> 0.5521 shotOffTarget: 0.3515 -> 0.4187 onTargetScoringAttempt: 0.112 -> 0.1334 aerialWon: 0.9891 -> 1.0 goals: 0.0451 -> 0.0537 expectedGoals: 0.0323 -> 0.0385
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 2 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=2`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.91 (squad xG sum 1.3935, xG scale 1.3728, xA scale 1.7141, team xA 1.24 = 0.65 x lambda)	`scoreline_reset=true` `team_lambda=1.913` `squad_xg_sum=1.3935` `squad_xa_sum=0.7254` `team_expected_assists=1.2435` `team_xa_to_xg_ratio=0.65`	expected_goals: 0.0385 -> 0.0529 expected_assists: 0.0361 -> 0.0619
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 7 by projected rating, fifa_overall 79.0)	`match_squad_rank=7` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv_filled` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 2 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=2` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.0515` `goal_rate=0.0537` `scorer_weight=0.062`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.06` `assist_rate=0.0453` `provider_weight=0.0685`	-

Mikel Oyarzabal F 82 1 0.44 0.32 38% 3 1.6 7.1 62.8 35 FBref Prior F_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost F_service raw output over 7 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0058, quality 0.694)	`market_value_eur=27000000` `quality_factor=0.6941` `boost_factor=1.0058` `reference_quality=0.505` `history_matches=44` `history_shrink=0.102`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0133)	`blend_weight=0.0133` `history_matches=44` `season_overlap_discount=1.0` `per90_to_match_factor=0.9091` `prior_reliability=0.2` `competition=La Liga`	wasFouled: 1.2792 -> 1.2904 wonTackle: 0.7353 -> 0.7255 goalAssist: 0.119 -> 0.1255 goals: 0.4055 -> 0.4001
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` duel_context={'opponent_faced': 'Florian Lejeune, Mujaid Sadick', 'matchup_zone': 'Striker vs Two Centre Backs — 1v2 Aerial/Physical Battle', 'duel_win_rate': 0.28, 'pressing_faced': 'LOW', 'numerical_situation': '1v2', 'xg_multiplier': 1.1183, 'skill_edge': 0.1065, 'height_edge': -0.026, 'attacker_skill': 0.6941, 'defender_skill': 0.4576, 'attacker_role': 'ST', 'attacker_mv_price_index': 1.0, 'defender_adjusted_parts': [{'player': 'Florian Lejeune', 'market_value': 1700000, 'role': 'CB', 'price_index': 0.75, 'adjusted_quality': 0.4374}, {'player': 'Mujaid Sadick', 'market_value': 4700000, 'role': 'CB', 'price_index': 0.75, 'adjusted_quality': 0.4777}], 'xg_override': 0.3206}	totalShots: 2.2955 -> 2.5671 shotOffTarget: 1.3968 -> 1.562 onTargetScoringAttempt: 0.8987 -> 1.005 bigChanceMissed: 0.4844 -> 0.5417 goals: 0.4001 -> 0.4474 expectedGoals: 0.2867 -> 0.3206 expectedGoalsOnTarget: 0.2834 -> 0.3169
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 2 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=2`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.91 (squad xG sum 1.3935, xG scale 1.3728, xA scale 1.7141, team xA 1.24 = 0.65 x lambda)	`scoreline_reset=true` `team_lambda=1.913` `squad_xg_sum=1.3935` `squad_xa_sum=0.7254` `team_expected_assists=1.2435` `team_xa_to_xg_ratio=0.65`	expected_goals: 0.3206 -> 0.4401 expected_assists: 0.1132 -> 0.194
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 1 by projected rating, fifa_overall 90.0)	`match_squad_rank=1` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv_filled` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 2 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=2` `allocation_applied=true` `goals_allocated=1` `goal_probability=0.356` `goal_rate=0.4474` `scorer_weight=0.5501`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.1763` `assist_rate=0.1255` `provider_weight=0.2086`	-

Orri Óskarsson F 79 1 0.63 0.33 90% 3 1.5 6.8 54.7 30 XGBoost F_service

Step-by-step (9 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost F_service raw output over 7 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0062, quality 0.641)	`market_value_eur=21000000` `quality_factor=0.6408` `boost_factor=1.0062` `reference_quality=0.505` `history_matches=28` `history_shrink=0.1515`	-
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=true`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` duel_context={'opponent_faced': 'Florian Lejeune, Mujaid Sadick', 'matchup_zone': 'Striker vs Two Centre Backs — 1v2 Aerial/Physical Battle', 'duel_win_rate': 0.276, 'pressing_faced': 'LOW', 'numerical_situation': '1v2', 'xg_multiplier': 1.1124, 'skill_edge': 0.0824, 'height_edge': -0.006, 'attacker_skill': 0.6408, 'defender_skill': 0.4576, 'attacker_role': 'ST', 'attacker_mv_price_index': 1.0, 'defender_adjusted_parts': [{'player': 'Florian Lejeune', 'market_value': 1700000, 'role': 'CB', 'price_index': 0.75, 'adjusted_quality': 0.4374}, {'player': 'Mujaid Sadick', 'market_value': 4700000, 'role': 'CB', 'price_index': 0.75, 'adjusted_quality': 0.4777}], 'xg_override': 0.4588}	totalShots: 2.4492 -> 2.7245 shotOffTarget: 1.3664 -> 1.52 bigChanceMissed: 1.1294 -> 1.2563 onTargetScoringAttempt: 1.0828 -> 1.2045 expectedGoalsOnTarget: 0.7421 -> 0.8255 goals: 0.4678 -> 0.5204 expectedGoals: 0.4124 -> 0.4588
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 2 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=2`	goals: 1.0 -> 0
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.91 (squad xG sum 1.3935, xG scale 1.3728, xA scale 1.7141, team xA 1.24 = 0.65 x lambda)	`scoreline_reset=true` `team_lambda=1.913` `squad_xg_sum=1.3935` `squad_xa_sum=0.7254` `team_expected_assists=1.2435` `team_xa_to_xg_ratio=0.65`	expected_goals: 0.4588 -> 0.6299 expected_assists: 0.0323 -> 0.0554
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 11 by projected rating, fifa_overall 74.0)	`match_squad_rank=11` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 2 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=2` `allocation_applied=true` `goals_allocated=1` `goal_probability=0.4674` `goal_rate=0.5204` `scorer_weight=0.7062`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.0539` `assist_rate=0.0` `provider_weight=0.0443`	-

Team totals 2 1.91 14 6.9

Bench Candidates (7 modeled substitutes with empirical playing time)

Player	Pos	Exp Min	App Prob	Substitute Modeling Details
Unai Marrero	G	4.4m	10.9%	Market-value substitute prior
Héctor Fort	D	14.1m	32.3%	Market-value substitute prior
Igor Zubeldia	D	24.9m	31.6%	Market-value substitute prior
Pablo Marín	M	22.2m	50.8%	Market-value substitute prior
Arsen Zakharyan	M	12.2m	69.6%	Market-value substitute prior
Takefusa Kubo	F	18.9m	44.7%	Market-value substitute prior

Rayo Vallecano — projected line vs season baseline (4-3-3)

Player Pos Min Proj G Proj xG Season xG/g Proj Sh Season Sh/g Proj Rtg Season Rtg Rec Model / FBref Prior

Augusto Batalla G 86 0 0.00 - n/a 0.0 6.9 54.2 24 FBref Prior G_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost G_service raw output over 3 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 0.9988, quality 0.472)	`market_value_eur=5800000` `quality_factor=0.472` `boost_factor=0.9988` `reference_quality=0.505` `history_matches=36` `history_shrink=0.122`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0090)	`blend_weight=0.009` `history_matches=36` `season_overlap_discount=1.0` `per90_to_match_factor=0.9524` `prior_reliability=0.1133` `competition=La Liga`	saves: 2.851 -> 2.8405
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` `duel_context={'opponent_faced': 'Orri Óskarsson, Mikel Oyarzabal', 'matchup_zone': 'Goalkeeper shot-stopping context', 'opponent_lambda': 1.8736}`	-
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 1 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=1`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.07 (squad xG sum 1.2060, xG scale 0.8881, xA scale 0.8881, team xA 0.75 = 0.70 x lambda)	`scoreline_reset=true` `team_lambda=1.071` `squad_xg_sum=1.206` `squad_xa_sum=0.8443` `team_expected_assists=0.7498` `team_xa_to_xg_ratio=0.7001`	shots_faced: 17.8 -> 18.2 expected_saves: 4.36 -> 4.46 high_pressure_saves: 1.5 -> 1.53
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 6 by projected rating, fifa_overall 79.0)	`match_squad_rank=6` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 1 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=1` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.0` `scorer_weight=0.0` `squad_weight_total=1.1385`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.0` `provider_weight=0.0` `squad_weight_total=0.7951`	-

Florian Lejeune D 90 0 0.06 0.06 ok 1 1.4 7.0 61.8 33 FBref Prior D_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost D_service raw output over 7 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 0.9980, quality 0.437)	`market_value_eur=1700000` `quality_factor=0.4374` `boost_factor=0.998` `reference_quality=0.505` `history_matches=46` `history_shrink=0.098`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0127)	`blend_weight=0.0127` `history_matches=46` `season_overlap_discount=1.0` `per90_to_match_factor=1.0` `prior_reliability=0.2` `competition=La Liga`	interceptionWon: 1.2522 -> 1.2702 wonTackle: 0.9032 -> 0.9172 totalShots: 1.1275 -> 1.1344 fouls: 1.4091 -> 1.4039 wasFouled: 1.0765 -> 1.0713
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` `duel_context={'opponent_faced': 'Mikel Oyarzabal', 'matchup_zone': 'Striker vs Two Centre Backs — 1v2 Aerial/Physical Battle'}`	-
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 1 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=1`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.07 (squad xG sum 1.2060, xG scale 0.8881, xA scale 0.8881, team xA 0.75 = 0.70 x lambda)	`scoreline_reset=true` `team_lambda=1.071` `squad_xg_sum=1.206` `squad_xa_sum=0.8443` `team_expected_assists=0.7498` `team_xa_to_xg_ratio=0.7001`	expected_goals: 0.063 -> 0.0559
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 3 by projected rating, fifa_overall 79.0)	`match_squad_rank=3` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 1 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=1` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.0544` `goal_rate=0.04` `scorer_weight=0.0447`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.0136` `assist_rate=0.008` `provider_weight=0.0122`	-

Andrei Rațiu D 90 0 0.07 0.05 ok 1 0.8 6.9 60.2 30 FBref Prior D_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost D_service raw output over 7 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0046, quality 0.650)	`market_value_eur=22000000` `quality_factor=0.6503` `boost_factor=1.0046` `reference_quality=0.505` `history_matches=42` `history_shrink=0.1064`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0092)	`blend_weight=0.0092` `history_matches=42` `season_overlap_discount=1.0` `per90_to_match_factor=0.9968` `prior_reliability=0.1333` `competition=La Liga`	fouls: 0.5562 -> 0.5511
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` `duel_context={'opponent_faced': 'Carlos Soler', 'matchup_zone': 'Right Mid vs Left Back — Wide Channel'}`	-
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 1 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=1`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.07 (squad xG sum 1.2060, xG scale 0.8881, xA scale 0.8881, team xA 0.75 = 0.70 x lambda)	`scoreline_reset=true` `team_lambda=1.071` `squad_xg_sum=1.206` `squad_xa_sum=0.8443` `team_expected_assists=0.7498` `team_xa_to_xg_ratio=0.7001`	expected_goals: 0.0736 -> 0.0654 expected_assists: 0.0832 -> 0.0739
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 9 by projected rating, fifa_overall 80.0)	`match_squad_rank=9` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 1 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=1` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.0633` `goal_rate=0.0789` `scorer_weight=0.0682`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.0712` `assist_rate=0.2162` `provider_weight=0.0924`	-

Adrià Pedrosa D 79 0 0.03 0.01 77% 1 0.4 6.7 46.9 32 FBref Prior D_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost D_service raw output over 7 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 0.9975, quality 0.437)	`market_value_eur=2700000` `quality_factor=0.4374` `boost_factor=0.9975` `reference_quality=0.505` `history_matches=35` `history_shrink=0.125`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0038)	`blend_weight=0.0038` `history_matches=35` `season_overlap_discount=1.0` `per90_to_match_factor=0.8797` `prior_reliability=0.0467` `competition=La Liga`	fouls: 1.09 -> 1.0954
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` `duel_context={'opponent_faced': 'Luka Sučić', 'matchup_zone': 'Left Mid vs Right Back — Wide Channel'}`	-
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 1 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=1`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.07 (squad xG sum 1.2060, xG scale 0.8881, xA scale 0.8881, team xA 0.75 = 0.70 x lambda)	`scoreline_reset=true` `team_lambda=1.071` `squad_xg_sum=1.206` `squad_xa_sum=0.8443` `team_expected_assists=0.7498` `team_xa_to_xg_ratio=0.7001`	expected_assists: 0.1159 -> 0.1029
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 16 by projected rating, fifa_overall 74.0)	`match_squad_rank=16` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 1 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=1` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.026` `goal_rate=0.0853` `scorer_weight=0.0329`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.0978` `assist_rate=0.1002` `provider_weight=0.0932`	-

Mujaid Sadick D 87 0 0.02 0.01 101% 0 0.1 6.8 40.3 10 XGBoost D_service

Step-by-step (9 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost D_service raw output over 7 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 0.9937, quality 0.458)	`market_value_eur=4700000` `quality_factor=0.4585` `boost_factor=0.9937` `reference_quality=0.505` `history_matches=6` `history_shrink=0.4545`	-
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` `duel_context={'opponent_faced': 'Mikel Oyarzabal', 'matchup_zone': 'Striker vs Two Centre Backs — 1v2 Aerial/Physical Battle'}`	-
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 1 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=1`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.07 (squad xG sum 1.2060, xG scale 0.8881, xA scale 0.8881, team xA 0.75 = 0.70 x lambda)	`scoreline_reset=true` `team_lambda=1.071` `squad_xg_sum=1.206` `squad_xa_sum=0.8443` `team_expected_assists=0.7498` `team_xa_to_xg_ratio=0.7001`	-
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 11 by projected rating, fifa_overall 84.0)	`match_squad_rank=11` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv_filled` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 1 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=1` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.0216` `goal_rate=0.0` `scorer_weight=0.0174`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.0042` `assist_rate=0.0` `provider_weight=0.0038`	-

Unai López M 81 0 0.06 0.03 108% 1 0.6 7.2 53.4 30 FBref Prior M_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost M_service raw output over 8 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 0.9976, quality 0.437)	`market_value_eur=2600000` `quality_factor=0.4374` `boost_factor=0.9976` `reference_quality=0.505` `history_matches=38` `history_shrink=0.1163`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0136)	`blend_weight=0.0136` `history_matches=38` `season_overlap_discount=1.0` `per90_to_match_factor=0.9012` `prior_reliability=0.18` `competition=La Liga`	totalShots: 1.1845 -> 1.1729 fouls: 0.955 -> 0.9647
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` duel_context={'opponent_faced': 'Beñat Turrientes', 'matchup_zone': 'Central Mid vs Central Mid — Box-to-Box Midfield Contest', 'duel_win_rate': 0.332, 'pressing_faced': 'LOW', 'numerical_situation': '1v1', 'xg_multiplier': 0.8713, 'skill_edge': -0.0476, 'height_edge': -0.04, 'attacker_skill': 0.4374, 'defender_skill': 0.5431, 'attacker_role': 'CM', 'attacker_mv_price_index': 0.9, 'defender_adjusted_parts': [{'player': 'Beñat Turrientes', 'market_value': 10600000, 'role': 'CM', 'price_index': 0.9, 'adjusted_quality': 0.5431}], 'xg_override': 0.0637}	aerialWon: 0.5488 -> 0.8 totalShots: 1.1846 -> 1.0322 shotOffTarget: 0.861 -> 0.7502 onTargetScoringAttempt: 0.3236 -> 0.282 goals: 0.102 -> 0.0889 expectedGoalsOnTarget: 0.0893 -> 0.0778 expectedGoals: 0.0731 -> 0.0637
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 1 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=1`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.07 (squad xG sum 1.2060, xG scale 0.8881, xA scale 0.8881, team xA 0.75 = 0.70 x lambda)	`scoreline_reset=true` `team_lambda=1.071` `squad_xg_sum=1.206` `squad_xa_sum=0.8443` `team_expected_assists=0.7498` `team_xa_to_xg_ratio=0.7001`	expected_goals: 0.0637 -> 0.0566 expected_assists: 0.151 -> 0.134
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 2 by projected rating, fifa_overall 76.0)	`match_squad_rank=2` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 1 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=1` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.055` `goal_rate=0.0889` `scorer_weight=0.0691`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.1254` `assist_rate=0.0986` `provider_weight=0.1072`	-

Pathé Ismaël Ciss M 90 0 0.06 0.10 47% 1 0.5 6.9 62.2 31 XGBoost M_service

Step-by-step (9 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost M_service raw output over 8 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 0.9976, quality 0.437)	`market_value_eur=1600000` `quality_factor=0.4374` `boost_factor=0.9976` `reference_quality=0.505` `history_matches=38` `history_shrink=0.1163`	-
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=true`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` duel_context={'opponent_faced': 'Beñat Turrientes', 'matchup_zone': 'Holding Mid vs Central Mid — Midfield Anchor Duel', 'duel_win_rate': 0.368, 'pressing_faced': 'LOW', 'numerical_situation': '1v1', 'xg_multiplier': 0.9536, 'skill_edge': -0.0476, 'height_edge': 0.016, 'attacker_skill': 0.4374, 'defender_skill': 0.5431, 'attacker_role': 'CDM', 'attacker_mv_price_index': 0.85, 'defender_adjusted_parts': [{'player': 'Beñat Turrientes', 'market_value': 10600000, 'role': 'CM', 'price_index': 0.9, 'adjusted_quality': 0.5431}], 'xg_override': 0.0627}	aerialWon: 1.4438 -> 1.1 totalShots: 0.6737 -> 0.6425 shotOffTarget: 0.5027 -> 0.4794 bigChanceMissed: 0.2374 -> 0.2264 onTargetScoringAttempt: 0.171 -> 0.1631
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 1 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=1`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.07 (squad xG sum 1.2060, xG scale 0.8881, xA scale 0.8881, team xA 0.75 = 0.70 x lambda)	`scoreline_reset=true` `team_lambda=1.071` `squad_xg_sum=1.206` `squad_xa_sum=0.8443` `team_expected_assists=0.7498` `team_xa_to_xg_ratio=0.7001`	expected_goals: 0.0627 -> 0.0557
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 5 by projected rating, fifa_overall 82.0)	`match_squad_rank=5` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv_filled` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 1 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=1` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.0542` `goal_rate=0.0691` `scorer_weight=0.0589`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.0288` `assist_rate=0.0608` `provider_weight=0.0365`	-

Pedro Díaz M 80 0 0.11 0.06 85% 2 1.4 6.9 58.0 33 FBref Prior M_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost M_service raw output over 8 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 0.9979, quality 0.437)	`market_value_eur=2900000` `quality_factor=0.4374` `boost_factor=0.9979` `reference_quality=0.505` `history_matches=43` `history_shrink=0.1042`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0023)	`blend_weight=0.0023` `history_matches=43` `season_overlap_discount=1.0` `per90_to_match_factor=0.8847` `prior_reliability=0.0333` `competition=La Liga`	totalCross: 0.9842 -> 0.9979
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` duel_context={'opponent_faced': 'Beñat Turrientes', 'matchup_zone': 'Central Mid vs Central Mid — Box-to-Box Midfield Contest', 'duel_win_rate': 0.378, 'pressing_faced': 'LOW', 'numerical_situation': '1v1', 'xg_multiplier': 0.9378, 'skill_edge': -0.0463, 'height_edge': 0.004, 'attacker_skill': 0.4402, 'defender_skill': 0.5431, 'attacker_role': 'CM', 'attacker_mv_price_index': 0.9, 'defender_adjusted_parts': [{'player': 'Beñat Turrientes', 'market_value': 10600000, 'role': 'CM', 'price_index': 0.9, 'adjusted_quality': 0.5431}], 'xg_override': 0.1245}	aerialWon: 0.11 -> 1.1 totalShots: 1.9991 -> 1.8748 shotOffTarget: 1.3477 -> 1.2639 onTargetScoringAttempt: 0.6514 -> 0.6109 goals: 0.1398 -> 0.1311 expectedGoals: 0.1328 -> 0.1245
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 1 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=1`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.07 (squad xG sum 1.2060, xG scale 0.8881, xA scale 0.8881, team xA 0.75 = 0.70 x lambda)	`scoreline_reset=true` `team_lambda=1.071` `squad_xg_sum=1.206` `squad_xa_sum=0.8443` `team_expected_assists=0.7498` `team_xa_to_xg_ratio=0.7001`	expected_goals: 0.1245 -> 0.1106 expected_assists: 0.1153 -> 0.1024
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 7 by projected rating, fifa_overall 75.0)	`match_squad_rank=7` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 1 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=1` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.1047` `goal_rate=0.1311` `scorer_weight=0.1148`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.0973` `assist_rate=0.1085` `provider_weight=0.0982`	-

Sergio Camello F 70 1 0.22 0.18 ok 2 1.5 7.2 49.2 31 FBref Prior F_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost F_service raw output over 7 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0011, quality 0.533)	`market_value_eur=10900000` `quality_factor=0.533` `boost_factor=1.0011` `reference_quality=0.505` `history_matches=32` `history_shrink=0.1351`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0123)	`blend_weight=0.0123` `history_matches=32` `season_overlap_discount=1.0` `per90_to_match_factor=0.7779` `prior_reliability=0.14` `competition=La Liga`	wasFouled: 1.2804 -> 1.292 goals: 0.3215 -> 0.3312 totalCross: 0.1353 -> 0.1427 onTargetScoringAttempt: 0.9342 -> 0.9407
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` duel_context={'opponent_faced': 'Jon Martin, Mamadou Sarr', 'matchup_zone': 'Striker vs Two Centre Backs — 1v2 Aerial/Physical Battle', 'duel_win_rate': 0.18, 'pressing_faced': 'MEDIUM', 'numerical_situation': '1v2', 'xg_multiplier': 0.7948, 'skill_edge': -0.0896, 'height_edge': -0.05, 'attacker_skill': 0.533, 'defender_skill': 0.732, 'attacker_role': 'ST', 'attacker_mv_price_index': 1.0, 'defender_adjusted_parts': [{'player': 'Jon Martin', 'market_value': 24000000, 'role': 'CB', 'price_index': 0.75, 'adjusted_quality': 0.732}, {'player': 'Mamadou Sarr', 'market_value': 24000000, 'role': 'CB', 'price_index': 0.75, 'adjusted_quality': 0.732}], 'xg_override': 0.2498}	totalShots: 2.2732 -> 1.8068 shotOffTarget: 1.3325 -> 1.0591 onTargetScoringAttempt: 0.9407 -> 0.7477 expectedGoalsOnTarget: 0.375 -> 0.298 bigChanceMissed: 0.3646 -> 0.2898 goals: 0.3312 -> 0.2632 expectedGoals: 0.3143 -> 0.2498
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 1 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=1`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.07 (squad xG sum 1.2060, xG scale 0.8881, xA scale 0.8881, team xA 0.75 = 0.70 x lambda)	`scoreline_reset=true` `team_lambda=1.071` `squad_xg_sum=1.206` `squad_xa_sum=0.8443` `team_expected_assists=0.7498` `team_xa_to_xg_ratio=0.7001`	expected_goals: 0.2498 -> 0.2219 expected_assists: 0.1251 -> 0.1111
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 1 by projected rating, fifa_overall 74.0)	`match_squad_rank=1` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 1 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=1` `allocation_applied=true` `goals_allocated=1` `goal_probability=0.199` `goal_rate=0.2632` `scorer_weight=0.2318`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.1052` `assist_rate=0.218` `provider_weight=0.1389`	-

Jorge de Frutos F 72 0 0.16 0.20 ok 2 1.2 6.8 62.5 36 XGBoost F_service

Step-by-step (9 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost F_service raw output over 7 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 1.0009, quality 0.533)	`market_value_eur=10900000` `quality_factor=0.533` `boost_factor=1.0009` `reference_quality=0.505` `history_matches=44` `history_shrink=0.102`	-
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=false`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` duel_context={'opponent_faced': 'Sergio Gómez', 'matchup_zone': 'Right Wing vs Left Back — Flank Battle', 'duel_win_rate': 0.44, 'pressing_faced': 'LOW', 'numerical_situation': '1v1', 'xg_multiplier': 0.9414, 'skill_edge': -0.0479, 'height_edge': 0.008, 'attacker_skill': 0.533, 'defender_skill': 0.6394, 'attacker_role': 'RW', 'attacker_mv_price_index': 1.0, 'defender_adjusted_parts': [{'player': 'Sergio Gómez', 'market_value': 14600000, 'role': 'LB', 'price_index': 0.7, 'adjusted_quality': 0.6394}], 'xg_override': 0.1812}	totalShots: 1.5184 -> 1.4294 shotOffTarget: 0.8093 -> 0.7619 onTargetScoringAttempt: 0.7091 -> 0.6675 expectedGoalsOnTarget: 0.3493 -> 0.3288 goals: 0.2819 -> 0.2654 expectedGoals: 0.1925 -> 0.1812 bigChanceMissed: 0.088 -> 0.0828
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 1 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=1`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.07 (squad xG sum 1.2060, xG scale 0.8881, xA scale 0.8881, team xA 0.75 = 0.70 x lambda)	`scoreline_reset=true` `team_lambda=1.071` `squad_xg_sum=1.206` `squad_xa_sum=0.8443` `team_expected_assists=0.7498` `team_xa_to_xg_ratio=0.7001`	expected_goals: 0.1812 -> 0.1609 expected_assists: 0.0565 -> 0.0502
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 10 by projected rating, fifa_overall 81.0)	`match_squad_rank=10` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv_filled` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 1 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=1` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.1486` `goal_rate=0.2654` `scorer_weight=0.2011`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=0` `assist_probability=0.049` `assist_rate=0.1657` `provider_weight=0.0628`	-

Álvaro García F 77 0 0.18 0.16 ok 1 1.0 6.9 55.2 38 FBref Prior F_service

Step-by-step (10 steps)

Step	Description	Key params	Changes
`xgboost_raw`	XGBoost F_service raw output over 7 target models	-	-
`market_value_boost`	Market-value quality scaling (boost 0.9978, quality 0.437)	`market_value_eur=1500000` `quality_factor=0.4374` `boost_factor=0.9978` `reference_quality=0.505` `history_matches=42` `history_shrink=0.1064`	-
`fbref_prior_blend`	FBref 2026-27 prior blended (weight 0.0101)	`blend_weight=0.0101` `history_matches=42` `season_overlap_discount=1.0` `per90_to_match_factor=0.8508` `prior_reliability=0.1467` `competition=La Liga`	totalCross: 3.5776 -> 3.5648 totalShots: 1.6597 -> 1.6508 fouls: 0.4655 -> 0.4725
`final_cap`	Stats clamped to historical constraints and shot-component identity	`stats_clamped=0` `shot_identity_applied=true`	-
`enriched_model_output`	Model output after duel-context enrichment (pre router post-processing)	`duel_context_applied=true` duel_context={'opponent_faced': 'Jon Aramburu', 'matchup_zone': 'Left Wing vs Right Back — Flank Battle', 'duel_win_rate': 0.362, 'pressing_faced': 'LOW', 'numerical_situation': '1v1', 'xg_multiplier': 0.8258, 'skill_edge': -0.0865, 'height_edge': -0.032, 'attacker_skill': 0.4374, 'defender_skill': 0.6297, 'attacker_role': 'LW', 'attacker_mv_price_index': 1.0, 'defender_adjusted_parts': [{'player': 'Jon Aramburu', 'market_value': 13900000, 'role': 'RB', 'price_index': 0.7, 'adjusted_quality': 0.6297}], 'xg_override': 0.2028}	totalShots: 1.6581 -> 1.3693 shotOffTarget: 1.1059 -> 0.9133 onTargetScoringAttempt: 0.5522 -> 0.456 goals: 0.252 -> 0.2081 expectedGoals: 0.2456 -> 0.2028 expectedGoalsOnTarget: 0.2418 -> 0.1997 bigChanceMissed: 0.2102 -> 0.1736
`scoreline_reset`	Team scoreline allocation: goal/assist counters reset to 0; the team's 1 goals are handed out as whole numbers by integer_goal_projection / integer_assist_projection	`scoreline_reset=true` `team_goals=1`	-
`team_lambda_rescale`	Player expected_goals/expected_assists rescaled to unconditional contributions so squad sum equals team lambda 1.07 (squad xG sum 1.2060, xG scale 0.8881, xA scale 0.8881, team xA 0.75 = 0.70 x lambda)	`scoreline_reset=true` `team_lambda=1.071` `squad_xg_sum=1.206` `squad_xa_sum=0.8443` `team_expected_assists=0.7498` `team_xa_to_xg_ratio=0.7001`	expected_goals: 0.2028 -> 0.1801 expected_assists: 0.0996 -> 0.0885
`rankings_attached`	Display-only provenance: no published number changes. Global ranking attached (match squad rank 8 by projected rating, fifa_overall 81.0)	`match_squad_rank=8` `squad_role=starter` `squad_rank_basis=projected_rating` `affects_projection=false` `ranking_source=sofascore_plus_fifa_csv` `ranking_status=combined`	-
`integer_goal_projection`	Team scoreline allocation: 1 projected goals distributed by scorer weight (xG x finishing x appearance)	`team_goals=1` `allocation_applied=true` `goals_allocated=0` `goal_probability=0.1648` `goal_rate=0.2081` `scorer_weight=0.1839`	-
`integer_assist_projection`	Team scoreline allocation: 1 projected assists distributed by provider weight (xA x creation x appearance; nobody assists his own goal)	`allocation_applied=true` `team_assists=1` `assists_allocated=1` `assist_probability=0.0847` `assist_rate=0.1809` `provider_weight=0.1106`	-

Team totals 1 1.07 15 6.9

Bench Candidates (7 modeled substitutes with empirical playing time)

Player	Pos	Exp Min	App Prob	Substitute Modeling Details
Emil Audero	G	4.4m	10.9%	Market-value substitute prior
Marash Kumbulla	D	13.6m	29.9%	Market-value substitute prior
Luiz Felipe	D	13.6m	29.9%	Market-value substitute prior
Isi Palazón	M	22.0m	49.9%	Market-value substitute prior
Óscar Valentín	M	29.7m	43.5%	Market-value substitute prior
Alemão	F	25.4m	56.2%	Market-value substitute prior

Antagonist / tactical summary

Home midfield dominates with ~52% possession expected (Jon Gorrotxategi vs Pedro Díaz). Total antagonist xG — Home: 1.18, Away: 0.88. Home danger zone: Striker vs Two Centre Backs — 1v2 Aerial/Physical Battle. Away danger zone: Striker vs Two Centre Backs — 1v2 Aerial/Physical Battle.

Midfield — possession 52.4% / 47.6%, dominance ratio 0.58, key duel: Jon Gorrotxategi vs Pedro Díaz

Mikel Oyarzabal xG 0.32Orri Óskarsson xG 0.46Luka Sučić xG 0.17Álvaro García xG 0.20Jorge de Frutos xG 0.18Sergio Camello xG 0.25

Summary

Antagonist matchup analysis for Real Sociedad vs Rayo Vallecano. Most likely scoreline: 2-1. Lambdas derived from 6 home and 6 away individual player duels. Head-to-head: Real Sociedad 3W-5D-2L vs Rayo Vallecano over 10 meetings.

Tactical parity between 4-4-2 and 4-3-3. Match outcome driven primarily by individual player quality. Home midfield dominates with ~52% possession expected (Jon Gorrotxategi vs Pedro Díaz). Total antagonist xG — Home: 1.18, Away: 0.88. Home danger zone: Striker vs Two Centre Backs — 1v2 Aerial/Physical Battle. Away danger zone: Striker vs Two Centre Backs — 1v2 Aerial/Physical Battle.

Generated 2026-11-03 14:38:27 UTC · Soccer prediction reporting