Sports Dashboard

MI Bivariate Poisson + Dixon-Coles + Elo

← Back to Blog
|Signal Test|REJECTED

Signal overperformance-decomposition: Rejected

Decomposing overperformance into shot_quality (shotXG - matchXG) vs finishing_luck (actual - shotXG) predicts regression magnitude. Teams with high finishing_luck component regress faster than teams where overperformance is explained by shot quality.. no-signal

Marginal ROI
-0.6%
entry-adjusted
N
145,182
CLV
+5.7%
Gates
7/10

Hypothesis

Decomposing overperformance into shot_quality (shotXG - matchXG) vs finishing_luck (actual - shotXG) predicts regression magnitude. Teams with high finishing_luck component regress faster than teams where overperformance is explained by shot quality.

Result

Threshold sweep (0.10-0.50) showed no viable threshold. Standalone N=133784 at all thresholds (no differentiation). Marginal ROI: -0.64% at 0.10, -0.02% at 0.50. V3 xG data only covers 2023+, so filter has no effect on majority of backtest period. Original +2.68pp claim not reproduced.

Gate Results

GateNameResult
1Pre-registeredPASS
2True standalonePASS
3Minimum NPASS
4Marginal ROIFAIL
5Bootstrap marginal significanceFAIL
6Matchday interleave OOSPASS
7Regime stratificationPASS
8Suspicious NPASS
9Practical significanceFAIL
10Walk-forwardPASS

Rejection Reason

no-signal

Failed gates: Marginal ROI, Bootstrap marginal significance, Practical significance

REJECTEDSignal: overperformance-decomposition|2026-04-11