Win probability
Game Prediction Model
- 20seasons of data2006–2025
- 5,133games trained on4,166 more held back for testing
- 64.7%called correctlyon games the model never saw
- ±2.1points of calibration driftwhen it says 70%, it happens about 70% of the time
23.2%decided by 3 points or fewer
47.5%decided by 7 points or fewer
8points, the median margin
A quarter of NFL games come down to a field goal, which is the ceiling every forecaster runs into. Predicting every game decided by more than a touchdown perfectly, and coin-flipping the rest, would still only reach 76.3%. Across 5,133 games, 2006–2025.
Week 9
15 games · 0 of them land in the coin-flip band, where the model has no usable call
- JAXatBALThu, Nov 5, 8:15 PMBAL 65%BAL -2.9Medium confidence
- CINatATLSun, Nov 8, 9:30 AMCIN 62%CIN -3.9Medium confidence
- DENatCARSun, Nov 8, 1:00 PMCAR 59%CAR -3.7Low confidence
- LARatWASSun, Nov 8, 1:00 PMWAS 62%WAS -3.4Medium confidence
- DETatMIASun, Nov 8, 1:00 PMMIA 58%MIA -3.5Low confidence
- CLEatNOSun, Nov 8, 1:00 PMNO 72%NO -4.8High confidence
- NYGatPHISun, Nov 8, 1:00 PMPHI 56%PHI -3.6Low confidence
- DALatINDSun, Nov 8, 1:00 PMIND 61%IND -3.0Medium confidence
- NYJatKCSun, Nov 8, 1:00 PMKC 58%KC -2.7Low confidence
- LVatSFSun, Nov 8, 4:05 PMSF 65%SF -3.9Medium confidence
- HOUatLACSun, Nov 8, 4:05 PMLAC 59%LAC -2.6Low confidence
- GBatNESun, Nov 8, 4:25 PMNE 66%NE -4.6High confidence
- ARIatSEASun, Nov 8, 4:25 PMSEA 63%SEA -4.5Medium confidence
- TBatCHISun, Nov 8, 8:20 PMCHI 58%CHI -2.2Low confidence
- BUFatMINMon, Nov 9, 8:15 PMMIN 55%MIN -2.6Low confidence
Calibrated Logistic Regression (core + market features) · walk-forward by season (train on all prior seasons, test on the next)