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 18
16 games · 8 of them land in the coin-flip band, where the model has no usable call
- TBatNOSun, Jan 10, 1:00 PMNO 56%NO -3.0Low confidence
- PITatBALSun, Jan 10, 1:00 PMBAL 58%BAL -3.7Low confidence
- SFatARISun, Jan 10, 1:00 PMSF 50%ARI -0.6Coin flip, no call
- DETatGBSun, Jan 10, 1:00 PMGB 54%GB -2.5Coin flip, no call
- PHIatNYGSun, Jan 10, 1:00 PMNYG 54%NYG -2.5Coin flip, no call
- NYJatBUFSun, Jan 10, 1:00 PMBUF 56%BUF -4.2Low confidence
- MIAatNESun, Jan 10, 1:00 PMNE 62%NE -3.8Medium confidence
- ATLatCARSun, Jan 10, 1:00 PMCAR 50%CAR -2.2Coin flip, no call
- LVatKCSun, Jan 10, 1:00 PMKC 54%KC -3.2Coin flip, no call
- JAXatINDSun, Jan 10, 1:00 PMIND 58%IND -3.3Low confidence
- LACatDENSun, Jan 10, 1:00 PMDEN 62%DEN -3.9Medium confidence
- TENatHOUSun, Jan 10, 1:00 PMHOU 56%HOU -3.0Low confidence
- SEAatLARSun, Jan 10, 1:00 PMLAR 51%LAR -2.6Coin flip, no call
- CLEatCINSun, Jan 10, 1:00 PMCIN 66%CIN -6.6High confidence
- CHIatMINSun, Jan 10, 1:00 PMMIN 51%MIN -3.6Coin flip, no call
- DALatWASSun, Jan 10, 1:00 PMWAS 55%WAS -2.6Coin flip, no call
Calibrated Logistic Regression (core + market features) · walk-forward by season (train on all prior seasons, test on the next)