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 8
14 games · 2 of them land in the coin-flip band, where the model has no usable call
- CARatGBThu, Oct 29, 8:15 PMGB 62%GB -2.2Medium confidence
- LVatNYJSun, Nov 1, 1:00 PMNYJ 65%NYJ -4.5Medium confidence
- TENatCINSun, Nov 1, 1:00 PMCIN 67%CIN -5.3High confidence
- MINatDETSun, Nov 1, 1:00 PMDET 60%DET -2.7Low confidence
- ATLatTBSun, Nov 1, 1:00 PMTB 54%TB -2.3Coin flip, no call
- CLEatPITSun, Nov 1, 1:00 PMPIT 61%PIT -4.4Medium confidence
- ARIatDALSun, Nov 1, 1:00 PMDAL 68%DAL -3.8High confidence
- BALatBUFSun, Nov 1, 1:00 PMBUF 67%BUF -3.9High confidence
- INDatJAXSun, Nov 1, 1:00 PMJAX 63%JAX -2.5Medium confidence
- LACatLARSun, Nov 1, 4:05 PMLAR 61%LAR -3.9Medium confidence
- NEatMIASun, Nov 1, 4:25 PMMIA 50%MIA -1.8Coin flip, no call
- KCatDENSun, Nov 1, 4:25 PMDEN 59%DEN -2.3Low confidence
- PHIatWASSun, Nov 1, 8:20 PMWAS 60%WAS -1.8Medium confidence
- CHIatSEAMon, Nov 2, 8:15 PMSEA 66%SEA -4.7High confidence
Calibrated Logistic Regression (core + market features) · walk-forward by season (train on all prior seasons, test on the next)