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 7
14 games · 0 of them land in the coin-flip band, where the model has no usable call
- NEatCHIThu, Oct 22, 8:15 PMCHI 58%CHI -0.8Low confidence
- PITatNOSun, Oct 25, 9:30 AMPIT 58%PIT -1.0Low confidence
- SFatATLSun, Oct 25, 1:00 PMATL 65%ATL -2.3High confidence
- CINatBALSun, Oct 25, 1:00 PMBAL 59%BAL -2.9Low confidence
- NYGatHOUSun, Oct 25, 1:00 PMHOU 66%HOU -3.9High confidence
- CLEatTENSun, Oct 25, 1:00 PMTEN 60%TEN -2.9Medium confidence
- TBatCARSun, Oct 25, 1:00 PMTB 59%TB -2.9Low confidence
- INDatMINSun, Oct 25, 1:00 PMMIN 62%MIN -2.0Medium confidence
- MIAatNYJSun, Oct 25, 1:00 PMNYJ 60%NYJ -2.5Low confidence
- DENatARISun, Oct 25, 4:05 PMARI 57%ARI -1.5Low confidence
- LARatLVSun, Oct 25, 4:25 PMLV 58%LV -1.7Low confidence
- GBatDETSun, Oct 25, 4:25 PMDET 64%DET -3.6Medium confidence
- KCatSEASun, Oct 25, 8:20 PMSEA 67%SEA -5.0High confidence
- DALatPHIMon, Oct 26, 8:15 PMPHI 55%PHI -2.5Low confidence
Calibrated Logistic Regression (core + market features) · walk-forward by season (train on all prior seasons, test on the next)