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 10
14 games · 1 of them land in the coin-flip band, where the model has no usable call
- WASatNYGThu, Nov 12, 8:15 PMNYG 59%NYG -2.1Low confidence
- NEatDETSun, Nov 15, 9:30 AMNE 56%DET -2.1Low confidence
- CARatNOSun, Nov 15, 1:00 PMNO 63%NO -4.0Medium confidence
- MINatGBSun, Nov 15, 1:00 PMGB 58%GB -2.0Low confidence
- KCatATLSun, Nov 15, 1:00 PMATL 64%ATL -3.1Medium confidence
- MIAatINDSun, Nov 15, 1:00 PMIND 67%IND -3.8High confidence
- BUFatNYJSun, Nov 15, 1:00 PMNYJ 61%NYJ -3.6Medium confidence
- HOUatCLESun, Nov 15, 1:00 PMCLE 58%CLE -2.3Low confidence
- JAXatTENSun, Nov 15, 1:00 PMTEN 58%TEN -1.1Low confidence
- LARatARISun, Nov 15, 4:05 PMARI 56%ARI -0.7Low confidence
- SEAatLVSun, Nov 15, 4:05 PMLV 54%LV -1.5Coin flip, no call
- SFatDALSun, Nov 15, 4:25 PMDAL 64%DAL -2.7Medium confidence
- PITatCINSun, Nov 15, 8:20 PMCIN 60%CIN -3.5Low confidence
- LACatBALMon, Nov 16, 8:15 PMBAL 68%BAL -5.4High confidence
Calibrated Logistic Regression (core + market features) · walk-forward by season (train on all prior seasons, test on the next)