Win probability · scoutindex-winprob-v1.5
The Elo rating going into any game only uses games played before it, so scoring the model across all 1999–2026 games is a fair walk-forward test — not a fit-and-score-the-same-data backtest.
Straight-up
64.3%
7,354 games · coin flip = 50%
Brier score
0.220
12% better than a coin flip
Log loss
0.630
coin flip = 0.693
Margin error
10.5 pts
bias -0.05
2026 season: 57.0% straight up, 0.237 Brier over 93 games. About 66–67% straight-up is the practical ceiling for any pregame NFL model.
Calibration
When the model says a team has a 70% chance, does it win about 70% of the time?
| Predicted | Games | Model | Actual |
|---|---|---|---|
| 50–55% | 1,458 | 52.5% | 53.8% |
| 55–60% | 1,403 | 57.5% | 58.3% |
| 60–65% | 1,276 | 62.4% | 59.9% |
| 65–70% | 1,103 | 67.4% | 65.7% |
| 70–75% | 860 | 72.5% | 72.3% |
| 75–80% | 640 | 77.4% | 76.4% |
| 80–85% | 392 | 82.3% | 83.4% |
| 85–90% | 182 | 87.1% | 87.4% |
| 90–95% | 37 | 91.2% | 91.9% |
By season
| Season | Games | Straight-up | Brier | Margin MAE |
|---|---|---|---|---|
| 2026 | 93 | 57.0% | 0.237 | 9.7 |
| 2025 | 284 | 64.4% | 0.221 | 10.1 |
| 2024 | 285 | 68.4% | 0.209 | 10.1 |
| 2023 | 285 | 60.0% | 0.233 | 10.5 |
| 2022 | 282 | 64.2% | 0.222 | 9.1 |
| 2021 | 284 | 60.9% | 0.232 | 11.1 |
| 2020 | 268 | 66.0% | 0.219 | 10.1 |
| 2019 | 266 | 64.7% | 0.222 | 10.5 |
| 2018 | 265 | 62.3% | 0.220 | 10.3 |
| 2017 | 267 | 64.8% | 0.217 | 10.5 |
| 2016 | 265 | 66.0% | 0.217 | 9.4 |
| 2015 | 267 | 67.8% | 0.220 | 10.0 |
| 2014 | 266 | 67.7% | 0.209 | 11.3 |
| 2013 | 266 | 66.9% | 0.214 | 10.0 |
| 2012 | 266 | 63.9% | 0.218 | 11.0 |
| 2011 | 267 | 64.4% | 0.216 | 10.8 |
| 2010 | 267 | 63.7% | 0.230 | 11.0 |
| 2009 | 267 | 69.7% | 0.208 | 11.6 |
| 2008 | 266 | 63.9% | 0.218 | 11.1 |
| 2007 | 267 | 63.3% | 0.218 | 11.1 |
| 2006 | 267 | 59.6% | 0.231 | 10.8 |
| 2005 | 267 | 68.2% | 0.208 | 10.4 |
| 2004 | 267 | 64.0% | 0.225 | 10.6 |
| 2003 | 267 | 64.0% | 0.221 | 10.9 |
| 2002 | 266 | 59.4% | 0.227 | 10.5 |
| 2001 | 259 | 64.1% | 0.226 | 10.3 |
| 2000 | 259 | 64.1% | 0.223 | 10.7 |
| 1999 | 259 | 61.4% | 0.224 | 10.5 |
Player projections · scoutindex-projection-v1
Among players who played the projected game. "Baseline" is the error from simply projecting the positional average — the model should beat it.
| Stat | Scored | MAE | Baseline MAE | Bias | In 10–90 band |
|---|---|---|---|---|---|
| Receiving yards | 1,573 | 20.5 | 24.2 | -4.1 | 66% |
| Receptions | 1,573 | 1.5 | 1.9 | -0.2 | 85% |
| Receiving TD | 1,108 | 0.3 | 0.3 | -0.0 | 96% |
| Rushing yards | 651 | 21.1 | 25.9 | -6.8 | 63% |
| Rushing TD | 465 | 0.4 | 0.4 | +0.0 | 97% |
| Interceptions | 186 | 0.6 | 0.6 | -0.0 | 96% |
| Passing TD | 186 | 0.9 | 1.0 | -0.0 | 95% |
| Passing yards | 186 | 69.7 | 72.3 | +1.4 | 79% |
Playoff odds · scoutindex-playoffs-v1
The playoff simulation is a season-long forecast, not a per-game prediction, so it can't be scored the same way. The honest check is end-of-season: did the teams it favored actually make it, and were the stated probabilities calibrated across many seasons. That needs weekly snapshots the model only started keeping this season — so this section is a placeholder until there's a full season of history to grade. Method: /methodology.
Recomputed nightly with the rest of the data. This week's predictions · Playoff picture · Methodology · Not for wagering — see Terms.