NBA Model Accuracy
Out-of-sample backtests of the per-game projection engine — projections vs. what actually happened, compared to the Marcel baseline
Per-game projections
Beat the Marcel baseline by 4% on per-game accuracy · p10–p90 coverage 66% (2025-26 backtest, 2,148 preds)
Player Accuracy
per-game rate vs naive · 2,148 predsskill 4%p10–p90 cov 66%
| Stat | N | Model/g | Base/g | Skill | Cov |
|---|---|---|---|---|---|
| Scoring | |||||
| Points | 358 | 2.53 | 2.68 | +6% | 64% |
| 3PM | 358 | 0.35 | 0.36 | +5% | 66% |
| Playmaking | |||||
| Assists | 358 | 0.7 | 0.73 | +5% | 63% |
| Defense & Boards | |||||
| Rebounds | 358 | 0.86 | 0.89 | +4% | 68% |
| Steals | 358 | 0.2 | 0.2 | -0% | 65% |
| Blocks | 358 | 0.14 | 0.15 | +5% | 67% |
Team Game Model vs Home-Court Baseline
backtest
| Season | Games | Brier | Base | Acc | Margin MAE | Total MAE |
|---|---|---|---|---|---|---|
| 2025-26 | 1,161 | 0.2046 | 0.2474 | 69.0% | 11.55 | 16.2 |
| 2024-25 | 1,162 | 0.2131 | 0.2480 | 66.3% | 11.35 | 16.0 |
| 2023-24 | 1,153 | 0.2133 | 0.2482 | 64.4% | 11.21 | 15.6 |
| 2022-23 | 1,157 | 0.2317 | 0.2439 | 62.5% | 10.38 | 15.0 |
| 2021-22 | 1,159 | 0.2272 | 0.2483 | 63.6% | 11.30 | 17.9 |
| 2020-21 | 1,034 | 0.2300 | 0.2489 | 61.2% | 11.06 | 24.3 |
| 2019-20 | 890 | 0.2151 | 0.2479 | 66.2% | 10.44 | 15.7 |
| Overall | 7,716 | 0.2193 | 0.2475 | 64.7% | 11.06 | 17.2 |
Opponent-adjusted point-differential ratings (SRS) → predicted margin, win probability, and total, scored walk-forward (ratings built only from games played before each matchup — no leakage). A lower Brier than the home-court baseline means real predictive skill. The model never sees the betting line.