For every bucket of predicted probability, what did the model actually hit? A well-calibrated model has actual win rate ≈ predicted prob. A negative gap means the model is overconfident in that bucket (dangerous); positive gap means it's underpredicting (safe). Rows turn red when |gap| > 5pp AND N ≥ 10.
Combined "all sports" is rarely meaningful — sports differ in market efficiency, signal availability, and base rates. Use the chips to drill into a single sport. N<50 (red) means the calibration is brittle; N≥200 (green) is trustworthy.
Filter: window=90d · sport=ncaaf
| Predicted-prob bucket | N | Mean predicted | Actual win rate | Gap (actual − predicted) | Brier |
|---|---|---|---|---|---|
| <50% | 119 | 49.8% | 44.5% | -5.3pp | 0.250 |
| 50-55% | 432 | 52.2% | 55.8% | +3.6pp | 0.247 |
| 55-60% | 152 | 57.0% | 61.2% | +4.2pp | 0.240 |
| 60-65% | 36 | 61.8% | 69.4% | +7.7pp | 0.219 |
| 65-70% | 25 | 67.6% | 60.0% | -7.6pp | 0.238 |
| 70-75% | 21 | 73.0% | 90.5% | +17.5pp | 0.116 |
| 75-80% | 36 | 77.3% | 77.8% | +0.5pp | 0.173 |
| 80-90% | 40 | 84.5% | 92.5% | +8.0pp | 0.073 |
| 90%+ | 99 | 94.3% | 99.0% | +4.7pp | 0.012 |