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=mma_mixed_martial_arts
| Predicted-prob bucket | N | Mean predicted | Actual win rate | Gap (actual − predicted) | Brier |
|---|---|---|---|---|---|
| <50% | 140 | 40.2% | 35.7% | -4.5pp | 0.228 |
| 50-55% | 83 | 52.5% | 61.4% | +9.0pp | 0.243 |
| 55-60% | 54 | 57.7% | 53.7% | -4.0pp | 0.247 |
| 60-65% | 61 | 62.1% | 62.3% | +0.2pp | 0.237 |
| 65-70% | 62 | 67.2% | 77.4% | +10.2pp | 0.184 |
| 70-75% | 32 | 72.1% | 78.1% | +6.1pp | 0.178 |
| 75-80% | 37 | 77.2% | 78.4% | +1.2pp | 0.168 |
| 80-90% | 41 | 83.4% | 80.5% | -2.9pp | 0.157 |
| 90%+ | 12 | 93.7% | 91.7% | -2.1pp | 0.075 |