Model calibration audit

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.

Calibration cohort by sport (click to filter):
All sports MLB N=3,582TENNIS_WTA N=909TENNIS_ATP N=873MMA_MIXED_MARTIAL_ARTS N=388NFL_PRESEASON N=145LALIGA N=116NCAAF N=95SERIEA N=87EPL N=86LIGUE1 N=84UCL N=61BUNDESLIGA N=53NBA N=12NCAAB N=11NHL N=9AMERICANFOOTBALL_NFL N=2

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=mlb

Overall: N = 3,582 · mean predicted 54.9% · actual win rate 49.7% · gap -5.2pp · Brier 0.251 · log-loss 0.695
Calibration by predicted-probability bucket.
Predicted-prob bucket N Mean predicted Actual win rate Gap (actual − predicted) Brier
<50% 594 45.5% 40.9% -4.6pp 0.244
50-55% 1340 52.2% 48.6% -3.6pp 0.251
55-60% 751 57.4% 55.3% -2.1pp 0.248
60-65% 674 61.6% 49.9% -11.7pp 0.263
65-70% 198 66.8% 58.1% -8.8pp 0.252
70-75% 14 71.7% 78.6% +6.8pp 0.170
75-80% 4 76.3% 75.0% -1.3pp 0.194
80-90% 6 84.8% 100.0% +15.2pp 0.023
90%+ 1 90.9% 100.0% +9.1pp 0.008