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=nba

Overall: N = 12 · mean predicted 55.8% · actual win rate 50.0% · gap -5.8pp · Brier 0.303 · log-loss 0.805
Calibration by predicted-probability bucket.
Predicted-prob bucket N Mean predicted Actual win rate Gap (actual − predicted) Brier
<50% 2 48.5% 100.0% +51.5pp 0.266
50-55% 6 52.0% 50.0% -2.0pp 0.256
55-60% 1 57.4% 100.0% +42.6pp 0.181
60-65% 1 63.6% 0.0% -63.6pp 0.404
65-70% 1 68.0% 0.0% -68.0pp 0.462
70-75% 1 72.0% 0.0% -72.0pp 0.518