Calibration

Odds bands

Expected versus observed rates by analytical band. Each band shows a 95% Wilson interval on the observed rate and a two-sided proportion z-test against the expected rate; the header reports a Pearson chi-square goodness-of-fit across every band. Differences are calibration deviations in synthetic samples — not betting tips. Patterns must survive re-seeding and sample-size checks before they are noteworthy.

Dataset synthetic-v1 · generated 2026-07-26T03:36:21.368Z

χ² 17.07 · df 9approx p 0.0473 significant bandsSynthetic calibration · not a wagering signal
Expected (margin-adjusted, price-implied) vs observed win rate per band, aggregated across all sport adapters.
BandEventsExpectedObserved95% WilsonDifferenceSignificance
+200 to +2998,00025.37%26.51%25.56% 27.49%+1.14%p 0.019
+500 to +79918,00012.82%12.54%12.07% 13.04%-0.27%within noise
+800 to +11996,0008.68%9.52%8.80% 10.29%+0.84%p 0.021
+1200 to +19992,0004.37%3.55%2.82% 4.45%-0.82%within noise
+2000 to +49994,0003.05%2.83%2.36% 3.39%-0.22%within noise
+110 to +1998,00037.49%37.30%36.25% 38.37%-0.19%within noise
-109 to +1094,00046.54%47.10%45.56% 48.65%+0.56%within noise
-149 to -1108,00051.73%51.45%50.35% 52.54%-0.28%within noise
-199 to -1504,00061.98%62.28%60.76% 63.76%+0.29%within noise
-499 to -3002,00072.73%69.55%67.50% 71.53%-3.18%p 0.001

Real-world use case

  • Reload a sport's statistics page with a different batch seed and check whether a flagged band stays significant — one-off flags are usually noise.
  • Compare a favourite band's tight Wilson interval against a longshot band's wide one to see why long prices need far more events before a deviation means anything.
More in the breakdowns guide →