back

Module 4 · Hypothesis testing the market

Where do we and the market disagree?

For each series, the model produces a Beta posterior over the single-game win probability. The sportsbook produces a moneyline pair whose de-vigged probabilities anchor the null hypothesis. A disagreement is recorded when the de-vigged market probability sits outside the equal-tailed 95% credible interval; a slate-level correction governs the false-positive rate when many series are tested at once.

de-vigging the moneylinethe rejection ruleslate-level correctionKelly fraction