Betting research

The Bozo Cave

Tools for doing the work yourself: what a player has actually done against a number, what a price really implies once the vig is stripped out, and whether an angle survives contact with out-of-sample data.

Read this first

Across 5,215 games since 2007 the closing spread misses the real margin by 0.02 points on average and the closing total by 0.45. I tested 72 angles against that history: nine beat the break-even rate, none survived correction for multiple comparisons, and of the eight best angles found in 2007–2018, one was still above water in 2019–2026. See the working →

So this page has no picks in it. Sides and totals are close to a coin flip after the vig; the value here is the arithmetic and the player-level history, which is where a careful person can still find something.

Pick a game. You get the spread and total in plain English, then every player prop the book is offering, sorted by how strongly that player's own history disagrees with the number.