NFL Model vs Market Disagreements

How we measure a disagreement

For every game our model gives each team a win probability. The betting market gives one too, hidden inside the odds. When the two numbers are far apart, that's a disagreement. Say the model makes the home team 60% to win and the market works out to 52%. That's an 8-point disagreement, and it's how we rank NFL games on this page.

We take the sportsbook's margin (the “vig”) out of the odds first, so we compare the model with a fair market number, not a price padded in the book's favor. The implied probability explainer walks through that math with a worked example.

What a disagreement does and doesn't tell you

It tells you the model reads the game differently from the market. It doesn't tell you who wins. A team at 60% still loses 4 games in 10, and the market is often right when the two split. We publish how the model's numbers hold up on the accuracy page, down to which probability ranges it over- and under-shoots. The methodology page shows the factors behind every number, so you can check the reasoning yourself.