How it works

We use the same free, public numbers anyone can pull up: pitching and hitting stats, batting orders, box scores. No insider info, no leaked data. If you wanted to, you could look at the exact same numbers we do.

For each game the model projects a player's stat, like a pitcher's strikeouts. It comes from two simple things: how good the player has been (their rate) and how much chance they'll get (their opportunity, like how many batters they'll face). Rate times opportunity gives a number we can stand behind.

The market sets a number for most of these stats, and that number is sharp, it soaks up a lot of information. We compare our prediction to theirs to see how the market reads the same public stats, and how well the model lines up. We track every difference so we can learn where the model is strong, where it is off, and keep improving it.

Every prediction goes on the record and gets graded against the closing number, the last and sharpest one before the game starts. We show the wins and the losses. We also track whether the market moved toward our side after we made the call, which is the truest test of whether a prediction was actually sharp.