Winner / Run line / Total: each team's strength comes from runs scored vs. allowed all season (better than win-loss, which is noisy in close games). The model compares the two teams, gives the home team a small bump (~54%), and folds in tonight's starting pitcher's ERA — regressed toward league average so an early-season fluke doesn't break it. From each team's expected runs it derives the chance to win, to win by 2+, and the total runs.
Strikeouts: built from the starter's strikeout rate per batter and his typical batters-faced per start, adjusted by how much the opposing lineup strikes out. Early-season rates are regressed toward league average.
What it ignores: bullpen, injuries, weather, park, and — for batter props — the day's lineup. It also can't price individual batter bets (home runs, hits, total bases) from team stats, so those aren't included. Because Kalshi's prices already reflect sharp money, a big edge usually means the model is missing something (often the pitcher) rather than free money — treat every edge as a question to investigate.
Live data from the free MLB Stats API. Kalshi prices are typed in by you. Everything runs in your browser; nothing is sent anywhere.