CWS games

Model probability next to the sportsbook line. Today 0, next 72 hours 0, completed 0.

Rating engine

Round

Status

Loading games…

Model, away and home: Our win probability for each team, frozen before the game.
Model line: Predicted home margin in runs; negative means home is favored.
Run line, total, ML: DraftKings (Pinnacle for non-P5 matchups). Run line is the home side, with home and away prices.
Pick results: Right or wrong for the model's moneyline pick and its spread pick, against the final score.
Gaps: Pregame only: model minus market, in percentage points (moneyline) and runs (run line).

How this works

Model probability is a per-game logistic on the rating difference, with home-field advantage at the regional host site and neutral elsewhere. We do not publish a totals model. Lines come from The Odds API.

How the numbers are produced. A 50,000-iteration Monte Carlo of the NCAA tournament under the PEAR Ratings rating engine. Each sim plays out 16 regionals (double-elimination), 8 super-regionals (best-of-3), then the Omaha 8-team double-elim + best-of-3 final. Win-CWS is the fraction of sims a team holds up the trophy.

Margin of error. 50,000 sims gives ±0.4pp 95% CI on every champion probability. Anything below 0.5% is cosmetic, we render “—” rather than publishing noise. The bootstrap CIs shown on top-of-field teams are per-1k-sim blocks.

Known biases. Our rating likely over-weights SEC and ACC conference quality, Georgia Tech, Mississippi State, and Alabama may be inflated; Oregon State and Oregon are probably undervalued. Pitcher matchups are not modeled, which means real-world upset rates are likely a floor, not a ceiling.

What the shading means. Row background is a 6-step gradient keyed to win-regional probability, darker means more likely to escape. Once games start, eliminated teams get a strikethrough overlay; the eventual champion gets a solid frame.