What it predicts

Home runs, away runs, and total probabilities

Separate home- and away-run estimators feed a calibrated discrete joint score distribution with ties removed. That distribution prices over, under, and push for the offered line.

prospective shadow only

Recorded component metadata

Some component metadata is absent from the published evidence. The verified system identifiers are on the overview; missing fields are not inferred.

Real-money authorized
No

Evaluation

Forecast quality against the market

Historical out-of-sample evidence for the historical totals cohort (retrosheet-game-date-bounded-fielding-2026-v16.0.0). Brier and log loss are forecast-error scores; lower is better for both.

Covered games
11,196
Model Brier
0.2503
Market Brier
0.2496
Model log loss
0.6937
Market log loss
0.6924

Open prospective live evaluation

Main totals and alternate lines

Main-total forecast quality and alternate-line calibration have separate evidence. An alternate quote is not an independent game, and a displayed ladder does not establish a validated betting model.

Review main-total scores and the alternate-line calibration gate

Per-game run contributions and global totals importance are not published yet. Inspect the game notebook for recorded run estimates and available lines.