Hits methodology
How the Hits model works
Per-game hit probability for each batter, surfaced as a research model — not part of the public official W-L record.
Inputs
- • Batter L5/L10/L20/season hit rate + handedness splits
- • Opposing pitcher contact quality, K%, pitch mix
- • Park hit factor
- • Weather (wind direction × stadium orientation)
- • Lineup slot + projected PA
Outputs
- • Hit probability per batter per game
- • Confidence band
- • Research-board status
- • When available: the sportsbook line and price the model compared against
Approach
Per-batter projection driven by rolling form (L5/L10/L20/season), handedness splits, opposing pitcher K% / contact quality, park hit factor, and weather.
Surfaced as research because hitter props can be highly sensitive to lineup slot, plate appearances, and price. The page is meant to explain probability and context, not overstate certainty.
Training data
Multi-year batter game logs joined to per-pitcher pitch-mix and Statcast metrics. Holdout ~10% temporal.
Calibration
The hit model is checked against thousands of settled batter games so displayed probabilities stay grounded in real hit rates.
The biggest practical question is whether the available price is worth it, so we keep this model in research while the public framing improves.
How a pick becomes official
Hit props stay research-first unless the matchup, lineup context, and sportsbook price are complete enough to evaluate cleanly.
If any of those pieces are missing, the row remains useful for exploration but is not treated as an official public pick.
Known biases and limits
- ! Hit research model is intentionally excluded from the public picks page until its public-facing performance framing is stronger.
- ! Lineup-spot dependent — a player who drops from leadoff to 8th sees a real PA reduction. We surface lineup status (projected/confirmed) but cannot pre-empt a same-day order shift.
Not for use as
- × Predicting one exact at-bat or one exact inning. These are probability models, not play-by-play fortune tellers.
- × Ignoring the sportsbook price. A model edge only matters if the available price is good enough.
- × Replacing late-breaking baseball judgment. Lineups, weather, scratches, and pitching changes can move after the page refreshes.