PropsBot’s MLB player prop models now incorporate opposing pitcher strength, batter-vs-pitcher history, and left/right platoon splits to deliver more accurate AI-driven predictions across all batter prop markets.
Added 3 new MLB feature sets that give batter prop models awareness of the opposing pitcher matchup.
Models now know how good the opposing starting pitcher is. Features include ERA, WHIP, strikeout rate, walk rate,
whiff rate, fastball velocity, and how rested the pitcher is. A batter facing a dominant ace will now be evaluated differently
than one facing a struggling reliever.
Models now account for the well-known left/right matchup advantage in baseball. Each batter’s historical performance is
tracked separately against left-handed and right-handed pitchers, capturing differences in batting average, power, plate
discipline, and strikeout tendencies.
Models now factor in how a specific batter has performed against the specific pitcher he’s facing. Some batters simply
own certain pitchers (and vice versa) — the models can now pick up on those individual matchup edges.
All 12 batter markets now receive the new features: Hits, Runs, RBIs, Home Runs, Total Bases, Walks, Strikeouts, Stolen Bases, Doubles, Triples, Singles, and Hits+Runs+RBIs.
Pitcher markets (Strikeouts, Hits Allowed, Walks, Earned Runs, Outs, Record a Win) are unaffected — these features describe the opposing pitcher, not the pitcher’s own performance.
Previous MLB models had no information about who was on the mound. The opposing pitcher is one of the strongest predictors of batter outcomes, and left/right matchup advantages are among the most established effects in baseball.
This brings MLB’s feature depth in line with our other sports (NFL has opponent defensive effects, NBA has usage and defense-vs-position, NHL has goalie strength).
