HR Matchup
Today's home run prop edges — batter power blend vs. arsenal matchup vs. park factor, ranked by edge over the book.
How HR Matchup Works
What it predicts
For each batter in today's slate who has a confirmed or predicted lineup spot, we estimate the probability they hit at least one home run. Predictions are sorted by edge — how much our probability exceeds the sportsbook's implied probability on the YES side of the HR prop.
Power Blend (60% Barrel / 40% xwOBA)
Each batter's intrinsic HR strength is expressed as a league-relative multiplier (1.00 = league average). The blend combines Barrel% (the cleanest predictor of HR rate) and xwOBA (broader contact quality). Above 1.30 indicates elite power. This is Marcel-shrunk toward the league mean, reducing noise for players with fewer PAs.
Arsenal matchup
For every pitch type the starting pitcher throws, we weight the batter's Barrel% and xwOBA against that pitch type by the pitcher's pitch-mix percentages. This produces a matchup multiplier: above 1.0 means the pitcher's arsenal plays into the batter's strengths. Small pitch-type samples (below 20 PAs) fall back to the batter's overall season Barrel%.
Park factor
A HR-specific 3-year rolling park factor adjusts the per-PA probability for the venue. Hitter-friendly parks (Coors Field, Great American Ball Park) push the multiplier above 1.0; pitcher's parks push it below.
Edge vs. the book
The YES/NO HR prop is devigged to a fair implied probability. Edge = our P(HR≥1) − implied probability. Positive edge means we think the batter is more likely to homer than the book's price suggests. Default book: Hard Rock Bet.
Qualifying thresholds
A batter appears in the Top Edges hero section only when edge ≥ 3.0% AND P(HR≥1) ≥ 10.0%. All batters with a lineup spot appear in the full-slate table regardless of edge; non-qualifying rows are dimmed.
Past-date grading
Use the date picker to browse historical predictions. Card borders and row backgrounds indicate outcomes: green = batter homered (hit), red = no HR (miss), yellow = DNP / lineup scratch. Past predictions are never rewritten.
Limitations
The model does not yet account for weather, bullpen usage, platoon splits inside core math, or real-time lineup changes after the 6:45 AM CT prediction run. Predicted lineups (not yet confirmed) are used when confirmed lineups aren't available and may occasionally be inaccurate.