Props Edge is a model, not a tipster. Every night it projects how each probable starting pitcher and each posted batter will do in a handful of stat lines, prices those projections as fair odds, checks them against a real sportsbook's number, and posts the two it likes most. The next morning every pick is graded against the box score and the running record is published on the track record page — wins, losses and all.
The projection
For pitchers the shape is always rate × volume:
- Strikeouts — the pitcher's own strikeout rate per batter faced, blended with how often the opposing lineup strikes out against that hand, then multiplied by how many batters he faces in a typical start.
- Hits allowed — the same idea flipped: his hit rate against, the lineup's hit rate, a small ballpark nudge, times batters faced.
- Outs recorded — innings per start this season turned into outs, recent form weighted heavier, with a small adjustment for how much traffic the opponent puts on.
- Earned runs — earned runs per batter faced, pulled toward the league average when the sample is thin, scaled by the opponent and the park, priced as a count that can blow up on one bad inning rather than trickle in evenly.
For batters it projects the chance of 1+ hit and 1+ RBI from the batter's rate against that pitcher hand, adjusted for how many hits tonight's starter actually gives up and how many plate appearances the batter's spot in the order should get — split between the starter and the bullpen based on how deep the starter is expected to go.
Win probability, not a hunch
A projection of 6.4 strikeouts against a 5.5 line isn't automatically a bet. The gap is scored against 2,130 completed 2026 starts: how often has a pitcher projected this far past the line actually cleared it? That empirical hit rate — not a textbook normal curve — is the published win chance.
Fair odds vs. the real price
The win chance is converted to fair odds — the price with no house cut. Where a sharp book (Pinnacle, then Bovada) has posted a number, the pick is re-priced against that line and the edge over the book's no-vig probability is shown. The edge is capped at 7 points, because a model that thinks it has a 30-point edge on a sharp line is wrong about something. More on fair odds →
The card
The headline card is two picks — one over, one under, from whichever market shows the biggest edge — plus the day's strongest 1+ hit calls and a handful of extra plays. The parlay payout shown is a fair-odds payout; a real book takes its cut and pays less.
Grading
Every morning the previous night's box scores are pulled from the MLB Stats API and each pick is marked win, loss or push. Pushes (exact landing on a whole number) don't count. The batter model's full calibration — claimed probability vs. what actually happened across every listed batter — is on the track record page next to the pick record.
None of this is advice to place a bet. It's a public, graded experiment in pricing baseball. Shop your own book's line before doing anything with it.