Player Props -- Today's Edges
Player props are the most inefficient market in sports betting. Books can't sharpen 1,000+ markets a day with the same precision they sharpen the moneyline. Our prop model attacks where the books are weakest.
Lines from DraftKings (your takeable book). PrizePicks & Sleeper lines are coming -- both run a different DFS pick-em model.
Model v1.1: pitcher_strikeouts, batter_home_runs, batter_hits, batter_total_bases. Projections use season stats + opponent-team K-rate + bench-player heuristic. Calibration loop is live -- each day's outcomes feed back into tomorrow's bias correction (see Learning). Low-confidence plays filtered to SKIP.
| Player | Matchup | Market | DK Line | Model Proj | DK O / U | Model % | Edge | Side |
|---|---|---|---|---|---|---|---|---|
| Loading props... | ||||||||
Top 8 player prop edges by magnitude.
For each player prop we project a full distribution, not a point estimate. Inputs include:
- Rolling 30-day stats regressed toward career baseline with shrinkage
- Opposing pitcher arsenal โ projected pitch mix and the hitter's xwOBA vs. each pitch
- Park & weather adjustments specific to the prop (HR props weight wind more heavily)
- Lineup position projection (top-of-order = more PAs)
- Game total / pace impact (high totals create more counting-stat chances)
- Pull/spray tendencies vs. defensive alignment
Output is P(prop) plus a Poisson-style or gamma distribution from which we compute the fair price.
Loading top plays...
Props are different from sides/totals. Three rules:
- Shop hard. Prop lines and prices vary book-to-book more than any other market. A 6.5 Ks line at โ115 vs. a 7.5 line at +100 across the same player can swing your edge 8โ12%.
- Beware juice creep. Books charge more vig on props (often 10โ15% built-in) than on standard markets (~4.5%). Your model probability needs to clear vig + edge to be a play.
- Mix-and-match correlations. Skip same-game-parlays. Books bake their projections into SGP pricing and the implied juice is brutal. Bet legs independently, sized via Kelly.