Build a parlay

Sports
Legs
Style
Pay at least
When
Bet types
Game lines are the bet types with a track record. Others are marked untested until their predictions prove out (see Record → Model check).
Same-game legs
Allowing them adds props from the same game (say, two players). Legs in one game can move together, so the hit chance is a rough estimate.
Highest price per leg
Minimum edge
Edge is our probability minus what the price needs to break even.

Picks

Parlays

Mixed across sports. Every leg is 55%+ likely and priced near fair.

Straight bets

The likeliest bets on the board: our model and Kalshi both rate each one 80% or better. Not guaranteed: an 85% lock still misses about 1 time in 7, and locks pay little.

Lock parlays

All locks

Games

Record

Every bet on the board goes through the same pipeline:

  1. Team ratings. Each team gets an offense and a defense rating, fit on every game over the last season or so. Recent games count more, and opponents' strength is built in.
  2. Game adjustments. Home field, rest and back-to-backs. In MLB, starting pitchers (FIP, recent starts) and park factors. For NFL and NHL, key injuries like a starting QB or goalie.
  3. Score simulation. Ratings become a full distribution of possible scores, built for each sport, which prices every moneyline, spread and total.
  4. Kalshi check. Every bet is a contract Kalshi lists, at Kalshi’s price. That price (and the sportsbook line) is blended with our model, and Kalshi’s fee is included in the break-even. A pick shows an edge only when the data genuinely disagrees with the price.
  5. Calibration. A walk-forward backtest tunes each sport so a "60%" pick hits about 60% of the time.

Grades: A clearly +EV and more likely than not. B +EV. C close to a fair price. D the price is too steep.

No model hits 100%. Parlays multiply the book's margin, and a 3-leg parlay of 70% legs still loses about two times in three. Bet only what you can afford to lose. If gambling stops being fun, call 1-800-GAMBLER.