How does sim-grading work?
Sim-grading is a way of pricing a bet that never asks a sportsbook what the odds should be. The game is simulated thousands of times, play by play, using the day's real lineups, starters and weather. How often the bet wins across those simulated games is its true probability. Only then does the price enter: the true probability is set against the probability the price implies, and the gap between the two is the expected value. If the price implies a better chance than the simulations found, the bet is negative EV.
Written by Jesse, NegativeEV. Last updated 12 August 2026.
- Sim-grading builds its own probability for a bet by playing the game out, instead of reading the market.
- Every game on the slate is simulated thousands of times, play by play, by machine-learning models trained on years of that sport's real results.
- The bet's true probability is simply how often it won across those simulated games.
- The price enters last. True probability against implied probability gives the expected value.
- The sportsbook's number never goes into the simulation, so the answer is free to disagree with the market.
- MLB, WNBA, PGA, and ATP bets can be checked. 10 free checks a day, unlimited signed in.
How does sim-grading price a bet?
Three steps, in this order.
- Simulate the game thousands of times before it starts, one play at a time.
- Count how often the bet won across those games. That count is its true probability.
- Compare that to the probability the price implies. The gap is the expected value.
The order is the whole point. The price is used once, at the end, and never as an input.
Where does the true probability come from?
From the simulated games themselves.
Each sport has its own machine-learning models, trained on years of that sport's results. The models drive every play of every simulated game. What comes out is a full spread of outcomes for the game and for the players in it. A bet's true probability is how often it landed inside that spread.
Real lineups, starters and weather go in. The sportsbook's price does not.
How is sim-grading different from comparing sportsbooks?
Odds tools read the market. Sim-grading builds its own number.
| Method | Where its probability comes from | What it can miss |
|---|---|---|
| Line shopping | The friendliest of several sportsbook prices | Still the market's opinion, just the best-priced version of it |
| Consensus and odds screens | The average across a group of books | If the books lean the same way, the consensus leans with them |
| Sim-grading | Thousands of simulated games, played out one play at a time | Anything that happens after a slate is simulated |
An odds screen can find the book that is out of step with the rest. It cannot tell whether the whole board is wrong. A probability built from a simulated game can.
What does a sim-graded check return?
A true win probability, and the expected value at the quoted price.
Paste a bet or upload a screenshot of the slip. Moneylines, totals, player props and full parlays all grade the same way. On a parlay, every leg is graded and then the whole ticket is priced. MLB, WNBA, PGA, and ATP are covered. 10 free checks a day, unlimited signed in.
This is a second opinion on a bet somebody already has in mind. There is no pick list and no claimed record. The number is the output, and what to do with it is the user's call.
What can sim-grading not do?
Price a bet it cannot match to a simulated game.
A bet on a game the simulator has not run comes back ungraded rather than guessed at, and a parlay with an ungraded leg is priced on the legs that remain.
Background: what negative EV means.
Related reads: what happens when a bet is checked and how accurate a simulator can be.