What an "edge" actually measures, why ours are leans rather than picks, and how the star rating condenses it.
Edge = model % minus market %
For every spread, total and moneyline we compute the model's probability from the unblended projection and compare it with the de-vigged market probability. The difference, in percentage points, is the edge. Positive edge means the model thinks the side is underpriced.
Why leans, never locks
Our game model tracks the closing line; it does not beat it. Backtests show small edges (inside about eight points) behave best — a huge edge usually means the model is missing news, not that the market is wrong. That is why rows past the band are tiered "longshot" and framed accordingly.
The star rating
Stars answer one question: how likely is this pick to land at this price? First the edge is shrunk to what our graded record supports — about three points of probability, never more, and nothing at all once the gap passes the playable band — then the pick is rated on that calibrated likelihood and the value left at the quoted price. Five stars is a likely winner at a better-than-fair price; zero stars means no value at this price, not a prediction of defeat. A big nominal edge earns nothing extra: the record says it is usually the model missing news.
The same formula runs on every Statz sport, so a four-star lean reads identically on football, cricket and here.
Expected value
EV converts edge at the book's actual price into expected profit per dollar staked. It is the sorting key inside the playable tier — and the reason the best book matters: the same side at a better price is simply a better bet.
Put it to work
- Best Bets The week's edges, tiered and star-rated.