Every prop that hits our +EV screen is captured at the first price we displayed — one record per book per game — and graded against the official box score the next morning. Wins, losses, and pushes settle on the line as shown; if a player doesn't play, the bet is void and the stake is returned, which matches how sportsbooks handle DNPs. Once a bet is recorded it is immutable: no deletions, no re-grades, no swapping in better odds after the fact.

Because that's the honest price. First-seen odds are what a bettor refreshing the page would actually have gotten, and they can't be gamed — recording the best price of the day, or re-pricing after line movement, would flatter the record. If a line moves after we log it, the record keeps the original price, for better or worse.

No. The simulator never looks at a whole day's board and picks the best of it in hindsight. It walks through the day in the order bets were first seen, in half-hour steps, and locks in the best of what was visible at each step — what a bettor refreshing the page at that time could actually have placed. A price it skips doesn't come back, and money it holds for later games can go unspent if nothing qualifies. That costs some return versus a hindsight pick, and it's the only version of the record a reader could have matched.

Because that's where the graded data says the edge is repeatable. Across the ledger, bets below 3% EV have run near break-even at flat stakes. On price, the edge is concentrated from short favorites up through small plus-money lines around +101 to +124, and it stops there: from +125 on, flagged bets have run flat at flat stakes — the well-documented favorite–longshot bias. The defaults concentrate the view on the reliable range, but every filter is adjustable, so you can inspect any slice of the ledger — including the ones that don't flatter us.

The bets and their grades are real and immutable — every one was publicly flagged on our +EV screens before its game started. The profit figures are a simulation over that ledger: stakes are modeled with quarter-Kelly or flat sizing under a 25% daily budget and 5% per-bet cap, because no one places every flagged bet. The flat-units record shown on every bet lets you check the raw results independent of any staking plan.

Six sports — NFL, NBA, MLB, NHL, WNBA, and NCAAF — across eight sportsbooks: DraftKings, FanDuel, Pinnacle, BetMGM, Caesars, BetRivers, Fliff, and Fanatics. The track record covers player props only; prediction-market and exchange EV bets (Polymarket, Kalshi, Novig) appear on the sport pages but are not part of this ledger.

A bet stays pending until the official game logs post — usually the next morning, though NFL and NCAAF results can take until Tuesday. Void means the player didn't play, so the stake is returned, exactly as a sportsbook would settle it. The honesty strip above the bet list shows won, lost, pushed, void, and pending counts for whatever window you've selected, so you can always see how much of the record is settled.

No — the grading is deliberately conservative. Many sportsbooks now void or refund player props when a player leaves a game early or plays limited minutes (“injury protection”); our ledger only voids a true DNP, so a bet a book would have refunded counts as a loss here if it would have lost. We also record every bet at the standard board price, never at promos or odds boosts. Both choices understate results on the bets shown, so outcomes with those protections and boosts applied may run modestly better — though no real bettor catches every first-seen price the ledger does.

This page exists to answer that question with data instead of a sales pitch. In theory, consistently taking positive-EV prices wins over a large enough sample — but variance rules short windows, and most sites only show you their winners. Here you can see every bet we flagged, graded against reality, over any window from a week to a year, with losing days left in. Past results are not a promise of future ones; bet responsibly.