Most of what a scanner flags is noise. Here is the part that is not.
We replayed a full season of bookmaker prices and asked one question: did the price we flagged beat the price the sharpest market closed at? On the whole list, no. In a narrow pocket, yes. This page shows both, with the limits first.
Confirmation on a second season (2023/24) is still pending. Until it is in, none of this is presented as proven.
Read this before the numbers
- One season. Everything below is 2024/25, six leagues. A second season (2023/24) is pending and is not in yet.
- We cleaned the data after seeing it. The cut-offs (edge at most 10 %, odds at most 5) were chosen after a first run produced absurd figures. That is a form of fitting.
- Profit is inside its error. The ROI intervals shown include zero: one season of bets cannot tell skill from luck on profit alone. Closing-line value is the evidence; profit is not.
- No account limits. The backtest assumes the price was there and the stake was accepted. Real bookmakers restrict winning accounts, which cuts real-world returns.
- Past performance is no guarantee of future results. Vigoff is an analytics tool, not betting advice. Bets can and do lose.
How we measure
- The sharp line. Pinnacle takes bets from professionals and runs on low margins, so its closing price is the best public estimate of the true chance of each outcome.
- Taking the margin off (Shin). A bookmaker's odds add up to more than 100 %. We remove that margin with the Shin method, which takes more off long shots than off favourites. On our data the three methods we tried differ very little (Premier League, under 0.001 log loss).
- Closing-line value (CLV). If you took 2.10 and the fair closing price ended at 2.00, you got a better price than the market's final view: positive CLV. Over hundreds of bets, average CLV above zero, and more than half of bets beating the close, is the standard sign that a price source has real information. It needs far fewer bets to show than profit does.
- The replay. At each snapshot at least 10 minutes before kickoff we compare every soft price with the de-vigged Pinnacle price, exactly as the live detector does, count the first time each event, bookmaker and selection shows an edge, and score it against the close. The signals are scored one by one, and also grouped so that identical prices at several books count once.
The numbers
Loading…
Signals are not independent: several bookmakers on the same network quote the same price, so one event and selection can appear as eight signals. The cluster row counts each once and is the one to trust. Its interval is wider, and it still stays above zero.
Where the signal is
Earlier is better
Prices close to kickoff are often stale ones that a bookmaker is about to fix. Signal sits a day or more out.
Bigger edges are worse
An edge of 6 % or more is usually a stale or mismatched price, not opportunity. We rank those last.
Odds
By league
Live, from our own database
Loading… If this stays empty, the live numbers are not reachable right now.
Everything here is counted from real detections; nothing is estimated. Live results track the edges Vigoff shows now against the actual closing line, and the sample is young.
Check it on your own bets
Free to use in your browser, no card. See the fair price, log your bets and watch your own closing-line value grow, then judge for yourself.
Try it free