The short answer
- In general betting terms, edge is the percentage gap between an estimate of the true probability of an outcome and the probability a specific bookmaker's odds imply.
- On FootyMetrics' Odds Outliers tool, edge does not come from a prediction model. It compares one bookmaker's odds against the average odds across the other bookmakers covering that same prop.
- A bigger edge is a bigger odds gap, not a bigger sure thing. It still needs enough bets over time before it shows up as real profit rather than noise.
- Any edge figure is only as good as the estimate feeding it, whether that estimate is a model's probability or, on Odds Outliers, the average of the other bookmakers.
Bettors use the word loosely. The Odds Outliers figure is precise, and narrower than a casual reading suggests.
What edge means in betting generally
Edge is the gap between odds and a better estimate of what those odds should be, turned into a percentage. The usual way it is set up is this: you, or a model, form an estimate of the true probability of an outcome, then compare it against the probability implied by a specific bookmaker’s odds. If your estimate is 58% and the bookmaker’s odds imply 55%, the edge is 3 percentage points in your favour. This is the same comparison what are fair odds covers in more depth, and it is directly tied to what is expected value: a positive edge is what makes a bet positive EV in the first place.
Edge does not always have to be measured against your own model. Some tools measure it against a consensus, the average odds across several bookmakers, on the idea that a wide, liquid market of odds is a decent stand-in for the true probability when no single sharp source is available. That consensus approach is exactly how FootyMetrics’ own Odds Outliers tool works, and it is worth being precise about, since it is a different method to a model-versus-market edge.
How FootyMetrics’ Odds Outliers figure actually works
FootyMetrics’ Odds Outliers tool takes the live odds for a prop from every bookmaker it tracks, works out the average, and shows the ones where a single bookmaker is a long way off the rest. There is no model behind it, just one bookmaker’s odds against the others, updated every minute. The biggest disagreement between a single book and the market average is shown as the edge, and every row needs at least two bookmakers offering odds on the same selection so the average stays meaningful.
That is a market-versus-market comparison, not a model-versus-market one. The outlier comes from the odds alone, with no model involved. Our model's probability is shown in its own column as a second check. A +8% edge means one bookmaker’s odds are 8% higher than the average of the other bookmakers offering odds on that same prop. The bigger the gap, the further out of step that one book is with everyone else.
| Bookmaker | Built-in substitution protection |
|---|---|
| Bet365 | Sub On Play On |
| Paddy Power | Super Sub |
| Sky Bet | Super Sub |
| ComeOn | Super Sub |
The exact wording of each feature lives in the bookmaker’s own rules. Sky Bet publishes its football market rules on a single hub, the place to check how its version treats a player who is substituted.
A bigger edge number is not automatically a bigger sure thing
A large edge percentage describes the size of an odds gap, nothing more. It still needs volume, meaning enough bets placed over time, before that gap turns into real, measurable profit rather than a run of results that could have gone either way by chance. A single bet on a wide-edge prop can still lose, because the underlying event is still uncertain, in exactly the same way a positive EV bet can still lose on any given try. See what is expected value for that in full. The case for acting on edge repeatedly is a long-run one, built up across enough attempts for the average to look like the average rather than a small, noisy sample.
The estimate behind the number
Every edge number depends entirely on the quality of the estimate it is measured against. For a model-based edge, that means the model’s own probability has to be a genuinely good read on the game, not just a number that happens to disagree with the market. For a consensus-based edge like Odds Outliers, it means the average of the other bookmakers has to be a reasonable stand-in for the true probability, which weakens if only a couple of thin, uncorrelated books make up that average, or if every book in the average shares the same blind spot. Get the underlying estimate wrong, whichever kind it is, and the edge percentage measures a gap against a bad benchmark, not against the truth. This is the same limitation covered on what is expected value: the maths is only ever as trustworthy as what you fed into it.