Every decimal odds price is a probability claim: 2.50 says the book thinks the team wins 40% of the time. Edge is the gap between that claim and your own estimate of the true probability, and one formula produces it: edge = (your true probability × decimal odds) − 1, read as a percentage. It is the number the whole odds-math chain exists to produce, and it is a long-run expectation rather than a promise about the next bet.

How to convert odds to a probability

The implied probability of a price is the break-even win rate the price assumes. For decimal odds the formula is:

Implied prob = 1 / decimal odds

At 2.50: 1 / 2.50 = 0.40 = 40.0%. You need to win 40% of the time just to break even.

Fractional odds (common in UK markets) convert the same way. Fractional odds of 3/2 mean you win 3 units for every 2 staked, so decimal equivalent = (3/2) + 1 = 2.50, and implied prob = 40.0%.

American odds split into two cases. A positive line (+150) means profit per $100 staked: implied prob = 100 / (150 + 100) = 40.0%. A negative line (−200) means stake required to win $100: implied prob = 200 / (200 + 100) = 66.7%.

All three formats produce the same implied probability when the market is the same.

Why the implied probabilities add up past 100%

Take a two-way market: Team A at 1.80, Team B at 2.10.

Implied prob A = 1 / 1.80 = 55.6%
Implied prob B = 1 / 2.10 = 47.6%
Total          = 103.2%

The total exceeds 100% by 3.2 percentage points. That excess is the overround: the percentage above 100% that the book's implied probabilities sum to. It's the bookmaker's built-in margin. A fair market would sum to exactly 100%.

You cannot treat either raw implied probability as the true probability of the outcome. The overround inflates both.

How to strip the margin out

Devigging is the process of removing the overround to recover a fair probability for each outcome. The proportional (multiplicative) method divides each implied probability by the total:

Fair prob A = (1 / 1.80) / 1.032 = 55.6% / 1.032 = 53.9%
Fair prob B = (1 / 2.10) / 1.032 = 47.6% / 1.032 = 46.1%
Total       = 100.0%  ✓

Two other methods exist. The additive method subtracts an equal share of the overround from each outcome's implied probability, treating all outcomes as equally over-priced.

The Shin method (named after the economist who derived it) accounts for the possibility of insider trading and skews the adjustment. It tends to produce more accurate fair probabilities on large asymmetric markets.

For most two-way markets the proportional method is accurate enough and easy to apply without a calculator.

The edge formula

Edge = (true_probability × decimal_odds) − 1   ×100 for a percentage

Two inputs, one output. The first is the decimal odds you can actually get on the bet. The second is your estimate of the true probability: a devigged/fair probability with the bookmaker's margin already removed, not the raw implied probability of the price you are betting.

If your true probability exceeds the break-even implied probability, (true_prob × odds) clears 1 and the leftover is your edge. Edge greater than 0 is exactly the same condition as "true probability greater than implied probability."

An equivalent form that sometimes appears in calculators: edge = (decimal odds / your fair odds) − 1, where fair odds = 1 / true probability. Both forms produce the same number.

A full worked example

Step 1: the price. A book offers a team at decimal odds 2.50. The implied break-even probability is:

Implied prob = 1 / 2.50 = 0.40 = 40.0%

Your devigged estimate puts the team's real chance at 45%. Because 45% exceeds 40%, value exists.

The edge:

Edge = (0.45 × 2.50) − 1
     = 1.125 − 1
     = 0.125 = +12.5%

That's $12.50 expected profit per $100 staked over many repetitions of this bet.

Sanity check in dollars. On a $20 stake at 2.50, a win returns $30 profit:

EV = (0.45 × $30) − (0.55 × $20)
   = $13.50 − $11.00
   = +$2.50

$2.50 / $20 = +12.5%  ✓ matches the edge

Flip the estimate. If your true probability were only 35%:

Edge = (0.35 × 2.50) − 1 = 0.875 − 1 = −12.5%

Negative edge. Skip it.

The formula also works for thin edges. A coin-toss priced at 2.10 gives: (0.50 × 2.10) − 1 = 0.05 = +5%. A sharper case: bettable odds 2.15 against fair odds 2.123 → 2.15 / 2.123 − 1 = +1.27%.

Real edge, but razor thin.

What edge size is worth betting

Not every positive edge is worth acting on. An edge threshold is the minimum number you require before placing a bet.

A 2–3% threshold is a reasonable starting point while you build a track record. Bets under roughly 2% tend to be break-even or losing once you factor in your own estimation error.

Practitioners commonly cite 2–5%+ as a working range, but the right cutoff depends on your estimate quality and bet volume. Treat it as tunable rather than fixed.

Treat large claimed edges with skepticism. On efficient markets, a "huge" edge usually means your probability estimate is off, not that you found free money.

Variance punishes small edges hard. A 0.5% edge is $0.50 of expected profit per $100 staked, and a normal losing run can wipe out hundreds of those before the edge pays off.

The formula is exact; your probability estimate is not. A thin edge can sit entirely inside your own estimation error. A threshold above 0% provides a margin of safety against that.

Set it too high and you barely get bets down. Too low and variance swallows the edge before the sample grows large enough to see it.

Why edge only shows up across many bets

A positive edge is a long-run average. On a single bet, variance dominates completely, making the edge invisible at n = 1.

Losing streaks happen even with a genuine edge. After 1,000 bets at 5% value, there is still roughly a 25% chance of hitting 10 losses in a row at some point. A losing streak doesn't mean the edge is gone.

Sample size is what converts edge into profit you can actually see. RebelBetting's user data, drawn from a self-selected base of active customers: among users placing 100+ bets per month, 79.7% were profitable.

At 500+ bets per month that figure reached 90.0%. At 1,000+ per month, about 95% of months were profitable.

The smaller the edge, the larger the sample you need before it shows through the noise. A 1–2% edge needs far more bets than a 10% edge to become statistically visible.

The full workflow

Price → implied prob (1 / odds) → devig → true prob → edge formula → bet if edge clears threshold

Each step feeds the next. Implied probability supplies the break-even number the formula uses to define "value." You cannot treat raw implied probability as true probability because it includes the book's cut. Devigging removes that cut and produces the true probability the formula needs.

None of it saves you if the estimate feeding the formula is loose. Below about 2% edge, ordinary estimation error can swallow the whole expected value, which is why the threshold sits above zero and why the quality of your probability ends up mattering more than the arithmetic wrapped around it.