Implied probability and bookmaker margin
Convert decimal odds into implied probabilities, measure the bookmaker's margin with the book sum, and see why prices that add up to more than 100% are a built-in cost to anyone who bets.
Learning objectives
- Convert decimal odds into raw implied probabilities with π = 1/o.
- Compute the book sum, overround and payout rate for a two-way or three-way market.
- Explain why raw implied probabilities overstate fair probabilities when the book sums to more than 100%.
- Show with numbers that backing every outcome in proportion loses exactly the margin, whatever the result.
From a price to a probability
Decimal odds state the total return per unit staked, stake included. At 2.10 a winning 1-unit stake returns 2.10, a profit of 1.10. The break-even probability is the win rate at which that price neither gains nor loses on average, and it is simply the reciprocal of the odds.
π = 1 / o| Decimal odds | Profit per unit if it wins | Raw implied probability |
|---|---|---|
| 1.50 | 0.50 | 66.67% |
| 2.00 | 1.00 | 50.00% |
| 2.10 | 1.10 | 47.62% |
| 3.40 | 2.40 | 29.41% |
| 3.60 | 2.60 | 27.78% |
| 10.00 | 9.00 | 10.00% |
Each row answers one question: how often must this pick win for the price to be fair to me? It does not yet answer the different question how likely does the market think this outcome is? For that we need the margin.
Book sum, overround and payout rate
Book sum β = Σ 1/oᵢ Overround = β − 1 Payout rate = 1/βTake a hypothetical match-result market for Kestrel Park v Dunmore at 2.10 / 3.40 / 3.60. The raw implied probabilities are 47.62%, 29.41% and 27.78%, which add to 104.81%. The overround is 4.81%, and the payout rate is 1/1.0481 = 95.41%: a bettor who spread money across all outcomes in the right proportions would get back 95.41% of it.
A two-way market works the same way. A hypothetical quote of 1.91 on each side gives 52.36% + 52.36% = 104.71%, an overround of 4.71% and a payout rate of 95.50%. If the two sides really are equally likely, each pick has EV = 0.5 × 1.91 − 1 = −4.5%.
Why summing to more than 100% matters
The probabilities of mutually exclusive, exhaustive outcomes must add to exactly 100%. When quoted prices imply 104.81%, the implied probabilities are, taken together, 4.81 points too high. That excess is the bookmaker's price for taking the bet, spread across outcomes in a way the quote does not reveal.
You can make the margin visible. Stake 45.43% of a unit on the home win, 28.06% on the draw and 26.50% on the away win, each share being π divided by the book sum. Whatever happens, the return is 0.954 units: a certain loss of 4.59%. No combination of picks in one book escapes the margin.
The margin works like a toll on every pick. To come out ahead over many picks, your probabilities must be better than the market's by more than the toll. A forecaster who is exactly as accurate as the margin-free prices still loses on average, because every price they act on is shorter than fair.
When a book sum does not apply
The book sum only makes sense when the outcomes are mutually exclusive and exhaustive. Match result, BTTS Yes/No, and Over/Under on a half line qualify. An anytime-goalscorer market does not: several players can score in the same match and nobody may score, so adding the implied probabilities of every listed scorer tells you nothing about the margin.
A low book sum also does not make any single price good value. It lowers the average cost of betting with that book, but an individual price is only worth taking if your probability clears its break-even bar, and the margin may be loaded unevenly onto some outcomes.
Book sums also work across bookmakers. Suppose a second hypothetical bookmaker quotes 2.20 / 3.30 / 3.50 for the same match, a book of 104.33%. Taking the best price for each outcome (2.20, 3.40, 3.60) gives a combined book of 102.64%: shopping between books roughly halves the margin you face. The lesson on line shopping develops this, including why the best quote is not always an executable one.
Worked example
Worked example · Reading a hypothetical 1X2 market
Kestrel Park v Dunmore, hypothetical match-result quotes: home 2.10, draw 3.40, away 3.60.
- Raw implied probabilities
1/2.10, 1/3.40, 1/3.6047.62%, 29.41%, 27.78% - Book sum
0.47619 + 0.29412 + 0.277781.04809 = 104.81% - Overround
1.04809 − 14.81% - Payout rate
1 / 1.0480995.41% - Margin-neutral stakes
Each π ÷ 1.04809 of a 1-unit budget45.43%, 28.06%, 26.50% - Return on any result
0.4543 × 2.10 = 0.2806 × 3.40 = 0.2650 × 3.600.954 units: a certain loss of 4.59% - First-pass fair prices
Proportional fair probabilities 45.43%, 28.06%, 26.50%; fair odds = 1/p2.201, 3.563, 3.773
The quotes 2.10 / 3.40 / 3.60 contain a 4.81% overround. To break even, a home pick must win 47.62% of the time, while the market's own margin-free estimate of the home win is only about 45.4%. The gap between those two numbers is the margin you pay on that outcome.
Practice · Measure a margin
Open the Remove the Margin lab with the preset 2.10 / 3.40 / 3.60. Confirm the book sum and the proportional fair probabilities. Then change the draw price to 3.60 and watch the book sum and the fair probabilities move.
Lab 2 · Remove the margin
Synthetic · deterministic| Method | Home | Draw | Away |
|---|---|---|---|
| Raw implied (with margin) | 47.62% | 29.41% | 27.78% |
| Proportional | 45.43% (2.201) | 28.06% (3.563) | 26.50% (3.773) |
| Additive | 46.02% (2.173) | 27.81% (3.596) | 26.17% (3.820) |
| Power k = 1.0461 | 46.02% (2.173) | 27.80% (3.597) | 26.18% (3.819) |
| Shin z = 0.0241 | 45.87% (2.180) | 27.87% (3.588) | 26.26% (3.808) |
- Proportional: Divides every implied probability by the book sum. Simple, but assumes the margin is spread in proportion to probability.
- Additive: Removes an equal amount from every outcome. Easy to explain; it can push long shots below zero.
- Power: Raises every implied probability to a common power. Takes relatively more margin from long shots.
- Shin: Derived from Shin's insider-trading model. Takes more margin from long shots; performed well in Štrumbelj (2014).
Reading the result. The methods agree on the favourite more than on long shots. None of them recovers the “true” probability: they are different assumptions about how the bookmaker spread its margin. Differences of a point or two can decide whether a price looks like value, so state which method you used.
Educational calculator. Uses only the numbers you enter or seeded synthetic data — no live prices, no staking, no real money.
Show what you should find
With 2.10 / 3.40 / 3.60 the book sum is 104.81% and the proportional fair probabilities are 45.43% / 28.06% / 26.50%. Raising the draw to 3.60 lowers the book sum to 103.17% (a smaller margin) and the proportional fair probabilities become 46.15% / 26.92% / 26.92%. A longer price on one outcome reduces the margin and shifts fair probability to the other outcomes.
Common mistakes
- ✗ Treating 1/o as the market's fair probability.1/o includes the margin. It is your break-even bar. The market's estimate is the margin-removed probability, which is lower for every outcome when the book sums above 100%.
- ✗ Adding up implied probabilities in a goalscorer market.Anytime-scorer outcomes are not mutually exclusive, so the sum is meaningless as a margin. Use book sums only for exclusive, exhaustive outcome sets.
- ✗ Confusing overround with payout margin.A 104.81% book has a 4.81% overround and a 95.41% payout rate (a 4.59% take). State which definition you use.
- ✗ Assuming a low-margin book means every price is value.A low margin lowers the average cost; it does not make an individual price beat your probability. Check each price against its break-even bar.
When it fails
Supporting research
- Sauer (1998): The Economics of Wagering MarketsJournal of Economic Literature · Literature reviewWagering prices are, to a first approximation, efficient forecasts, so margin-removed prices are a serious estimate of outcome probabilities, while known exceptions such as the favourite–longshot bias remain.
- Hegarty & Whelan (2024): Comparing two methods for testing the efficiency of sports betting marketsSports Economics Review · Peer-reviewed methodsRaw inverse odds and normalised probabilities are not interchangeable: efficiency tests built on raw inverse odds can mislead, which is why we separate break-even bars from fair probabilities.
Check your understanding
Practice only: saved on this device, no XP and no account record.