Commission is easy to spot: it is a line item taken from a winning bet on an exchange, after the fact. The overround works differently. It is built into every price in a market before you ever see them, spread across every outcome at once. That is why it is easy to miss and impossible to avoid simply by picking a different side of the same market.
The formula
Add up the implied probability of every outcome in a market. In a market with no margin, that sum is exactly 100%, with every possible result accounted for once. Real markets sum to more:
overround = (implied probability, outcome 1) + (implied probability, outcome 2) + … - 100%
Worked example
A three-way football market, home, draw, away, priced at 2.20, 3.40 and 3.20:
- Implied probabilities — 45.45%, 29.41%, 31.25%
- Sum — 106.12%
- Overround — 6.12%
That 6.12 cents of extra "certainty" bought for every dollar of true 100% is the bookmaker's structural margin on the market as a whole. It is not a fee attached to any single bet within it, but a markup spread across all three prices simultaneously.
Why it is a tax on the market, not a fee on your bet
Commission is charged against the specific bet that wins. Change which selection you back, and you change whether commission applies to your winnings at all. The overround does not work that way. It is already priced into home, draw and away alike before you choose between them.
Picking the draw instead of the home side does not let you escape it, because the 6.12% is not sitting on one outcome. It is distributed across the relationship between all three. The only way to reduce how much of it you personally pay is to find a price, anywhere, closer to what you believe the true probability actually is. The overround shrinks your edge on every selection in the market, not just one of them.
Stripping it out
The simplest way to estimate "true" probability is to normalise: divide each outcome's implied probability by the sum of all of them, so the total comes back down to exactly 100%:
- Home - 45.45% ÷ 106.12% = 42.83%
- Draw - 29.41% ÷ 106.12% = 27.72%
- Away - 31.25% ÷ 106.12% = 29.45%
Each outcome loses roughly its own proportional share of the 6.12%. It is the standard first-pass method, and it is an approximation rather than an exact recovery of the bookmaker's real view. It assumes the margin was spread evenly across outcomes in proportion to their own price, which is a simplifying assumption, not a fact about how bookmakers actually set prices.
Margin depends heavily on which market it is
The 6.12% above is a mid-range figure. Deep, heavily traded markets on major fixtures tend to carry noticeably less. Competition between bookmakers for the same popular bets pushes prices closer together and margins down.
A tight two-way market at 1.95 and 1.95 carries only about a 2.56% overround. A niche prop market, a lower-tier fixture, or a same-game combination bet typically carries far more, since there is less competitive pressure and often less liquidity behind the price.
A two-way market at 1.70 and 1.70, for example, carries roughly 17.65%. Same formula, nearly seven times the margin, simply because of the type of market.
Checking the overround before backing anything in an unfamiliar or thinly traded market is worth the ten seconds it takes, since the margin alone can already exceed whatever edge a bettor believes they have found.
Where the approximation bends: the favourite-longshot bias
A well-documented pattern in betting markets, across many sports and many decades of data, is that bookmaker margins are not actually spread evenly by that measure. Longshots tend to carry proportionally more margin than favourites do, a pattern researchers call the favourite-longshot bias.
Its likely causes are debated. Bettors systematically overvaluing the excitement of a big-priced win is one common explanation among several, but its practical effect on the simple proportional method above is consistent: it tends to slightly overstate a longshot's true chance and slightly understate a favourite's.
This happens because the method assumes an even spread that the real market does not quite deliver. More sophisticated stripping methods exist specifically to correct for this, including Shin's method and various power or logarithmic adjustments, but they also introduce additional assumptions of their own.
None of these methods make the proportional method wrong to use. They simply make it useful to understand which direction its error tends to run.
A quick habit worth building
Summing implied probability takes seconds and answers a useful question before backing anything: how much of this price is genuine market opinion, and how much is structural markup that would be there no matter what actually happens? A price that looks generous in isolation can still sit in a market carrying 15% or more of overround. A price being better than a competitor's does not necessarily mean it is better than the margin sitting underneath it.
Check the book and the normalised probabilities on any market with the implied probability calculator. The overround it shows is the same number sitting in every price in that market, whichever one you end up backing.