How to calculate a bookmaker's margin

To calculate a bookmaker’s overround, convert every decimal price in one complete market into its reciprocal, add those figures and subtract one. With hypothetical match-result odds of 2.00, 3.50 and 4.00, the total is 1.035714: the overround is 3.5714 percentage points above 100%. That describes the prices offered across the market. It does not mean the bookmaker will earn exactly 3.5714% of the stakes, or that every customer has that precise expected loss.

The calculation is useful when comparing the price environment surrounding a tip. It becomes misleading when an isolated selection, different settlement rules or a payout percentage are treated as interchangeable. This guide works through a complete example and shows where the arithmetic stops. All prices and stakes below are invented for explanation; they are not current offers or recommendations to bet.

Start with a market that genuinely adds up

A standard three-way football result market has three mutually exclusive outcomes: home win, draw and away win. Exactly one settles as the winner under the specified rules. Before collecting prices, check the event, period and settlement terms. A market for the result after normal time cannot simply be mixed with a selection that includes extra time. Nor can a price for qualification replace a match-result price because the team names look the same.

Take the prices from the same bookmaker and the same moment when the purpose is measuring that bookmaker’s displayed market. Saving a timestamp helps you distinguish a real comparison from a collection of screenshots taken hours apart. If one selection is suspended or missing, the market is incomplete. Do not fill the gap with a price from another provider and continue calling the result the first bookmaker’s margin. You have constructed a different collection of prices.

Why double-chance prices cannot replace the three outcomes

The choices home-or-draw, draw-or-away and home-or-away overlap. If the home team wins, two of those selections win. Adding their reciprocal prices therefore does not create the same partition as home, draw and away. The basic formula assumes an exhaustive set of mutually exclusive outcomes. A calculator can still return a number after unsuitable inputs, but the presence of a percentage on screen does not make that number an informative overround.

The calculation, without hidden rounding

For decimal odds O, the implied figure is 1/O. Add the implied figures to obtain S. The overround is S − 1, commonly expressed by multiplying that difference by 100. Keep several decimal places until the final result. Rounding each implied percentage to a whole number before adding them can make close markets appear equal, or create differences that the original prices do not support.

Here S equals 29/28. Subtracting one gives 1/28, or 3.5714% when written as a percentage. Calling it 3.5714 percentage points above a 100% book makes the reference explicit. You may also see “book percentage” used for the full 103.5714%. Always inspect the definition before comparing two tools: one may report the full book, another the excess, and another the complement of theoretical payout.

Pinnacle’s educational margin calculator uses the relationship between decimal odds, implied probabilities and their sum. It is material published by an operator, not independent confirmation that the resulting numbers describe true sporting probabilities. The arithmetic can be checked directly from the displayed prices. A commercial description of a low margin should not substitute for checking the actual market and the selection you intend to assess.

Theoretical payout is a different percentage

For this complete fixed-odds example, define theoretical payout as 1/S. It equals 28/29, or 96.5517%. Its complement is 3.4483%, rather than the 3.5714-point overround. These figures use different denominators. The first measures an excess over a book of one; the payout complement expresses a shortfall relative to the total implied book. Subtracting the overround directly from 100% would give 96.4286%, which is not 1/S.

A stake allocation explains the distinction. Suppose a hypothetical total of 29 units is split into 14 units on the home win, eight on the draw and seven on the away win. Every winning outcome returns 28 units: 14 × 2.00, 8 × 3.50 or 7 × 4.00. The combined net result for that allocation is −1 unit whichever outcome occurs. The total return divided by total stake is 28/29. This is an arithmetic illustration, not a reason to place all three bets.

If one unit represents €25, those 29 units are €725 staked and the 28-unit return is €700, leaving −€25. Changing the chosen unit size changes the cash illustration but not the ratio. Return includes the winning stake coming back; it is not profit. This example deliberately assumes the three prices remain available and the same settlement rules apply. Real availability and any additional charges must be checked separately.

Why the bookmaker’s actual profit is not predetermined

Now replace the balanced allocation with 29 units all placed on the home win at 2.00. If the home side wins, the bookmaker pays 58 units against 29 received: the result from those bets is −29 units before other costs. If it loses, those bets return nothing and the bookmaker keeps the 29 units. The quoted market has exactly the same overround in both cases. The distribution of money and the sporting outcome change the realized result.

Actual business profit also cannot be read from one event’s odds table alone. The example above concerns stakes received and returns paid for that small set of bets; it does not account for operating expenses or other business items. Avoid turning a theoretical payout into an audited profit statement. Equally, a customer’s realized loss over a short run is not a direct measurement of the market’s overround. One describes prices; the other describes outcomes actually experienced.

Removing the excess does not reveal the true probabilities

The simplest normalization divides each reciprocal by S: pᵢ = (1/Oᵢ)/S. Our example produces 48.2759%, 27.5862% and 24.1379%. These numbers sum to 100% by construction. Their inverse prices would be approximately 2.0714, 3.6250 and 4.1429. Calling these “fair odds” requires a stated modelling assumption. They are prices implied by proportional adjustment, not a discovery of what will really happen on the pitch.

Proportional normalization is an assumption about how to remove the excess. It scales every implied figure by the same factor. It does not prove that the original margin was distributed that way across favourites and outsiders. Clarke, Kovalchik and Ingram’s 2017 methodological paper compares adjustment methods and discusses the favourite–longshot bias that proportional normalization does not capture. Its findings on the datasets studied do not establish one universally best method for every sport and market.

This matters when somebody takes the normalized 48.2759% home figure, compares it with the original 50% break-even figure and announces a precisely measured disadvantage. That difference follows from the adjustment method. Without an independently justified probability estimate, it is not evidence of the selection’s true expected return. Adding decimal places improves numerical precision; it does not remove uncertainty about the underlying sporting probability.

A lower overall margin may still hide a worse selection price

Consider another hypothetical provider offering 1.95, 3.70 and 4.20 under identical rules at the same moment. The reciprocal sum is approximately 1.021186, so its overround is 2.1186 points. That is below the first market’s 3.5714 points. Yet the home-win price is 1.95 instead of 2.00. For a one-unit winning home bet, the net is +0.95 units rather than +1.00. The cheaper overall market does not give the better price on every selection.

Compare the actual selection price as well as the complete market. Record the offered odds, the odds genuinely available to you and whether the settlement contract is the same. A change from 2.00 to 1.95 is consequential even when the provider advertises a favourable average margin. Market-level cost is a useful contextual measure, while a particular tip must be assessed using its own obtainable price and your uncertainty about the event.

Keep price comparisons separate from performance comparisons

A snapshot of overround is not a tipster’s yield. Yield measures net results relative to stakes over a defined record; the guide to yield explains that calculation. Similarly, comparing a posted price with a later closing price asks a different question. The closing line value guide helps you examine that movement without treating it as a guaranteed future profit.

For repeated reviews, keep an observation sheet containing the date, event, full market, rules, all prices, S, overround and the particular price being followed. Add your own accepted price separately if a bet was actually made. Missing data should remain marked missing. This prevents a good-looking reconstructed book from disguising the fact that one price was never available when the tip reached you. A small, consistent record is more useful than a large collection of incomparable percentages.

Using this information when reviewing Tippium profiles

The Tippium directory lets you explore tipster profiles and their published figures. Tippium is a platform for publishing, tracking and subscribing to picks; it does not take bets or execute them at the displayed odds. Automatic profiles and demonstration data exist, so directory figures are not proof of real cash profits or an external performance audit. Identity verification also does not certify a strategy’s profitability.

If you want to follow public picks, you can create an account and compare the published information with your own observations. Active premium picks require a subscription. A pick can be edited or cancelled before its first event starts; afterwards those actions are blocked. A confirmed paid account restart can delete the history and statistics. Those limits matter when deciding how much historical evidence a profile actually provides.

Betting can cause financial harm. Understanding margin does not create an advantage or guarantee a return. Set limits you can afford, never chase losses, and stop if gambling is affecting your finances or wellbeing.