How to Interpret Football Odds When Probabilities Exceed 100%

At a price of 2.00 for a home win, the implied probability of that outcome is 50%. Put 3.50 next to the draw, and 4.00 next to the away win, and the three prices indicate an overall probability of 103.57%. A football match has three possible outcomes, and one of them must happen. The other 3.57 percentage points of probability is built into the prices.

A £10 stake at 2.00 would return £20, and a £10 stake cashed out at x2 in a round of crash games would also return £20, assuming the cash out was taken before the game stopped. Neither gives any information on the likelihood of either outcome.

The Calculation Hiding in Decimal Odds

A quoted price for a match, such as 2.00, shows what the bookmaker believes the probability is for that outcome. Another quoted price may show what a different bookmaker believes the probability is for the same outcome. A prediction percentage requires a forecasting model, and assumes certain things, often based on historical data. A prediction percentage for a match can also be hidden in decimal odds. To find it, divide 100 by the quoted price.

At 2.00, the implied prediction is 50%. At 3.50, it is approximately 28.57%. At 4.00, it is 25%. The three different figures are not three independent predictions. They represent three possible outcomes for a single match.

Full-time selection Hypothetical decimal odds Raw implied probability
Home win 2.00 50.00%
Draw 3.50 28.57%
Away win 4.00 25.00%
Combined — 103.57%

No actual fixture is represented by these prices. They were chosen to illustrate the arithmetic. Actual quotes for a match can differ, including between the first listing and kick-off.

Where the Extra 3.57 Points Come From

The combined percentage above 100 is called the overround. Here it is 3.57 percentage points. It is the margin incorporated in the set of prices. It does not mean there is a 3.57% chance of a fourth result or that an individual £10 bet incurs a 35.7p charge.

Treating the raw implied probabilities as actual and mutually exclusive outcomes creates a contradiction, as they would add up to more than 100%.

Excess can be removed for comparison using normalisation, by dividing each by the total of 103.57%.

Using the adjusted figures, there is approximately a 48.3% chance of a home win, a 27.6% chance of a draw and a 24.1% chance of an away win.

It does not prove they are the true chances. The calculation assumes that the margin built in to each price can be distributed across the outcomes in the same way. Actual prices do not need to work this way.

The difference between the true probability and what is paid out is the same as the difference between the numbers of sectors occupied on a wheel in a game of Spin and Win. The larger payment does not necessarily mean the outcome is more likely.

A greater decimal price results in a larger conditional payout and a lower raw implied probability.

The Gap Between a Prediction and a Price

Take the example of a forecasting model assigning a 60% probability to the home team in the hypothetical example above. The price of 2.00 still implies a 50% probability, regardless of what the model may think. Both can co-exist in the same example and need not be determined the same way.

The model can take into account recent results, the venue, the state of the teams and other factors relevant to the match. Its forecast can depend on how it weights those factors. The bookmaker’s price is a separate figure and is determined within the context of the other prices in that market.

The forecast is not a defined outcome. Even an accurately calibrated 60% forecast says nothing about the next case. Rather, it says a similar forecast should lead to success 60% of the time across numerous examples.

The draw is important too

In football, if a match ends as a draw after regular time and stoppage time, then that is an official draw. Leaving the draw price out of a 1X2 calculation would give a n incomplete picture of that market.

The Calculation Changes in a Two-Way Market

Now consider a completely different market: over/under 2.5 goals. The Over is priced at 1.80 and the Under is priced at 2.05.

The implied probabilities are 55.56% for the Over and 48.78% for the Under. The total of the two probabilities adds to 104.34%. The overround is 4.34 percentage points.

The number of outcomes has reduced from three to two, but the arithmetic is the same. All mutually exclusive outcomes that complete the market have been included.

A full-time home win and over 2.5 goals are different outcomes. It is possible for both to occur and for both to fail. Taking their implied probabilities does not reveal the overround, since they do not form a complete set of alternatives.

Odds do change with new information. Calculating with prices from one time defines the market as it is quoted at that time. A calculation with the home win price from before the game started and the draw price after the game started would give a total that would never exist as a single market.

What the Figures Establish

At decimal odds of 2.00, a stake of £10 yields a total of £20. The three-way market, however, has an overround of 3.57 percentage points. The first calculation gives the expected (or conditional) payment; the other gives how the prices meld together.

Neither calculation determines the ultimate winner. The forecast percentage, however, gives an additional estimate, which may differ from the market.

Football betting is an entertainment activity and, to be responsible, one has to know when to stop.