Probability and price

What value betting means in football

Value exists when your defensible probability estimate implies a better chance than the offered odds. It is a comparison between an uncertain estimate and a price—not a promise that the next bet will win.

· 8 min read

Value starts with an independent probability

First estimate the outcome without using the price as the answer. A football prediction model can provide that estimate if its method and validation are credible. Then compare the estimate with the return offered by the market.

The expected-value formula

Expected value = (estimated probability × decimal odds) − 1

Multiply the probability as a decimal by the available decimal price, then subtract one. A result above zero is positive expected value under the estimate.

Expected value is an average across repeated comparable decisions. It does not describe what must happen in one match.

Worked example: 55% at decimal odds of 2.10

The arithmetic is exact, but the 55% input is an estimate. If the true chance were only 45%, the same 2.10 price would have negative expected value.

Why a good value decision can lose

Positive expected value does not eliminate variance. Even a correctly estimated 55% outcome fails 45% of the time. Short sequences can therefore contain several losses without showing whether the underlying estimate was good.

  • Separate decision quality from one result.
  • Judge a fixed method across a sufficiently large complete sample.
  • Track the exact price that was genuinely available.
  • Revisit calibration when forecast groups repeatedly miss their stated rates.

Account for bookmaker margin

Raw implied probabilities across all outcomes usually add to more than 100%. That excess is the overround. Removing it gives a simple market-based fair-probability estimate, though different margin models can produce different answers.

Normalised probability = raw implied probability ÷ sum of all outcome probabilities

This proportional method rescales the outcome probabilities so that they total 100%.

A disciplined value checklist

  • Define the selection and market precisely.
  • Use a probability produced before seeing the result.
  • Convert the available odds into implied probability.
  • Remove margin consistently when comparing market beliefs.
  • Record the estimate, price, timestamp and later settlement.
  • Review complete results rather than selected wins.

Limits and responsible use

Model error, stale prices, stake limits, account restrictions and changing markets can make realised results differ from a clean calculation. Past value estimates and historical records do not guarantee profit.

Sources and further reading