Fair odds and value: what does the price tell you?

A fair coin has a 50% chance of landing heads. Decimal odds of 2.00 return £20 from a winning £10 stake. At those odds, repeated independent £10 bets have zero expected profit before any costs: half win £10, half lose £10.

At 1.80 the same winning bet earns only £8, while a loss still costs £10. At 2.20 a win earns £12. The event has not changed; the price has. That is the basic distinction between predicting a winner and finding value.

Fair odds = 1 ÷ probability. A 25% estimated scoring chance corresponds to 4.00; 40% corresponds to 2.50. The word estimated matters: a model’s fair price is conditional on its inputs, not a fact about the future.

Take a two-player market priced at 1.90 on both sides. Each implies 52.63%, so together they add to 105.26%. The extra 5.26 percentage points are the overround. They are not a prediction that both players can win.

Proportional margin removal divides each implied probability by the total. Here that restores 50% per player, or fair odds of 2.00. Shin and odds-ratio methods allocate the adjustment differently on uneven markets. They are alternative estimates, not ways to uncover a guaranteed true price.

Use a complete, mutually exclusive market. You cannot remove the margin from a list of anytime goalscorers by normalising their probabilities: several players can score in the same match.

If the price falls to 2.30, the same 40% estimate gives −8% EV. If the player is benched and the estimate drops to 30%, even 2.80 gives −16% EV. A good-looking price gap is a reason to check minutes, team news and the market mapping.

1. Check the fixture and update time

   A recorded reference price may have moved. Missing data means there is no comparison, not that the model has found no value.

2. Read the player assumptions

   Likely minutes, starting status and penalty responsibility can all change a goalscorer estimate.

3. Compare your actual price

   Use the exact market and bookmaker rules. The board’s reference is not a promise of the best available odds.

4. Keep research and published tips distinct

   A model disagreement is research. Latest hits show recent successful examples, not every signal or an audited profit record.