Expected value: putting a number on a bet

"Value" is the whole game, and expected value (EV) is how you measure it. EV turns a probability and a price into a single number: your average profit per bet if you could place it thousands of times.

The formula

For a bet at decimal odds O that you believe wins with probability p, staking 1 unit:

EV = p × (O − 1) − (1 − p)

In words: the chance you win times what you'd win, minus the chance you lose times the 1 unit you'd lose. If EV is positive, the bet pays more than the risk deserves — that's a value bet. If it's negative, you're overpaying.

A worked example

Say the model gives a team a 60% chance (p = 0.60) and the odds are 2.00 (O = 2.00):

EV = 0.60 × (2.00 − 1) − 0.40 = 0.60 − 0.40 = +0.20

That's +0.20 units per bet, or +20% — strong value. Now the same 60% chance at odds of 1.50:

EV = 0.60 × 0.50 − 0.40 = 0.30 − 0.40 = −0.10

Same team, same probability, but at 1.50 the bet loses 10% on average. The price flipped a good bet into a bad one.

Why this is the point

• A favourite can be a bad bet (great chance, worse-than-fair odds).
• An underdog can be a good bet (lower chance, generous odds).
• The only question EV asks is: does the price beat the true probability?

Your p is an estimate, so EV is only as honest as your probability. FinalSkore's confidence bands are one source of p; the market's implied probability is another. When they disagree, EV tells you whether the gap is worth backing.

Positive EV doesn't mean the bet wins — it means that at this price, betting it repeatedly makes money. Any single bet can still lose.