Expected value is the arithmetic that separates a “good bet” from a “bet on the team you like.” It answers one question: if you could replay this exact wager thousands of times, at this exact price, would you come out ahead on average? A bet is positive EV (+EV) when the odds on offer pay you more than your honestly-estimated probability of winning justifies. It’s negative EV (-EV) when the reverse is true — which, after the bookmaker’s margin is baked in, is where most casual bets sit by default.

The mechanics are simple once you separate two numbers that people constantly blur together: the price’s implied probability, and your own estimate of the true probability. Decimal odds convert to implied probability with 1 ÷ odds — so 2.50 implies a 40% chance. If you genuinely believe the true chance is higher than 40%, the price is generous and the bet is +EV. If you believe it’s lower, the bookmaker has priced you correctly (or has the edge), and taking it is -EV no matter how confident you feel about the outcome.

This is why EV and “picking winners” are different skills. You can be right about who wins and still make a bad bet, if you paid too short a price for that certainty. You can also be wrong about the outcome — lose the bet — and still have made a correct, +EV decision, because a single result is one draw from a probability distribution, not proof the decision was flawed. That distinction is the single hardest thing for new bettors to internalize.

Example

Say Wimbledon has two players in a fictional quarterfinal: Elena Vukovic and Marta Delgado. A bookmaker prices Vukovic at 2.50 to win the match. Convert that to implied probability: 1 ÷ 2.50 = 40%.

You’ve watched both players all season. Vukovic has won 7 of her last 9 matches on grass, has a big first serve that neutralizes Delgado’s best weapon (her return), and Delgado is carrying a minor thigh niggle from the previous round. Weighing all that, you land on a true win probability for Vukovic of 48% — 8 percentage points above what the price implies.

You stake $200 at 2.50. Work out EV directly:

  • If Vukovic wins (48% of the time): profit = stake × (odds − 1) = $200 × 1.50 = $300
  • If Vukovic loses (52% of the time): loss = −$200

EV = (0.48 × $300) + (0.52 × −$200) = $144 − $104 = +$40

There’s a shortcut that gives the same answer as a percentage of stake: EV% = (probability × decimal odds) − 1. Here that’s (0.48 × 2.50) − 1 = 1.20 − 1 = 0.20, or +20% EV. Multiply by the $200 stake and you get the same $40.

That $40 is not what you expect to win on this specific match — you’ll either be up $300 or down $200 when the match finishes, nothing in between. The $40 is what this bet is worth on average, if the same 48%-probability, 2.50-price situation repeated hundreds of times. Make enough decisions with a genuine 20% edge and the results converge toward that average; make one such bet and anything can happen.

Key Points

  • A single result never confirms or denies your EV: Losing a +EV bet doesn’t mean the decision was wrong, and winning a -EV bet doesn’t mean it was right. Judge the process — was your probability estimate honest and better-informed than the market’s — not the scoreboard from one match.
  • Your probability estimate is the whole game: The formula is trivial; getting 48% right for Vukovic instead of a wishful 55% is the actual skill. An EV calculation is only as good as the estimate you feed into it, so it’s worth stress-testing your numbers against recent form, head-to-head record, conditions (grass, hard-court, wet outfield in cricket, wet track in Formula 1) and injury news before trusting them.
  • Shop the price before you fault the model: The same Vukovic bet is worthless at 2.05 (implied 49%, above your 48% estimate) and strongly +EV at 2.50. Comparing odds across several bookmakers on the same market often swings a bet from break-even to genuinely profitable, or vice versa.
  • Overround eats your edge before you start: Add up the implied probabilities of every outcome in a market — a football match’s home/draw/away, say — and it’ll exceed 100%. That excess is the bookmaker’s built-in margin, so your true-probability estimate needs to beat not just the fair price but the inflated one actually on offer.
  • Closing line value is your best reality check: If you consistently bet a price and the market moves in your direction before the event starts (the closing price is shorter than what you took), that’s independent evidence your probability estimates are sound — a far more reliable signal over time than whether any individual bet won.
  • +EV tells you whether to bet, not how much: Once you’ve found a genuine edge, staking is a separate decision — flat-staking a fixed unit or applying a fractional Kelly approach both protect you from one bad estimate wiping out a string of good ones.