Expected Value: The Only Number That Decides a Bet

The one line of arithmetic that separates a good bet from a bet that merely won.

Expected value is the number that decides whether a bet is worth placing. Not confidence, not form, not how the fixture looks — EV. It is a single line of arithmetic, and understanding it properly restructures how you approach every market.

The calculation

For a bet at decimal odds O, with your estimated probability p that it wins, the expected profit per unit staked is:

EV = (p × (O − 1)) − (1 − p)

The first term is what you win, weighted by how often you win it. The second is what you lose, weighted by how often you lose. A positive result means the bet makes money over many repetitions; a negative result means it loses money, however good it feels.

Worked example: a selection priced at 2.60 that you estimate at 45%. EV = (0.45 × 1.60) − 0.55 = 0.72 − 0.55 = +0.17. Seventeen pence of expected profit per pound staked. Over a thousand such bets at £10, that is £1,700 — and a stretch of thirty losing bets somewhere in the middle that will feel like proof the method is broken.

The uncomfortable implication

EV is indifferent to whether a bet wins. A +EV bet that loses was still the right bet. A −EV bet that wins was still the wrong bet. If you evaluate decisions by outcomes, you will systematically learn the wrong lessons — reinforcing bad bets that happened to land and abandoning good ones that happened not to.

This is the single hardest habit in betting, and it is the one that separates people who improve from people who cycle through strategies.

Where p comes from

Everything rests on your probability estimate, and a sloppy estimate makes the formula worse than useless — it dresses a guess in decimals. Three things improve it.

• Start from base rates. What proportion of matches like this one, historically, ended in this outcome? Anchor there before adjusting.
• Adjust for specifics conservatively. Team news, motivation and conditions matter, but the temptation is to over-adjust for the story you find most compelling.
• Check your calibration. Group your bets by estimated probability and see whether the ones you called 60% actually won about 60% of the time. If they won 45%, your edge is imaginary and the fix is recalibration, not more analysis.

Accounting for the margin

Compare your estimate against the margin-adjusted market probability, not the raw implied one. In a market totalling 105%, an outcome priced at 2.00 implies 50% raw but only 47.6% once the overround is stripped out. Skip that step and you will find phantom edges in every market you look at.

Turning EV into stake size

EV tells you whether to bet. It does not tell you how much. Once you have a positive number, size it with a staking plan — flat staking for most bettors, fractional Kelly if your calibration checks out. The two decisions are separate, and merging them is how people end up staking their bankroll on the bet with the biggest theoretical edge, which is usually the one with the biggest estimation error.

The realistic scale of an edge

A sustained edge of 2% to 5% is a genuinely good result for a serious bettor. Anything advertised well above that, over a large sample, is either a small sample presented as a large one or not true. Calibrate your expectations to those numbers, and a positive-EV approach becomes a slow grind rather than a transformation — which is exactly what it is.