Expected Value Is Boring Until It Stops You Buying a Bad Bet

A winning bet feels correct. A losing bet feels wrong.

That emotional accounting is understandable and useless.

One result cannot tell you whether the price was good. A coin can land tails after you correctly accept 3.00 on heads. A 1.20 favourite can win after you paid a terrible price.

Expected value asks what the same decision would produce on average if repeated under the same probabilities.

The formula

For a simple bet:

EV = (probability of winning × profit) - (probability of losing × stake)

You stake €10 at decimal odds of 2.50. The profit if you win is €15.

You estimate the true chance at 45%.

EV = (0.45 × €15) - (0.55 × €10)

EV = €6.75 - €5.50 = +€1.25

On your estimate, the bet has positive expected value of €1.25 per €10 staked.

The formula is the easy part

The difficult number is 45%.

Where did it come from?

A model, historical data, team news, market comparison or a feeling after watching highlights?

Expected value can make weak thinking look scientific. The answer is only as good as the probability estimate.

If the true chance is 38% instead:

EV = (0.38 × €15) - (0.62 × €10)

EV = €5.70 - €6.20 = -€0.50

A seven-point change turns the bet from attractive to poor.

Positive EV does not mean likely to win

A 10.00 outcome can be +EV and still lose nine times out of ten.

Suppose the true chance is 12%. The implied chance at 10.00 is 10%.

The bet has an edge, but an 88% chance of losing on each attempt.

This is why value betting can produce long losing stretches. The mathematics describes an average over repetition, not a smooth salary.

Negative EV can win repeatedly

A heavily favoured outcome may win several times while remaining overpriced.

Imagine a true 80% chance offered at 1.20. The implied probability is 83.3%.

You can win eight of ten bets and still receive too little on the wins to compensate for the losses over time.

Casinos and sportsbooks survive because many negative-EV bets win often enough to feel safe.

The bookmaker margin matters

The odds already contain a commercial markup.

Before evaluating your view, remove or at least estimate the overround. Otherwise, you may compare your probability with a number that includes the bookmaker’s fee.

Longshots deserve extra caution because research has found that unlikely outcomes can be systematically overpriced in many betting markets. [1]

A dramatic payout can still be a poor purchase.

EV and bankroll are separate questions

A positive expected value does not tell you how much to stake.

A thin edge with high uncertainty should not receive a reckless bet merely because the spreadsheet shows a plus sign. Variance can destroy a bankroll before the theoretical advantage has time to appear.

Stake sizing, limits and uncertainty deserve their own discipline.

“I have an edge” is not permission to make the rent part of the experiment.

A practical workflow

1. Convert odds to implied probability
2. Estimate the bookmaker margin
3. Build your own probability range, not one magical number
4. Calculate EV at the low and high ends
5. Compare the price across operators
6. Record the closing odds and result
7. Review whether your estimates were consistently calibrated

A range is more honest than false precision.

You might estimate a team between 44% and 48%. If the bet is only +EV at 48%, the edge is fragile. If it remains +EV at 44%, the case is stronger.

My verdict

Expected value is not a system for winning every bet. It is a language for asking whether the reward justifies the risk.

It also forces a useful separation:

Was the decision good?

Did the result go my way?

Those questions often have different answers.

A single winner can flatter bad judgment. A single loser can insult good judgment. Expected value is the boring friend who waits for enough evidence before choosing sides.