+EV EDUCATION

What Is +EV Betting? Expected Value Formula & Examples

inplayLABS · Published October 9, 2026 · Updated October 10, 2026 · AI-assisted educational guide

+EV betting means a wager has positive expected value: under your probability estimate, its average net return is greater than zero. For a simple win-or-lose bet, EV per unit is win probability × decimal odds − 1. The difficult part is verifying the probability—not just doing the arithmetic.

Looking for tools to organize your research? See the LABS Data Suite ↗.

Start with the price and a probability estimate

For an ordinary win-or-lose cash wager, decimal odds include the returned stake. If the odds are 2.00, a winning $10 wager returns $20 in total, for $10 net profit. A losing wager has −$10 net profit.

Let p be your estimated win probability and d be the decimal odds. For a one-unit stake, expected net profit is:

EV per unit = p × (d − 1) − (1 − p) = p × d − 1

This formula assumes no push or other settlement outcome and no fees. It is a probability-weighted average of net profit, not a payout prediction for a single bet. Smarkets explains the same win/loss method in its expected-value guide.

A hypothetical example

Suppose a fictional model estimates a 52% chance of winning at decimal odds of 2.00. Expected net profit per unit is 0.52 × 2.00 − 1 = 0.04 units, or 4% of the stake. A $10 example stake has $0.40 expected net profit under that model.

That does not mean the wager returns $0.40. It wins or loses according to its settlement. The positive expectation depends on the 52% estimate being valid. If the real probability were 48%, the same price would instead have −4% expected net profit per unit.

Break-even probability is a useful first check

With the same simple win/loss assumptions, the break-even probability is 1 ÷ decimal odds. At 2.00 it is 50%. At 1.90 it is approximately 52.63%. A claim that a 52% win rate is profitable at 1.90 therefore fails this basic check before fees.

Implied probability describes a price. It is not automatically the true probability. When comparing a full market, remember that quoted implied probabilities can include bookmaker margin. Removing margin also requires a method and does not magically reveal the correct forecast.

Test sensitivity before trusting a single number

A small change in the probability estimate can reverse the conclusion. At odds of 1.90, a 54% estimate gives 2.6% EV per unit. A 52% estimate gives −1.2%. If uncertainty spans both estimates, reporting only the optimistic percentage hides a meaningful limitation.

Use a range when the estimate is uncertain. Ask whether the argument survives a less favorable but plausible probability, a worse available price and the actual fees. A number produced by a calculator is only as reliable as its inputs.

Research starts with context. Explore LABS player-prop and live game-state tools to see what you can investigate with the relevant access.

What could invalidate a claimed edge?

Stale price: an attractive quote may no longer be available when checked.
Changed conditions: lineup, role or game-state changes can make an earlier model unsuitable.
Data leakage: a historical test can accidentally use information unavailable before the event.
Selection bias: testing many filters and reporting only the best result overstates confidence.
Settlement differences: pushes, voids, dead heats or commissions require the appropriate calculation.
Correlated outcomes: related selections cannot be treated as independent without justification.

Keep predicted EV and realized ROI separate

EV is a model-based expectation. ROI describes settled historical results. A positive realized ROI does not by itself validate the model, and a losing short period does not by itself refute it. Both interpretation and sample size matter.

Maintain a complete bet log, including losses. If you record a probability estimate, save the estimate, price and time before the event. That makes later review more useful than reconstructing a favorable explanation after a result.

Use research to ask better questions

LABS offers sport-specific research and live game-state tools with the relevant product access. Those tools can help organize the evidence you examine. They do not establish that every observed pattern is a profitable bet.

For an upcoming-game preview, separate confirmed facts from projections. Record the source and timestamp for injuries, lineups and market prices. Treat unknowns as unknowns. An interesting matchup is a research question; it becomes a supported +EV claim only if the price and probability evidence justify it.