Expected value (EV) measures whether a bet offers a positive long-run edge after accounting for odds and estimated win probability. It is central to quantitative sports betting research—not a promise of short-term profit.
The EV formula
For decimal odds O and model probability p:
EV = (p × O) − 1
Expressed as a percentage: EV% = ((p × O) − 1) × 100.
Example: model probability 55%, decimal odds 2.00 EV = (0.55 × 2.00) − 1 = 0.10 → +10% EV.
Negative EV means the market price is unfavourable relative to your estimate.
Implied probability from odds
Bookmaker decimal odds imply probability 1/O, but margins push implied probabilities above 100%. Compare implied vs model probability and bookmaker margins.
Why positive EV still loses often
Even +10% EV selections lose most individual bets when probabilities are below 50%. Variance dominates small samples. See why models lose with an edge and Kelly criterion for bankroll sizing.
Closing line value as a sanity check
If your model consistently finds +EV at bet time but closes below market, execution or calibration may be wrong. Study closing line value.
Common mistakes
1. Using win rate instead of probability-weighted returns 2. Ignoring bookmaker margin when converting odds 3. Cherry-picking date ranges after results are known 4. Treating small-sample ROI as proof of skill
AlphaBench application
Models on AlphaBench only flag opportunities that pass conservative EV thresholds and risk caps. Review live decisions on the research dashboard and historical context on model performance.
Limitations
EV depends entirely on probability quality. Overconfident models create imaginary edge. This content is educational—not betting advice.