What Is Expected Value in Sports Betting?
Expected value (EV) is the long-run average return on a bet. Here is the formula, worked examples at +150 and -200, and why EV is the figure that decides whether a bet is worth making.
Expected value, or EV, is the average amount a bet returns if you could replay it thousands of times. It is the one number that tells you whether a wager is worth making, because it folds together how often you win and the price you get when you do.
The formula
EV is the probability of winning times the profit, minus the probability of losing times the stake:
EV = (P(win) × profit) − (P(loss) × stake)
A positive result means the bet makes money over the long run. A negative result means it bleeds money, no matter how good it feels in the moment.
Your model gives a team a 45% chance to win, and a sportsbook offers them at +150 (risk $100 to win $150).
EV = (0.45 × $150) − (0.55 × $100) = $67.50 − $55 = +$12.50
That is +$12.50 of expected profit per $100 risked. The book thinks the team wins 40% of the time (that is what +150 implies); your model thinks 45%. The 5-point gap is the edge, and the positive EV is what it is worth in money.
Why win rate lies
It is tempting to judge a bettor by how often they win. Price breaks that instinct. A 60% win rate at -200 still loses money, because you are risking $200 to win $100 on bets that fail 40% of the time. A 40% win rate at +200 is comfortably profitable. Win rate versus ROI is its own topic, but the short version is that price decides profit.
EV combines the two into one figure, which is why Fairline ranks every play by edge and EV, the numbers that already fold price into the win probability.
Where the edge comes from
Positive EV requires a probability estimate that is more accurate than the book’s price after the house margin is stripped out. That is the entire job of a model: produce an honest win probability for each side, convert it to a fair price, and bet only when the book’s number is generous enough to clear the vig. EV is the scoreboard for whether the model is actually finding those spots.