Every bet you will ever make has one honest number attached to it: its expected value — what one unit staked on that bet returns on average if it could be replayed thousands of times. Positive EV bets make money over the long run no matter how any single night goes. Negative EV bets lose money no matter how clever they feel. Everything else in sharp betting — models, line shopping, bankroll math — exists to find and protect positive EV.
The formula (one line, no calculus)
EV = (win probability × profit if you win) − (loss probability × stake). Say you back a team at +120 and believe it wins 48% of the time: EV = (0.48 × 1.20) − (0.52 × 1.00) = +0.056 — about +5.6 cents per dollar staked. Same price, but a 42% win probability? EV = (0.42 × 1.20) − (0.58 × 1.00) = −0.076, and the bet bleeds 7.6 cents per dollar forever. The EV calculator runs this for any price and probability instantly.
Why win rate alone lies to you
A 60% win rate sounds like winning — and it loses money at −200, because that price demands 66.7%. A 35% win rate sounds like losing — and it prints at +250, which only needs 28.6%. Win rate without price is half a fact. The price's demand is called the implied probability, and EV is simply the gap between your estimate and that demand, converted into money.
Where positive EV actually comes from
The sportsbook's margin — the vig — means the average bet starts around −4% EV. To get positive you need an estimate that is genuinely better than the market's in some spot: a situational edge the line underweights (rest, travel, scheduling), a number that moved past fair value, or a backtested pattern the market is slow to price. That last one is what a betting model is for — it replaces “I feel good about this” with a win rate measured over hundreds of historical games.
What a realistic edge looks like
- −4% to −5% — the average casual bet. The vig, working as designed.
- +1% to +3% — a real, sustainable edge. This is where most profitable bettors actually live.
- +3% to +7% — excellent, usually in softer markets like props and smaller leagues.
- +15% and up — almost always a broken probability estimate, not a broken market. Recheck the sample before celebrating.
Small edges compound: +3% EV at 10 bets a week is roughly 15 units a year at flat stakes. The discipline is refusing the −4% bets in between — every pass on a negative EV bet is worth exactly what it would have lost.
Put a number on your next bet
Before your next wager, estimate the win probability honestly, run it through the EV calculator against the posted price, and only bet if the number is positive. Want the estimate itself to be more than a feeling? Build a model and let five seasons of history do the estimating — free, no account needed.