EV is the only score that matters over a season: what it measures, the maths on a −110 bet, and why price beats picking winners.
One number decides whether a bet was good
Every bet you place either had positive expected value or it did not, and nothing that happens after kickoff changes which. Expected value is the average profit or loss a bet would produce if the same situation ran thousands of times: your win probability times what you win, minus your lose probability times what you risk. It is the only honest scoreboard in betting, because it grades the decision rather than the outcome.
This is the hardest idea in the sport to internalise, because results shout and EV whispers. A terrible bet cashes and feels brilliant. An excellent bet loses and feels like a mistake. Over one weekend the two are indistinguishable; over a season, EV is nearly all that remains.
The maths on a single −110 bet
Take a standard spread bet at −110 — risk 110 to win 100. The price implies a break-even win rate of 52.4%. Now suppose your true chance of covering is 55%. The expected value per 110 staked is 0.55 × 100 − 0.45 × 110 = 55 − 49.5 = +5.5. Roughly five percent of your stake, on average, every time you make this exact bet.
Flip the numbers and the same arithmetic bites. At a 50% true win rate — a genuine coin flip — the −110 price grinds you down: 0.50 × 100 − 0.50 × 110 = −5 per 110 risked. That is the vig doing its quiet work, and it is why "I pick winners about half the time" is a losing record, not a respectable one.
EV comes from price, not from winners
The revelation that reorganises most bettors' thinking: the same team, in the same game, can be a good bet at one price and a bad bet at another. A favourite at −3 (−105) and the identical favourite at −3.5 (−115) are different propositions with different EV, even though every argument about the teams applies equally to both.
That is why disciplined bettors talk about numbers, not teams. "I like the Bills" is a television opinion. "I like the Bills at anything better than −3" is a bet. Line shopping — taking the best available quote on the side you already want — is the purest EV there is: identical opinion, better price, mathematically more money.
Where positive EV actually hides
The market's consensus price, with the vig stripped out, is the best public estimate of any bet's true probability. Positive EV means beating that estimate — knowing something, or weighing something, better than the closing market does. That bar is high. It is not cleared by fandom, by touchdown montages, or by a tout's record screenshot.
It can be cleared at the margins: books shade prices toward popular sides, secondary markets get less attention than main lines, and prices disagree across books more often than efficiency theory says they should. Our model hunts the first kind of gap — it projects every game unblended by the market, converts the difference into percentage points of edge, and publishes the result as a lean with a star rating. Small edges, honestly labelled, are what real +EV looks like.
EV is the destination; variance is the road
A +5% EV bet still loses 45 times in a hundred. Positive expectation guarantees nothing about any bet, any week, or even any month — it promises only where the average settles if you keep making good decisions. This is why bankroll rules exist: sizing keeps you solvent long enough for the average to arrive.
The practical routine is short. Estimate the probability first, look at the price second, and bet only when the first number beats the second with room to spare. When you cannot tell whether it does, the bet with the best EV is usually no bet at all.
Put it to work
- Best Bets Model probability vs de-vigged price, expressed as EV and stars.
- Odds Board The same side at every book — take the best price or none.