Expected value in betting: why a winning bet can still be a bad bet

Your buddy hit a +900 longshot last night and will not shut up about it. Great read, he says. Called it.

He did not call anything. He got paid on a bad bet. And until you can see why a winning bet can be a bad bet, and a losing bet can be a good one, you are gambling in the dark while telling yourself it is skill.

This is the idea that separates bettors who last from bettors who fund the ones who last. It is called expected value.

The lie: it won, so it was a good bet

Here is the belief we are breaking:

It won, so it was a good bet.

This is the most natural mistake in gambling, and the most expensive. It fuses two different things, the decision and the result, and judges the decision by the result. But the result is one noisy sample. The decision is the thing you repeat a thousand times, and only the decision has an expected value.

A good bet is one where the price you got was better than the true odds. Whether that particular bet won or lost is mostly noise on top. Judge the price, not the outcome. That is the whole reframe.

What expected value actually is

Expected value, EV, is what a bet is worth on average if you could make it over and over. The formula is plain:

EV = (probability of winning x profit if you win) minus (probability of losing x amount you stake)

If that number is positive, the bet makes money in the long run. If it is negative, it loses. One bet tells you almost nothing. Ten thousand bets tell you everything, and EV is the average of those ten thousand collapsed into a single number you can compute before you bet.

Work a clean example. You bet 100 dollars at +100 (even money, double your money). You believe the true chance is 55%.

• Win: 0.55 x 100 = 55
• Lose: 0.45 x 100 = 45
• EV = 55 minus 45 = +10 dollars per 100 staked

Positive. Good bet. It can absolutely lose tonight. Over a season of bets like it, you print. Now flip your buddy's longshot: a +900 bet he thinks is 20% but is really 8%. The book priced 8% fairly at +900 (roughly). He is betting a number that is negative EV and got lucky. Good result, bad bet.

The number: 52.4% at -110

Here is the anchor everything hangs on.

At the standard price of -110, you risk 110 to win 100. To break even you must win often enough to cover that vig. The break-even win rate is 52.38% (100/110 divided out, or 110/210). Call it 52.4%.

Read that slowly. A coin flip, 50%, loses money at -110. You are not fighting the other team. You are fighting a price built so that being right half the time still bleeds you. To make money you must be right more than 52.4% of the time at that price, or get a better price on the same opinion.

That single number is why casual bettors lose even when they win "about half." Half is not enough. The vig set the bar above half on purpose.

How to find positive EV: remove the vig

Positive EV means betting a price that pays better than the true probability. To spot it, you convert the odds to an implied probability and strip out the book's cut.

Convert American odds to implied probability:

• Negative odds: implied % = odds / (odds + 100). So -110 = 110/210 = 52.4%.
• Positive odds: implied % = 100 / (odds + 100). So +150 = 100/250 = 40%.

Now the catch. Add up both sides of a market and you get more than 100%. That extra is the vig. A -110 / -110 market implies 52.4% + 52.4% = 104.8%. That 4.8% over 100 is the house's cut baked into the price. To get the market's honest estimate, you remove the vig (divide each side by the total). The de-vigged fair price on that market is 50/50.

So the hunt for value is simple to state, hard to do: find bets where your estimated true probability is higher than the de-vigged implied probability. That gap is your edge. Here is the shape of it:

Bet	Your true estimate	De-vigged fair	Price you can get	Verdict
Team A moneyline	58%	54%	-110 (implied 52.4%)	+EV, bet it
Team B total over	50%	52%	-110 (implied 52.4%)	-EV, pass
Longshot +900	8%	8%	+900 (implied 10%)	-EV, pass

You do not need a big gap. You need a real one, repeated, at a price you can actually get. Which raises the obvious question: how do you know your estimate is right? You mostly do not, on any single bet. You find out over time by tracking whether you consistently beat the market's closing price, which is closing line value, the closest thing to a live scoreboard for whether your bets are +EV.

Process over outcome

This is the mental habit that changes everything. Grade your bets by the price you got, not by whether they won.

• A +EV bet that lost was still a good bet. Make it again.
• A -EV bet that won was still a bad bet. Stop making it.

If you flinch at that, you are still judging decisions by results, and the book loves you for it, because it means a lucky win will keep you making negative-EV bets until the average catches up. The average always catches up. Variance is loud in the short run and silent in the long run, and EV is what is left when the noise dies.

Edge size dictates how long you wait

A positive expected value is a rate, not a schedule. It tells you the bet pays on average, not when. The smaller the edge, the longer the wait before your record reflects it, and this is exactly where beginners misread their own results.

A thin edge of a couple of percentage points over the break-even bar needs a large sample before the signal climbs above the noise. A fatter edge surfaces much faster, because each bet contributes more drift in the right direction. The practical consequence is uncomfortable: a real but small edge can produce a losing month, and a non-existent edge dressed up as a big one can produce a winning month, and only volume sorts them out.

This is why expected value never travels alone. EV gives you the direction, bankroll sizing keeps you alive until the direction pays, and closing line value confirms the direction was real in the first place. Pull any one of the three out and the other two stop working. A positive-EV bettor with no bankroll discipline goes broke during the wait. A positive-EV bettor with no CLV tracking never finds out whether the edge was ever really there.

A worked scenario: two bettors, one season

Meet two bettors. Both place 500 bets this season at -110.

Bettor A bets on feel. He is genuinely good at reading games and wins 51% of the time. Sounds like a winner. It is not. At -110, break-even is 52.4%, so 51% is below the line. Over 500 bets he is a steady loser, and he will spend the whole season blaming bad beats, because 51% feels like winning. His problem is not luck. His problem is that his price required 52.4% and his edge delivered 51%.

Bettor B is a worse handicapper but a better bettor. She wins only 50% of her raw opinions, but she shops for the best number every time and often bets a price better than -110, sometimes +100, sometimes -105. Because her average price is better, her break-even bar is lower than 52.4%. Same raw skill as a coin flip, but she clears her bar and grinds out a small profit.

The difference is not talent at picking winners. It is respect for the price. Bettor A ignored EV and lost with a better record. Bettor B respected EV and won with a worse one. Over 500 bets, the math does not care who looked smarter on any given Sunday.

The three places your edge actually comes from

Positive EV has to come from somewhere real. In practice there are only a few honest sources, and it helps to know which one you are claiming.

1. A better estimate than the market. You genuinely think a team's true probability differs from the de-vigged line, and you are right often enough to matter. This is the hardest and rarest edge, because the closing market is very sharp.
2. A better price than the market. Same opinion as everyone, but you got a superior number by shopping books or betting before the line moved. This is the most reliable everyday edge and it requires no genius, just discipline. It is covered in line shopping.
3. A structural or promotional edge. Boosted odds, a soft market a book is slow to correct, or a price distortion. Real, but usually small and short-lived.

If you cannot name which of these your bet relies on, you probably do not have an edge, you have a hunch wearing a formula. Naming the source keeps you honest.

Where to be careful

• Your probability estimate is the hard part. The formula is easy. Estimating the true probability better than the market is the entire game, and the market is sharp. Assume you are wrong until closing line value says otherwise.
• Small samples lie. A 60% record over 20 bets is well within luck. You need hundreds of bets before a win rate means much, a point we expand in how to verify a betting claim.
• Price is not fixed. The same opinion is +EV at one book and -EV at another, because the price differs. Shopping the number is free EV, covered in line shopping.

Where a service fits

A picks service is, at heart, a claim about probability: "this is +EV at this price." That claim is only worth money if it is true and verified. A tool like ParlayScience markets pick cards with a stated edge percentage, which is literally an EV claim, plus a suggested stake. That is the right information to hand a bettor. The honest move is to treat the stated edge as a hypothesis and confirm it against the closing line before you trust it. You can see how ParlayScience frames its edge on Whop and hold it to the verification standard in our review.

Estimating probability without a fancy model

The EV formula needs your true probability estimate, and that is where most people freeze, because they assume you need a statistical model. You do not need one to start. You need a disciplined way to turn odds into probabilities and a habit of comparing your read to the market.

Start from the market itself, because it is a strong estimate you can anchor to. De-vig the line to get the market's fair probability, then ask one honest question: do I have a specific, nameable reason to think the truth differs from this number? Not a vibe, a reason. A key injury the line has not fully absorbed, a scheduling spot, a matchup the market tends to misprice. If you cannot state the reason in a sentence, your estimate is just the market's estimate with extra confidence, and you should pass.

When you do have a reason, quantify it modestly. If the fair line says 52% and you genuinely believe a factor moves it a couple of points, call it 54%, not 65%. Beginners blow up by turning a small, real edge into a huge, imaginary one. Small honest edges compound. Large fantasy edges detonate. Over time, the feedback loop that tells you whether your adjustments are any good is closing line value: if your reasons are real, the market tends to move toward your side by close. If it keeps moving away, your reasons are stories, and the number is telling you so.

The point is that EV is not gated behind a data-science degree. It is gated behind honesty: an honest conversion of odds to probability, an honest reason for any disagreement with the market, and an honest, modest size on the edge you claim.

The takeaway

A good bet is a good price, full stop. Not a win. A price better than the true odds, bet repeatedly, sized sensibly. You must clear 52.4% at -110 just to break even, so "about half right" is a slow loss. Find the gap between your honest estimate and the de-vigged market, bet it small and often, and judge yourself by the process, not last night's result. The wins and losses are weather. EV is the climate.

Bet only what you can afford to lose. If gambling stops being fun, it is time to stop. Help is available (in the US, call 1-800-GAMBLER). 21+, where legal.

FAQ

What is positive EV in sports betting? Positive expected value means the price you got pays better than the true probability of the bet, so it makes money over the long run even though it can lose any single time. You find it by comparing your honest probability estimate to the market's de-vigged implied probability and betting only when yours is higher.

How can a winning bet be a bad bet? Because the outcome is one noisy sample and the decision is what you repeat. If you bet a price worse than the true odds and happened to win, the decision was still negative EV and will lose over many repetitions. Grade the price, not the result.

What is the 52.4% break-even? At standard -110 odds you risk 110 to win 100, so you must win 52.38% of the time just to break even. A 50/50 coin flip loses money at that price because the vig sets the bar above half. To profit you must beat 52.4% or get a better price.

How do I remove the vig from odds? Convert both sides to implied probability, add them (they total more than 100%, and the excess is the vig), then divide each side by that total to get the fair, de-vigged probability. Compare that fair number to your own estimate to judge value.

How do I know if my bets are actually +EV? On any single bet you mostly cannot. Over time you track closing line value, whether you consistently beat the market's closing price. Consistently beating the close is the best available evidence that your bets are positive EV. Screenshots of wins are not.

Can I be a winning bettor with a losing record? Yes, in principle. If you consistently bet prices better than -110, your break-even bar drops below 52.4%, so you can profit while winning fewer than 52.4% of bets. This is why shopping for the best number matters as much as picking the right side. Price and record together determine profit, not record alone.

Is expected value the same as guaranteed profit? No. Expected value is the long-run average of a bet repeated many times. Any single positive-EV bet can lose, and a run of them can lose, because variance dominates the short term. EV tells you what the bet is worth on average, not what will happen tonight. There is no guaranteed profit in betting.

Do I need a statistical model to bet with positive EV? No. You need a disciplined way to convert odds to probability, the market's de-vigged line as an anchor, and an honest, specific reason before you disagree with it. Size any edge modestly and let closing line value tell you over time whether your reasons are real. A model can help, but discipline and honesty about your estimate matter far more than fancy math.