Positive EV Betting: Expected Value, Fair Prices and Variance

Positive EV betting, often written +EV betting, means placing only bets whose expected value is above zero: bets that should make money on average if you could repeat them thousands of times at the same price and the same true chance. EV stands for expected value, the average profit or loss per bet over the long run. It says nothing about whether tonight's bet wins.

The idea is the same one behind value betting, which covers building your own price and checking it against the closing line. This guide takes a narrower, more mechanical angle. It works through the EV formula with American and decimal odds, explains how many bettors use the fair price of a sharp market as their estimate of the true probability, and spends most of its time on the part people underestimate: how small a real edge is, and how many bets it takes before the results mean anything.

The expected value formula

Expected value weighs each possible result by its chance and adds them up. For a simple bet that either wins or loses:

EV = (probability of winning × profit if it wins) - (probability of losing × stake)

With decimal odds there is a shortcut: EV per unit staked = (p × decimal odds) - 1, where p is the true probability written as a decimal. Multiply by the stake to get EV in money.

Start with a $100 bet at +150 on a team with a true 42% chance. A win pays $150 profit, so EV = (0.42 × $150) - (0.58 × $100) = $63 - $58 = +$5, or +5% of the stake. The decimal shortcut agrees: +150 is 2.50, and (0.42 × 2.50) - 1 = 0.05.

Now a $110 bet at -110 on a genuine 50/50. A win pays $100, so EV = (0.50 × $100) - (0.50 × $110) = $50 - $55 = -$5, or -4.55% of the stake. Finally, a £50 bet at decimal 1.95 on a football (soccer) result with a true 53% chance: (0.53 × 1.95) - 1 = +0.0335, which is +3.35% of the stake or about +£1.68.

American odds need one extra step. A plus price is the profit on a $100 stake, so +150 pays $150. A minus price is the stake needed to win $100, so -110 pays $100 on $110, or $90.91 on $100. If you prefer the decimal shortcut, convert first: +150 is 2.50 and -110 is 1.909. Our odds calculator does the conversion.

The EV percentage is the number to compare between bets. A +$5 EV on a $100 bet and a +$5 EV on a $500 bet are very different edges: 5% and 1%.

+EV and -EV: why almost every bet starts negative

The margin a bookmaker builds into its prices is the reason most bets are -EV. On a standard point spread both sides are often -110. If the game really is a coin flip, the -110 example above applies to both sides: each loses 4.55% on average. To reach zero EV at -110 you need to win 52.38% of the time, the break-even rate explained in our implied probability guide.

Above that line, EV turns positive quickly. At a true 55% the same $110 bet has EV = (0.55 × $100) - (0.45 × $110) = $55 - $49.50 = +$5.50, a 5% edge. The distance between losing 4.55% and winning 5% is under five percentage points of win rate. That is the whole playing field for a spread bettor, and skilled bettors usually work at the bottom end of it.

+EV is a property of a price and a probability together. The same team can be +EV at +130 and -EV at +110. Favourites, underdogs, overs, unders and props can all be either; the label tells you nothing about the side, only about the price relative to the chance. It also means a +EV bet that loses was still a good bet, and a -EV bet that wins was still a poor one.

Using a sharp no-vig price as the true probability

The hard input in the formula is p. Nobody knows the true chance of a game. Positive EV bettors usually solve this by borrowing it from the market. Some markets, typically those with high limits, heavy betting and low margins, have a good record of pricing outcomes accurately, especially close to the start. They are often called sharp markets. Strip the margin out of a sharp price and you have a reasonable estimate of the true probability. The next step is to look for other bookmakers whose prices are higher than that fair price, usually because they have been slower to move.

Here is the full calculation on an NBA moneyline. A sharp market prices the favourite at -125 and the underdog at +115. The implied probabilities are 125 ÷ 225 = 55.56% and 100 ÷ 215 = 46.51%, a total of 102.07%. Dividing each by 1.0207 gives fair probabilities of 54.43% and 45.57%, which are fair prices of 1.837 and 2.194 in decimal, or about -119 and +119 in American odds.

Another sportsbook offers the underdog at +125, which is 2.25 in decimal. EV = (0.4557 × 2.25) - 1 = +2.53%, or +$2.53 on a $100 bet. The same book prices the favourite at -120, which is better than the sharp -125 but still worse than the fair -119: EV = (0.5443 × 1.833) - 1 = -0.21%. Beating the sharp price is not enough; you have to beat the fair price after the margin has been removed.

Decimal markets work the same way. A sharp football 1X2 market at Home 2.40, Draw 3.40, Away 3.10 adds up to 103.34%, which gives fair probabilities of 40.32%, 28.46% and 31.22%. If a bookmaker offers the draw at 3.60, EV = (0.2846 × 3.60) - 1 = +2.46%, about +49p on a £20 stake. The no-vig calculator handles both two-way and three-way markets.

The method has weak points. Proportional margin removal, used here, is only one way of doing it; other methods put more of the margin on the underdog, which would shrink the edge on the +125 bet. The sharp market itself can be wrong, particularly early in the week or in thin markets. And the gap can close within minutes, since a price that lags the market is exactly what other bettors are looking for too.

Why real edges are small

In busy markets, prices from different bookmakers rarely sit far apart, so the edges this method turns up are usually small: something in the region of 1% to 4%. An apparent edge much larger than that is more often a warning than a gift. It may be a stale line after team news, a market you have misread, such as a different overtime rule, or a price the bookmaker can void as an obvious error under its terms.

Small edges have two consequences. The first is that money comes slowly. At an average edge of 2.5%, $100 bets return an expected $2.50 each. To expect $1,000 of profit you need $40,000 of turnover, which is 400 bets at $100.

The second is that small errors in p matter a great deal. At decimal odds of 2.25, every percentage point of probability is worth 2.25 points of EV. In the NBA example, a fair probability of 45.57% gives +2.53%. If the true chance is 45%, the edge drops to +1.25%. At 44.5% it is +0.1%, and at 44% it is -1%. A single point of error in the fair price, which is about the size of the gap between two margin removal methods on some markets, can turn a +EV bet into a -EV one. That is why careful bettors set a minimum edge, often a couple of per cent, before they act.

EV per bet vs over 1,000 bets: variance and sample size

Expected profit grows in a straight line with the number of bets. Luck grows more slowly, with the square root of the number of bets. That one fact explains why +EV betting looks random for a long time and then, if the edge is real, slowly stops looking random.

Take a simple model: $100 bets at even money (+100, or 2.00) where you win a true 51.5% of the time. EV per bet is (0.515 × 2.00) - 1 = +3%, or $3. Each bet ends at +$100 or -$100, and the standard deviation (the typical size of the swing around the average) of one bet is about $100, since $200 × √(0.515 × 0.485) = $99.95. Over n bets the expected profit is $3 × n and the standard deviation is about $100 × √n. About 95% of results land within two standard deviations of the expected figure:

• 100 bets: expected profit $300, standard deviation about $1,000. The 95% range runs from about -$1,700 to +$2,300, and you finish behind roughly 38% of the time.
• 1,000 bets: expected profit $3,000, standard deviation about $3,160. The range runs from about -$3,320 to +$9,320, and you are still behind roughly 17% of the time.
• 10,000 bets: expected profit $30,000, standard deviation about $10,000. The range is roughly +$10,000 to +$50,000, and the chance of being behind falls to around 0.1%.

The expected profit only clears two standard deviations at around 4,400 bets. Win rate tells the same story. Over 1,000 bets, a true 51.5% bettor will usually show a win rate somewhere between about 48.3% and 54.7%, a band wide enough to include both a losing -110 bettor and a very good one.

These figures use a normal approximation and assume your 51.5% is exactly right and every bet is independent. Real betting is messier: odds vary, edges vary and your probabilities carry errors of their own. The point survives the simplification. After a few hundred bets, profit and loss is a poor guide to whether an edge is real, which is why EV bettors also track closing line value, a signal that settles much faster than results.

Variance is also why stake size matters. The Kelly criterion sets a stake from the edge and the odds. For the +125 underdog, full Kelly is 2.53% ÷ 1.25 = 2.02% of the bankroll and quarter Kelly is about 0.51%, or roughly $10 on a $2,000 bankroll. The Kelly criterion calculator does the sum, and the bankroll management guide compares it with flat staking. Small, steady stakes let an edge work over thousands of bets; bigger ones mostly make the swings bigger.

Limits and restrictions when you win

The sharp-price method finds bookmakers that are slow to move their prices, and those bookmakers know it. Most sportsbooks' terms allow them to reduce the maximum stake on an account, or close it, at their discretion, and customers who regularly take prices above the sharp fair line are among the most likely to be limited. An approach that depends on one or two slow bookmakers may not last, so judge it on the stakes you can actually place, not on the ones you could place at full size.

Voids are the second practical cost. A price far out of line with the rest of the market may be settled as an obvious error, often called a palp, under the bookmaker's terms. That removes exactly the bets with the biggest apparent edge, which is one more reason to distrust very large EV figures.

Do not try to get round a restriction. Opening accounts in other people's names or hiding where you are betting from breaks the terms, can lead to winnings being withheld and may break the law. Use only operators licensed where you live: betting law differs by country and, in the US, by state, and India has prohibited online money games, including betting, since 1 May 2026.

None of this changes the arithmetic of EV. It changes how much of it you can actually collect.

Keeping records of expected value

Records for +EV betting need one column most bettors leave out: the expected value at the moment of the bet. Log each bet with:

• date, sport, market and bookmaker
• odds taken and stake
• the sharp fair probability you used and the margin removal method
• EV in per cent and in money
• the fair closing price, once the event starts
• result and profit or loss

Add up the EV column and you have an expected profit line to set against actual profit. Suppose that after 400 bets of $100 at an average edge of 2.5%, your expected profit is $1,000 but you are $300 down. That looks bad. Yet with a standard deviation of about $2,000 over 400 bets near even money, a $1,300 shortfall is well within normal luck. It does not prove the method works either. The closing price column is the better early test: if the fair closing price is regularly shorter than the price you took, your EV figures were probably real.

Split the figures by sport, market and bookmaker. A positive overall number can hide one market where your fair prices are consistently wrong, often a thin market where the sharp price was not very sharp.

Before you place a +EV bet

Run the same routine on every bet, so your records measure the method rather than your mood.

• Take the fair probability from a market you have reason to trust, as close to the start as possible, and note how you removed the margin.
• Convert the price you can get to decimal and calculate EV = (p × odds) - 1.
• Check that the edge clears your minimum and survives a one-point error in p.
• Confirm the price has not moved since you did the sum.
• Size the stake from a fixed bankroll, with a flat stake or a fraction of Kelly.
• Log the bet with its EV, then add the fair closing price later.

A bet that passes every step can still lose, and a run of them can lose for months. If betting stops feeling under control, our responsible gambling page lists limit tools and places to get support.