Expected value in one line
Expected value is what a bet returns on average per unit staked: EV = (chance of winning × odds) − 1. If a coin flip is offered at odds of 2.10 when the fair price is 2.00, the chance of winning is 50%, so EV = 0.5 × 2.10 − 1 = +5%. Stake 100 kr and you expect to make 5 kr on average.
'On average' matters. The bet still loses half the time. +EV is a statement about the price, not about the next result.
A worked example, step by step
Say Pinnacle prices a tennis match at 1.95 for each player, and a Swedish bookmaker offers 2.10 on the first player.
- Pinnacle's implied chances: 1 ÷ 1.95 = 51.3% for each player, 102.6% together. The 2.6% above 100 is the margin.
- Remove the margin: each player's fair chance is 51.3 ÷ 102.6 = 50%, so the fair odds are 2.00.
- The edge at 2.10: 0.50 × 2.10 − 1 = +5%.
- The stake: full Kelly is (2.10 × 0.50 − 1) ÷ (2.10 − 1) = 4.5% of the bankroll, so half Kelly is about 2.3%.
Then check the price you can actually get. If the bookmaker has already cut 2.10 to 2.02, the edge is down to +1%, and most of the reason to bet is gone. With uneven prices, such as a favourite against an outsider, the margin isn't split evenly; Sharp Bets uses Shin's method, which puts more of it on the outsider.
Where do the fair odds come from?
Nobody knows the true chance of a football match. What you can do is borrow the best estimate available: the odds at Pinnacle, a bookmaker known for low margins that welcomes sharp bettors and moves its prices quickly when smart money arrives.
Every bookmaker builds a margin (the 'vig') into its odds, so the implied chances add up to more than 100%. Removing that margin, with a method such as Shin's, leaves an estimate of the fair chance for each outcome. Compare that with the odds at another bookmaker and you can see where the price is too generous.
The idea has been tested in public. In a 2017 study, Lisandro Kaunitz, Shenjun Zhong and Javier Kreiner used the average of the market's own odds as the fair price and bet wherever a bookmaker offered more. It was profitable over ten years of historical closing odds and over five months of betting with real money, and the paper describes how the bookmakers then restricted their accounts.
Market value and model value
Some tools call a bet value when their own prediction model rates a team higher than the odds do. That is a different thing from what's described above. A model is one opinion; a liquid market like Pinnacle's is thousands of informed opinions, including team news the model may not have.
When a model claims an edge of 30 or 40% on a thin market, such as a lower league or a women's international, the far likelier explanation is that the model is wrong, not that the bookmakers missed something that big.
Value measured against the market is smaller, usually a few percent, and it can be checked: if the price you took keeps beating the closing line, the value was real. Ask any tool which kind of value it shows, and whether it can show its closing-line record.
When the edge looks too good
Free value-bet lists often lead with edges of 50% or more, nearly always on long odds in small leagues. Those are almost never value. A price that far from the market is usually a line the bookmaker hasn't updated yet, a typing error, or two different markets compared as if they were the same. Bookmakers' terms let them void bets placed at an obvious error, so the 'edge' tends to disappear after you've bet.
Real value against a sharp market is mostly a few percent, now and then 10 to 20%. Above that, check the same market at Pinnacle and see whether the price is still there a minute later before you trust it. The closing line settles it afterwards: an error closes nowhere near the price you took.
How big real edges are: our numbers
Of the 5,366 picks Sharp Bets has posted, the median edge when found was 3.06%. 18.7% had an edge of 5% or more, and only 2.9% reached 10%. Most real value is small.
Not all of it lasts until the start, either. On the 4,954 picks with both numbers, the average edge when found was 3.97%, and the average closing-line value was 2.61%: about 66% of the edge was still there at the close. Part of the rest is the market moving after the price was found, and part is the first estimate being a little generous. So judge a price by what it's likely to be worth at the close, not by the number on the screen.
Why one bet proves nothing
A 5% edge is small next to the swings of luck. Over 100 bets at odds around 2.00 a fair-priced bettor's profit routinely lands 10 units above or below expectation. You only see the edge over hundreds of bets, which is why the closing line (see our guide to closing line value) is a better early signal than profit.
Sharp Bets' own record shows how slow profit is to prove anything. With the spread of results our picks have, and an edge the size of our average closing-line value, profit alone needs around 12,000 bets before it can rule out luck. The closing line gets there in about 60.
The luck meter on My Bets does this calculation for you: it shows the range your results should fall in and whether you're inside it.
Staking: how much per bet?
The Kelly criterion sizes a stake to the edge: f = (odds × chance − 1) ÷ (odds − 1). Full Kelly is aggressive, so most people use half Kelly or less. Sharp Bets suggests stakes in units: one unit is a fixed slice of your bankroll, typically 1%, so the advice works whatever your bankroll is.
Common mistakes
- Chasing the biggest number on the list. Edges far above the rest are usually errors (see above), and they're the bets most likely to be voided.
- Betting a price that has already moved. The edge belongs to the odds you actually get, so check them first; on Sharp Bets, + Log asks for the odds you got and tells you whether the edge is still there.
- Comparing different markets. A match winner that includes overtime isn't the same bet as one settled after regular time, and an Asian handicap isn't a European one. Fair odds only apply to the exact same market.
- Staking by feel. A bigger stake on the pick you like most undoes the maths. Size every stake from the edge and the odds.
- Judging the method by last week. A few dozen results say almost nothing; the closing line and a few hundred bets do.
The practical catch
Prices move, often within minutes: in mid-October 2026 about one in three prices on Sharp Bets' board had fallen below the edge threshold by the next check, 2 to 15 minutes later. So check the odds you can actually get before you bet. And bookmakers limit winning customers, so a real edge tends to have a shelf life on any single account.
Betting is never risk-free, and a positive edge doesn't stop long losing runs. Only stake what you can afford to lose.
Common questions
Is +EV betting the same as value betting?
Yes. Both mean betting at odds that are higher than the fair odds for the outcome. +EV (positive expected value) is the maths term, value bet the everyday one.
How big is a normal edge?
A few percent per bet. Sharp Bets' picks average +2.61% closing-line value over 4,954 checked picks. An edge of 20% or more is rare and is usually a data error rather than real value.
Is +EV betting legal in Sweden?
Yes. Choosing which odds to bet on is ordinary betting, and winnings at Swedish-licensed bookmakers are tax-free. A bookmaker may limit your account if you keep winning, which its terms allow.
How many bets before the edge shows?
Usually several hundred. Over 100 bets at odds around 2.00, luck alone moves the result about ten units either way, so closing-line value is the earlier and more reliable signal.
Why is the edge smaller at the start than when the pick was found?
Prices keep moving after a pick is found. On Sharp Bets' picks the average edge when found was 3.97%, and 2.61% was left at the close. The closing figure is the better guide to what the price was really worth.