Two words get thrown around constantly in betting: edge and expected value.
They are related. They are not the same number.
The classic mix-up goes like this: "I make it 55%, the book needs 52.4%, so I've got a 2.6% edge — that's a 2.6% return." It isn't. A probability edge is measured in percentage points of win probability. Expected value is measured in money, and expected ROI is that money as a share of what you risk.
Same bet. Different questions. And the one thing that connects them is the price.
The three probabilities in every bet
Before any EV math, keep three numbers apart:
- Break-even probability — set by the price you can actually bet. At -110 it is 52.38%. (See How Implied Probability Works.)
- No-vig market probability — the market's centered view with the sportsbook margin removed. (See How to Remove the Vig.)
- Your probability — what your own handicap says.
Probability edge is your probability minus the break-even probability.
Expected value puts your probability together with the payout.
The whole idea in one line
OutpriceTheBook · Education
Probability
How often you believe it wins
Price
What the book pays when it wins — and what you lose when it doesn't
Expected value
The average result per bet if your probability is right
A simple EV calculation
Take a -110 price and an estimate of 55%.
At -110, a $100 risk wins $90.91 in profit. If your estimate is 55%, you lose the other 45% of the time.
EV = (win probability × profit if win) - (loss probability × stake)
EV = (0.55 × $90.91) - (0.45 × $100)
EV = $50.00 - $45.00 = +$5.00
Expected ROI is $5.00 ÷ $100 = +5.00%.
So this one bet has:
- Break-even probability: 52.38%
- Your probability: 55.00%
- Probability edge: +2.62 percentage points
- Expected value per $100: +$5.00
- Expected ROI: +5.00%
The edge and the ROI are not the same number. Here's why.
Why edge and EV are different numbers
There is a clean relationship between them:
Expected ROI = probability edge × decimal odds
Decimal odds are your total return per $1 risked, stake included. -110 is about 1.909 in decimal.
2.62 points × 1.909 ≈ 5.00%
That one line explains everything. The same probability edge is worth more at longer odds, because the payout is bigger. A 4-point edge on a +150 underdog (decimal 2.50) is a 10% expected ROI. A 4-point edge on a -150 favorite (decimal 1.667) is only about 6.67%.
So keep the two numbers doing different jobs:
Probability edge: How far is my estimate above the hurdle the price sets?
Expected value: What is that gap worth per bet, at this payout, if my estimate is right?
Same probability, different prices
Hold your opinion fixed at 55% and change only the price.
Expected value, side by side
OutpriceTheBook · Education
One opinion — 55% — at three different prices
Price A
-110at 55% to win
Break-even52.38%
Edge (pts)+2.62
Win pays ($100)$90.91
+$5.00 EV
Expected ROI +5.00%
Price B
-120at 55% to win
Break-even54.55%
Edge (pts)+0.45
Win pays ($100)$83.33
+$0.83 EV
Expected ROI +0.83%
Price C
-130at 55% to win
Break-even56.52%
Edge (pts)-1.52
Win pays ($100)$76.92
-$2.69 EV
Expected ROI -2.69%
At -130, break-even jumps to 56.52%, a win pays only $76.92, and EV becomes:
EV = (0.55 × $76.92) - (0.45 × $100) = $42.31 - $45.00 = -$2.69
The team didn't change. Your estimate didn't change. Only the price did — and it turned a +EV bet into a -EV one.
Same price, different probabilities
Now flip it. Hold the price at +120 (break-even 45.45%) and change only your estimate.
Expected value, side by side
OutpriceTheBook · Education
One price — +120 — at three different estimates
You say 40%
+120at 40% to win
Break-even45.45%
Edge (pts)-5.45
Win pays ($100)$120.00
-$12.00 EV
Expected ROI -12.00%
You say 45%
+120at 45% to win
Break-even45.45%
Edge (pts)-0.45
Win pays ($100)$120.00
-$1.00 EV
Expected ROI -1.00%
You say 50%
+120at 50% to win
Break-even45.45%
Edge (pts)+4.55
Win pays ($100)$120.00
+$10.00 EV
Expected ROI +10.00%
This is why the quality of the estimate matters so much. Moving your number from 45% to 50% doesn't feel like much. At +120 it is the difference between a small losing bet and a +10% expected ROI.
A favorite example
You believe a favorite wins 70% of the time.
- At -180, break-even is 64.29%. A win pays $55.56 on $100. EV = (0.70 × $55.56) - (0.30 × $100) = +$8.89.
- At -260, break-even is 72.22%. A win pays $38.46. EV = (0.70 × $38.46) - (0.30 × $100) = -$3.08.
You can be right that the favorite is much more likely to win and still make a bad bet. Betting isn't a contest to identify the likeliest winner. It is a question of whether the payout compensates you for the chance of losing.
An underdog example
An underdog is +150, so break-even is 40.00%. You make it 44%.
EV = (0.44 × $150) - (0.56 × $100) = $66.00 - $56.00 = +$10.00
That's a +10% expected ROI on a bet you expect to lose 56% of the time. The edge comes entirely from being paid +150 for something you believe happens 44% of the time.
Positive EV does not mean likely to win. It means that, if your probability is right, repeating the same bet at the same price would come out ahead on average.
Liking a side and still being -EV
This is the most common way good handicappers lose money.
You like Team A. Your work says 58%. The best price you can find is -150, which needs 60.00%.
EV = (0.58 × $66.67) - (0.42 × $100) = $38.67 - $42.00 = -$3.33
You were right about the team. You were wrong to bet it — at that price. Passing on a side you like is not a lack of conviction. It's reading the price correctly.
Price shopping changes EV, not your opinion
Same 55% opinion, two books:
- Book 1 offers -115: break-even 53.49%, EV +$2.83 per $100
- Book 2 offers -105: break-even 51.22%, EV +$7.38 per $100
You didn't learn anything new about the game. You found a better price, and the value of the same opinion more than doubled. Over hundreds of bets, that gap is often the difference between a winning record and a losing one.
"A 60% chance to win" means nothing without the price
People say "I think they win 60% of the time" as if it were a recommendation. It isn't one yet.
- At -130 (break-even 56.52%): EV +$6.15 per $100. A good bet if the 60% is real.
- At -170 (break-even 62.96%): EV -$4.71 per $100. A bad bet with the exact same opinion.
A probability only becomes a decision after you pair it with a price.
No-vig vs break-even: two different hurdles
Suppose the no-vig market makes a side 50%, but your book asks -110, which requires 52.38%.
If you estimate the side at 54%, you are:
- 4 points above the no-vig market,
- but only 1.62 points above the price you actually pay.
"My model is four points above the market" says you disagree with the market. It does not tell you the EV of the bet. Only the price you can actually get does that.
The formula doesn't know if your number is good
Type 70% into any EV calculator and almost every reasonable price looks attractive. The calculator will do exactly what you tell it. That doesn't make 70% credible.
A probability should come from something sturdier than confidence, a hot streak or a narrative — predictive team strength, confirmed availability, the playing environment, the specific matchup. If the input is weak, the output is just a confident-looking number. How to Build Your Own Fair Betting Line covers how to build that input.
Estimating your probability honestly
The hardest part of EV isn't the arithmetic. It's being honest about how good your probability really is.
Three habits help:
- Start from the market, then move away for a reason. The no-vig market is a strong default. If you can't name the specific information that moves you off it — an availability change, a matchup, a weather effect — your estimate probably shouldn't be far from it.
- Shrink big disagreements. When your number is miles away from the market, the most likely explanation is usually that you're missing something, not that the market is asleep. A large gap deserves more checking, not a bigger bet.
- Write the number down before you see the price. Once you've seen -150, it is remarkably easy to "decide" the team is a 62% shot.
An honest 54% beats an optimistic 58% every time, because only one of them is real.
Small edges, big swings
Even a genuinely good bet loses often. That's what makes EV hard to trust in the short run.
Take the first example — a true 55% shot at -110, worth +$5.00 per $100. Bet it 200 times at $100 and the expected profit is $1,000. But the results spread widely around that number.
Working through the exact binomial math, a bettor with that real edge is still behind after:
| Number of bets | Expected profit | Chance of still being behind |
|---|---|---|
| 100 | +$500 | about 31% |
| 200 | +$1,000 | about 22% |
| 500 | +$2,500 | about 11% |
| 1,000 | +$5,000 | about 5% |
Roughly one bettor in five with a real 5% edge is still losing money after 200 bets. That's not bad luck in any special sense — it's what normal variance looks like. It's also why you can't judge a handicapper, or yourself, on a few weeks of results.
EV is not a bet size
A positive EV tells you the bet is worth making at all. It doesn't tell you how much to put on it.
Sizing depends on things EV ignores: how uncertain your probability is, how much you already have riding, whether your bets are correlated, and how deep a losing run you can sit through. A +10% EV bet built on shaky inputs can deserve a smaller stake than a +3% bet you understand well.
Keep the two decisions separate. First, is this bet worth making? Then, at what size does it fit my risk? The Daily Risk & Exposure tool handles the second question against limits you set yourself.
Results don't grade the decision
A +EV bet that loses wasn't a mistake. A -EV bet that wins wasn't a good decision. One game is one draw from a range of outcomes.
That's why a process has to be judged over a large sample. A useful short-run signal is whether you consistently get better prices than the market closes at. If you regularly bet numbers that move in your direction by game time, your timing and pricing may be finding value early. It isn't proof on its own, but it says more than one night's record.
Common mistakes with EV
Calling probability edge "EV"
A 3-point edge is not a 3% return. Multiply by the decimal odds.
Forgetting the price you actually get
An estimate becomes a bet only at a specific price. Re-check it before you bet.
Treating your estimate as certain
The calculator can't see how shaky your inputs are. You have to.
Sizing up because the EV looks big
A big calculated edge from weak information isn't a strong bet. Bad inputs make impressive outputs.
Grading EV by tonight's result
EV is a long-run concept. Tonight is one outcome.
A practical way to use both numbers
- Convert the price you can get into break-even probability (Odds Converter).
- Check the no-vig market so you know where the market is centered (No-Vig Calculator).
- Build your own probability.
- Calculate the probability edge and the EV at the real price (EV Calculator).
- Ask how much you trust the inputs behind your probability.
- Re-check the price right before you bet.
The takeaway
Probability edge tells you how far your estimate is from the price.
Expected value tells you what that gap is worth at the payout you're getting.
The price connects them, and the price is the one thing you can always improve without knowing anything new about the game. Good betting decisions are about buying probabilities at good prices — not demanding certainty from uncertain events.