↓ Skip to main content

Positive EV Betting: How It Works and Where It Breaks Down

·5 mins

Positive EV betting means only placing bets where the price you get is better than your best estimate of the fair price. In practice most +EV bettors don’t build models. They take a market they consider efficient, remove its margin to get a fair probability, and look for other books offering a better price on the same outcome. The edge per bet is usually small (often low single-digit percentages), the results are noisy, and the bettors who do it consistently tend to run into stake limits. All three points matter as much as the math.

Odds on this page are illustrative examples, not live prices.

The method in four steps

  1. Pick a reference market. Usually a book or exchange that takes large bets and moves quickly on information.
  2. Remove the margin. Convert its two prices to probabilities and normalize them to 100%. That is the fair probability. The methods are in no-vig odds.
  3. Compare with other books. Find a price at another licensed book that pays more than the fair odds.
  4. Compute the edge. EV per $1 = fair probability × decimal odds − 1.

Worked example

Reference market: Team A -120, Team B +105.

  • Implied: 54.55% and 48.78%. Sum = 103.33%.
  • Fair (multiplicative): Team A 54.55 / 103.33 = 52.79%; Team B 48.78 / 103.33 = 47.21%.
  • Fair odds: Team A -111.8, Team B +111.8 (decimal 2.118).

Another book offers Team B at +115 (decimal 2.15).

EV = 0.4721 × 2.15 − 1 = +0.0150, so +1.50% per dollar staked, or $1.50 per $100.

If the same book has Team A at -110 (decimal 1.909): 0.5279 × 1.909 − 1 = +0.78%. Smaller, but still above fair on this method.

Those numbers rest on one assumption: that the reference market’s de-vigged price is close to the truth. Change the de-vig method or the reference book and the edge changes. A 1.5% edge can disappear inside the gap between methods.

Why this can work at all

Bookmakers don’t all move at the same speed, and consensus prices are informative. Lisandro Kaunitz, Shenjun Zhong and Javier Kreiner built a strategy on exactly this idea for soccer: use the average odds across bookmakers as the probability estimate and bet when one bookmaker’s price is out of line. In their 2017 paper they report a 3.5% return in a 10-year simulation on closing odds and an 8.5% return across 265 real-money bets over five months.

The same paper is the clearest warning about what happens next (see limits below).

The variance problem

A 1.5% edge is real money over many bets and almost invisible over a few. Using the +115 example (true probability 47.21%, $100 per bet), the exact binomial odds of being behind:

Bets Expected profit Chance you are behind
100 +$150 44.4%
250 +$376 42.4%
1,000 +$1,502 33.8%
5,000 +$7,511 16.1%

The standard deviation of a single $100 bet here is about $107, roughly 70 times the expected profit per bet. After a thousand bets, a genuinely +EV bettor still has about a one-in-three chance of showing a loss. That is why the usual measure of progress is not profit but closing line value, which tells you whether your prices were good long before results converge.

Where it breaks down

1. Limits. Kaunitz and colleagues write that a few months into real-money betting, bookmakers “started to severely limit our accounts”, cutting allowed stakes below their fixed $50 bet or requiring manual review, and they ended the experiment. US licensed books reserve the same power in their own rules:

  • DraftKings’ Massachusetts house rules: it “may lower the Maximum Payout limit for any customer” at its sole discretion, “for any reason or no reason” (implemented Aug 26, 2025).
  • BetMGM’s Massachusetts house rules: it “will determine minimum and maximum wager amounts per patron” (effective Mar 28, 2025).

Both checked Nov 5, 2026. A strategy that depends on stale prices is a strategy the books have every incentive to stop.

2. Voided “errors”. The biggest apparent edges are often mistakes. DraftKings’ Massachusetts rules define an “Error” to include bets at odds “meaningfully different” from what similarly situated US sportsbooks offer, and BetMGM’s rules list wagers at odds “materially different” from the general betting market as obvious errors it may void or resettle (both subject to Massachusetts Gaming Commission approval). A 15% “edge” is more likely a voided bet than a windfall.

3. Model risk. Your fair price is an estimate. If the reference market is wrong, every bet you place against it inherits that error.

4. Speed. +EV prices exist because one book is slow. By the time you see it, it may have moved.

5. Vig on the reference. The margin you remove is not spread evenly in any way you can observe. Different de-vig methods can disagree by more than your edge, especially on longshots.

6. Taxes and costs. Winnings are taxable, and tool subscriptions come out of a small edge. Run your own numbers.

Doing it with software

Most +EV tools automate steps 1 to 4 across many books and alert you to prices above fair. They differ in which books they cover, which reference markets they use, which de-vig method they apply and what they charge. We compare those published details in best EV betting tools. Before relying on any tool, find out how it de-vigs and which market it treats as “sharp”, because that choice is the edge.

A realistic checklist

  • Record the fair probability, the price taken and the closing price for every bet.
  • Treat edges under about 2% as fragile; a different de-vig method may erase them.
  • Be skeptical of very large edges; they are often errors that can be voided.
  • Size each bet as a small fraction of bankroll; variance at these edges is large.
  • Expect limits if you are consistently on the right side of stale prices.

Only adults 21 and over (or your state’s legal age) can bet. If you or someone you know needs help, call or text 1-800-MY-RESET, and see our responsible gambling page.