Almost every bet needs you to be right about something. An arbitrage bet, or arb, is the exception. You bet both sides of the same market at two different sportsbooks, at prices that leave you ahead whichever side wins.
Here's the shape of one. Book A has a pitcher at over 6.5 strikeouts, priced at +120. Book B has the same pitcher at under 6.5 strikeouts, priced at -110. Put $465 on the over and $535 on the under, and this is what happens:
| Result | Over 6.5 at +120 ($465) | Under 6.5 at -110 ($535) | Net |
|---|---|---|---|
| 7 or more strikeouts | Win $558 | Lose $535 | +$23 |
| 6 or fewer strikeouts | Lose $465 | Win $486 | +$21 |
You staked $1,000 in total and you finish about $22 ahead either way. You never had to have an opinion on the pitcher.
Why arbs exist
No single sportsbook will offer you both sides of a market at prices like that. A book's own over and under are priced so that the book keeps a margin whichever side wins.
Arbs come from two books disagreeing. As we covered in line shopping for player props, books build their numbers differently and some are slower to move than others. When one book has moved on a piece of news and another hasn't, the best over at one and the best under at the other can add up to a profit for you instead of a margin for them.
Player props produce more of these than spreads and totals do. They get less attention, limits are lower, and a stale number can sit for a while before anyone corrects it.
How to tell if two prices make an arb
Convert each price to its implied probability, which is the win rate you'd need to break even at that price, then add the two together.
| Odds | Implied probability |
|---|---|
| +120 | 45.5% |
| +105 | 48.8% |
| +100 | 50.0% |
| -105 | 51.2% |
| -110 | 52.4% |
| -120 | 54.5% |
If the two sides add up to more than 100%, there's no arb. If they add up to less than 100%, there is one, and the further below 100% the bigger the return.
| Side one | Side two | Combined | Return on total stake |
|---|---|---|---|
| -110 | -110 | 104.8% | No arb |
| -105 | +105 | 100.0% | Break even |
| +100 | +105 | 98.8% | 1.2% |
| -110 | +120 | 97.8% | 2.2% |
| -120 | +135 | 97.1% | 3.0% |
The first row is what a single book offers you on its own market. The rest only show up when you take each side from a different book.
Sizing the two bets
An arb only works if the stakes are in the right proportion. The aim is for both bets to return the same amount, so the result doesn't matter.
The rule: each side's share of your total stake is its implied probability divided by the combined figure.
In the strikeout example, the under at -110 is 52.4% and the over at +120 is 45.5%, for a combined 97.8%. On a $1,000 total:
- Under: 52.4 ÷ 97.8 = 53.5% of the total, so about $535
- Over: 45.5 ÷ 97.8 = 46.5% of the total, so about $465
Both bets then return roughly $1,022. Get the split wrong and you haven't locked in anything. You've made a lopsided bet on one side with a small hedge on the other.
Check the line, not just the price
On player props, both sides have to be on the same line. Over 6.5 at one book and under 6.5 at another covers every result.
If the lines differ, it's a different trade. Over 6.5 and under 7.5 leaves a window where both bets win, which is a middle. Over 7.5 and under 6.5 leaves a hole: if he finishes on exactly 7, both bets lose. Prices that look like an arb mean nothing if the lines leave a gap.
What can go wrong
Arbs are low risk, not no risk. The math is certain. Getting both bets placed as planned is not.
The second price moves before you bet it. You place one side, and by the time you open the other app the price has changed. Now you're holding a single bet you never wanted. Place the side you think is more likely to move first, and have the second book open and ready.
One book voids its side. Books can cancel bets taken at a price they consider an obvious mistake. If one side is voided and the other stands, you're left with a straight bet. A price that is far off the rest of the market is the likeliest to be cancelled.
The two books settle differently. If the player doesn't play, most books void the prop, but the rules vary by book and by sport. A bet that's void at one book and live at the other is no longer an arb. Read both books' prop rules before you rely on them.
Limits. The stake you need on one side may be more than that book will take. A partial fill leaves the two sides out of proportion.
Fees. Prediction-market exchanges charge fees. A 2% fee on winnings turns +105 into roughly +103, and that can be the whole margin on a thin arb. Work out the return after fees.
Your account. Each side of an arb is an ordinary bet at a licensed book. Books don't have to welcome it, though. Customers who only ever bet a book's off-market prices tend to see their limits cut.
How big are arbs, really?
Small. Most are a percent or two of the total staked. That's why arbers think in terms of turnover: a 2% return isn't much on one bet, but the same bankroll can be used again as soon as the bets settle.
Be suspicious of big ones. An arb showing 8% or 10% is far more often a mistake at one of the books, a mismatched line or a stale price than a gift. Those are also the ones most likely to be voided.
How to find arbs
Compare every book on the same prop. An arb only appears when you can see the best over and the best under side by side, across books.
Look right after news. Injuries, lineup changes and pitching scratches move some books faster than others. The gap between the fast books and the slow ones is where arbs open up.
Be quick, then check again. Arbs close in minutes. Before you place the first bet, confirm both prices are still live in the books' own apps.
Keep money in several books. You can't bet the other side if your balance is sitting somewhere else.
Keep records. Log both sides, the stakes and the result. It's the only way to see what you're really making after voids, moved prices and fees.
Finding them without the spreadsheet
Hunting arbs by hand means checking every prop at every book, converting prices and working out stakes before the gap closes. Most are gone by the time you've done the arithmetic.
Betravon's EdgeFinder runs that scan across the books Betravon covers and lists cross-book arbs on one board, with the return worked out after exchange fees and the stake for each side. You still check the prices in each book's app, and you still place the bets yourself. You just don't have to find the gap or do the division.
Whichever way you find them, the rule is the same: an arb is only an arb once both bets are placed, on the same line, at the prices you saw.