Arbitrage Betting Explained: How It Works and Its Limits

Published September 5, 2026 · 9 min read · By the OddsLighthouse Editorial Team

Every so often, two sportsbooks disagree with each other by just enough that betting both sides of the same game locks in a profit no matter who wins. That's the entire idea behind arbitrage betting, and it sounds, at first pass, like a math trick that shouldn't exist in a business built on taking the other side of your money. It does exist — but it's rarer, smaller, and more fragile than it sounds, and understanding exactly why is more useful than the pitch most "guaranteed profit" videos give you.

What arbitrage betting actually is

Every priced bet implies a probability. A moneyline of -110 implies a win probability of about 52.4%; a moneyline of +130 implies about 43.5%. If you add up the implied probabilities of every outcome in a single market at a single book, the total normally comes out above 100% — that gap above 100% is the book's built-in margin, sometimes called the vig or the hold. Arbitrage betting looks for the opposite situation: a case where two different books, pricing the same game independently, end up with implied probabilities that sum to less than 100% once you take the best number available on each side. When that happens, there's a way to split your money across both outcomes such that you come out ahead regardless of the result — a "middle" between two books' own margins rather than inside either book's own hold.

This only works across separate books. Betting both sides of a market at the same book never produces an edge, because a single book's own prices are built to keep that total above 100% for exactly this reason. Arbitrage requires two books that don't coordinate with each other and have, for whatever reason, landed on prices that no longer agree.

A full worked example with real numbers

Say an NFL underdog is priced at +130 at one sportsbook, while the favorite in the same game is available at -110 at a second book. Converting each to decimal odds and implied probability:

Add those two implied probabilities together and you get 95.9% — a full 4.1 percentage points under 100%. That gap is the arbitrage margin. To lock in a matching profit no matter which side wins, you don't split your money 50/50; you split it in proportion to each side's implied probability. With a combined $1,000 to risk: stake roughly $453 on the underdog at Book A and roughly $547 on the favorite at Book B.

That 4.3% figure is unusually generous for a real-world example — genuine arbitrage windows in mainstream US markets are far more often in the 1-2% range, and they close within minutes as books adjust to each other's numbers.

Why these gaps show up between books at all

Sportsbooks don't set prices by consulting each other in real time. Each book manages its own liability based on its own customers' betting patterns, its own risk models, and how quickly its own trading team reacts to news. A book that's taken heavy action on one side of a game may shade its own line to balance its book, even if that shading temporarily disagrees with where a competitor's line sits. Promotional pricing — a boosted odds offer on a specific bet, for instance — can also create a short-lived arbitrage window against a book running standard pricing on the same game. None of this is a flaw in any individual book's math; it's simply what happens when dozens of independently operated pricing engines react to the same event on different timelines.

These gaps tend to be largest on lower-liquidity markets — player props, smaller conferences, in-play lines right after a big scoring play — because fewer sharp bettors are actively trading those markets back toward consensus. Mainstream NFL and NBA point spreads and game totals, by contrast, get so much betting volume and so much attention from professional bettors that meaningful arbitrage gaps rarely last long enough to act on.

The practical limits that keep it from being free money

The math above is real, but four practical problems keep arbitrage from being the risk-free income stream it's sometimes marketed as.

• Books limit or ban arbers. Sportsbooks track betting patterns closely, and a customer who consistently bets both sides of games across different accounts looks nothing like a typical recreational bettor. Most major US books will quietly reduce that customer's max bet sizes or restrict their account once the pattern is clear, which caps how much volume any one arbitrage bettor can realistically run through a given book over time.
• Timing risk is real. The price you see when you calculate an arb isn't guaranteed to still be there by the time your second bet is placed. Odds can move or a book can void a bet for a pricing error in the seconds it takes to place two separate wagers at two separate books, turning what looked like a locked-in profit into an unbalanced, one-sided risk.
• Capital gets tied up across multiple accounts. Running real arbitrage volume means keeping funded accounts open at several different sportsbooks simultaneously, plus managing withdrawals and deposits across all of them — overhead that eats into the return on a strategy that's already only clearing a percentage point or two per opportunity.
• Opportunities are small and infrequent. A 4% margin like the example above is well above what most real arbitrage windows offer, and a bettor has to move fast and check odds across many books constantly to catch the smaller, more common 1-2% gaps before they close.

None of this means arbitrage betting is a myth — the math is sound, and the strategy genuinely exists and gets used, including by software tools built specifically to scan many books for these gaps in real time. It means the "guaranteed profit, no risk" framing skips past the operational friction that determines whether it's actually worth doing at the scale most people imagine.

Arbitrage vs. line shopping vs. middling

These three ideas get confused with each other because they all involve comparing prices across books, but they're structurally different. Line shopping is simply betting one side of a game at whichever book offers the best price for that side — it doesn't require betting both sides, and it's a good habit for every bettor, arbitrage or not. Middling involves betting both sides of a game at two different point-spread numbers (say, -2.5 at one book and +4 at another) hoping the final score lands in the gap between them, which would win both bets; unlike arbitrage, middling isn't risk-free — if the score doesn't land in that specific window, you can lose one side while only pushing or winning the other. Arbitrage is the only one of the three built to guarantee a profit purely from mismatched moneyline or spread pricing, independent of the final score.

Is it worth setting up for yourself?

For a casual bettor, manually scanning multiple books for arbitrage opportunities is a lot of ongoing effort for a return that's usually a small fraction of what a $1,000 bankroll could earn elsewhere, once you factor in the time spent, the number of accounts you need open, and the real chance that a book limits your action once a pattern shows up. For someone who already treats betting as a numbers-driven side activity and doesn't mind managing several sportsbook accounts at once, it can be a legitimate, low-variance way to generate a modest return — but it's a logistics-heavy strategy, not a shortcut, and it works best paired with the same habits described in our guides on how betting odds work and closing line value, since all three rely on the same underlying skill: reading what a price actually implies before you bet it.