Arbitrage Betting Explained (and Why It's Harder Than It Looks)

By Cobia Picks · Updated July 3, 2026 · ~7 min read

Arbitrage betting — "arbing" — is the closest thing to a sure thing in this whole space, which is exactly why it sounds too good to be true and mostly is. The idea is simple: when two venues price the same event differently enough, you back every outcome across those venues so that no matter who wins, your total payout exceeds your total stake. A locked profit, decided before the game even starts. This article walks through what an arb actually is, a worked two-way example with the stake math, where these gaps come from, and the unglamorous reasons they're far harder to cash than a spreadsheet makes them look.

What an arbitrage actually is

A two-way event has two outcomes — say Team A wins or Team B wins. Any single venue prices those two outcomes so that their implied probabilities add up to more than 100%. That extra slice is the vig, the house's cut, and it's why you can't just bet both sides on one book and profit.

An arb appears when you can assemble the two sides from different venues such that the combined implied probability drops below 100%. When that happens, the two prices are "off" relative to each other, and you can carve up your stake so every outcome pays back more than you laid out in total.

The quick test: add up the implied probabilities

Convert each price you're getting into an implied probability, add them, and if the sum is under 1.00 (100%), you have an arb. The size of the gap under 100% is roughly your locked margin. In decimal-odds terms, take 1/odds for each side and sum them; under 1.00 means arb.

A worked two-way example

Suppose the same game is priced two ways:

• Venue 1: Team A to win at decimal 2.10 (implied 1/2.10 = 47.6%).
• Venue 2: Team B to win at decimal 2.05 (implied 1/2.05 = 48.8%).

Add the implied probabilities: 47.6% + 48.8% = 96.4%. That's under 100%, so there's an arb worth about 3.6% before costs.

Splitting the stake

Say you have $1,000 total to deploy. To lock the profit, you split it so both outcomes return the same amount. The stake on each side is proportional to that side's implied probability:

• Team A (Venue 1): $1,000 × (0.476 / 0.964) = $493.78 at 2.10 → returns $493.78 × 2.10 = $1,036.94.
• Team B (Venue 2): $1,000 × (0.488 / 0.964) = $506.22 at 2.05 → returns $506.22 × 2.05 = $1,037.75.

Whichever team wins, you get back roughly $1,037 on a $1,000 outlay — a ~3.7% locked profit, no prediction required. On paper that's flawless. Now for why "on paper" is doing a lot of work in that sentence.

Where arbs come from

Arbs exist because no two venues price identically at the same instant. The gaps typically come from:

• Line differences between books. Different sportsbooks weight news, injuries, and money flow differently. One shades a line before another catches up, and for a few minutes the two disagree.
• Slow reactions to news. A late scratch or lineup change moves one venue instantly and another with a lag. The stale side is temporarily mispriced.
• Prediction-market vs sportsbook gaps. This is the interesting one. A prediction market like Polymarket is priced by traders buying and selling shares, not by a book setting a line. Its price for "Team A wins" moves on order flow and can drift meaningfully from a sportsbook's number — especially on niche games, overnight, or right after news. When the market's implied probability and the book's disagree, a cross-venue arb opens up.

That last point is why prediction markets matter to arbers: they're a genuinely different pricing mechanism sitting alongside traditional books, so cross-venue divergences show up more often than between two sportsbooks that copy each other. But those same markets have their own frictions — spreads, liquidity limits, and settlement rules — which feeds directly into the honest part.

The honest reality: why it's harder than it looks

If arbing were as clean as the math, everyone would do it and it would vanish. Here's what the worked example above quietly ignores:

The margins are thin

Real arbs are usually well under 3% and frequently under 1%. Fees, spreads, and the small price you give up crossing a bid-ask can erase a 1% edge entirely. A "3.6% arb" that looks fat on screen can be a break-even or losing trade once every cost is paid.

Limits and getting booked out

Sportsbooks hate arbitrageurs. Bet into stale lines often enough and you get limited — your max stake quietly drops to a few dollars — or your account is restricted entirely. The very act of profiting reliably gets you shut down, so the ceiling on this "guaranteed" strategy is enforced by the venues themselves.

Execution risk: the leg you can't fill

An arb only works if you get both sides. In practice you place one leg, then reach for the second — and the price has moved, or the liquidity is gone, or the market moved against you before your second order fills. Now you're holding one naked side of a bet you never wanted, exposed to exactly the outcome risk arbing was supposed to remove. On thinner prediction-market books, the size you can actually fill at the quoted price may be far smaller than the number on screen.

Prices move before you're done

Arbs are short-lived. The gap that existed when you spotted it can close in seconds as other traders and the venues correct. By the time you've calculated stakes and clicked twice across two sites, the number you were arbing may no longer exist. Speed and pre-funded balances on both venues are the difference between a real arb and a half-filled mess.