Arbitrage Betting

Arbitrage — an "arb", or "surebet" — is the one betting strategy where the profit is genuinely certain at the moment the bets are struck. That's why it attracts attention, and why the honest version of this article spends four fifths of its length on the reasons it rarely survives contact with a real account.

How an arb works#

Every decimal price implies a probability: 1 ÷ odds. Add up the implied probabilities of every outcome in a market and you get the book. At a single bookmaker that total is always above 100% — the excess is the margin, which is how the book makes money.

But you don't have to take all your prices from one bookmaker. Take the best available price for each outcome from wherever it lives, and occasionally the total lands under 100%. That's an arb.

Say Book A has the home side at 2.10 and Book B has the away side at 2.00 in a two-way market:

Split a 1,000-unit total in proportion to each implied probability: 487.80 on the home side and 512.20 on the away. Home wins and you collect 487.80 × 2.10 = 1,024.39. Away wins and you collect 512.20 × 2.00 = 1,024.39. Identical either way, 24.39 units up on the 1,000 risked.

The three-way version works the same, and is more common in football because the draw is often where books diverge. Try the defaults in the arbitrage calculator below — 2.60 / 3.50 / 3.10 sums to 99.29%, which locks about 0.7%, or 7 units on a 1,000-unit outlay. That is a realistic size for a real football arb.

Note what the calculator does when there is no arb: it shows you the guaranteed loss from splitting a stake across every outcome. That's the honest answer for almost every market you'll check.

Why arbs exist at all#

Two bookmakers have to disagree by more than their combined margins. That's a high bar — with 4-5% margins on each side, the underlying opinions must differ by more than that before any gap opens.

It happens for three reasons. Books have different customer bases and shade prices toward what their customers back. They react to news at different speeds, so for a few minutes after a lineup drops one book is stale. And some books quote markets they don't model carefully — lower divisions, women's football, obscure cups — copying a feed rather than pricing it.

Every one of those causes is short-lived by nature. An arb is a temporary disagreement, and the temporary part is doing a lot of work.

The four things that kill it in practice#

Line movement between legs. You cannot place two bets simultaneously at two websites. In the seconds between them the second price can move, and the arb is thin enough that it doesn't need to move far. Take the example above: if the away side shortens from 2.00 to 1.90 before you get the second bet on, the total becomes 47.62% + 52.63% = 100.25% — you have just locked in a guaranteed loss of 0.25%. A +2.44% certainty became a −0.25% certainty because you were slow.

Voided legs. If one leg is voided — a market suspended and settled dead, a player who doesn't start, a bet cancelled under a bookmaker's palpable-error rule — you are no longer hedged. You hold a full-size one-way position at a price you specifically chose because it was out of line. Palpable-error clauses are especially awkward here, because the mispriced price is the exact one that created the arb.

Differing settlement rules. Two books can settle the same event differently and both be within their terms. Postponements and abandonments have different cut-offs. Some markets include stoppage time only, others extra time. Own goals, dead heats and "player must start" conditions vary between operators. When the rules diverge you can lose both legs of a bet you thought was risk-free.

Account limiting. This is the one that actually ends it. Arbitrage leaves an obvious signature: odd stake amounts, bets placed only into prices that are out of line, activity within seconds of a market move, and a portfolio that never touches a bookmaker's promotions. Books monitor for exactly this. The consequence is not a ban but a limit — your maximum stake quietly drops from four figures to double digits, and the strategy still works while being worth nothing.

Arbitrage versus value betting#

They're often mentioned together and they are opposites in almost every respect.

Value betting asks you to hold an opinion about probability and be right about it, then survive the variance while the expected value accumulates. Arbitrage asks nothing about probability at all — you're exploiting a disagreement between two prices, not disagreeing with either of them. That's why it needs no sample size to verify: there's nothing statistical to prove, only arithmetic to check.

The ceiling on scale#

The reason arbitrage is "mostly not worth it" isn't the percentage — 2% risk-free would be extraordinary if you could repeat it indefinitely. It's that you can't.

Your capital has to sit spread across a dozen bookmakers to be ready when a gap opens, so most of it is idle most of the time. Each arb absorbs only as much as the smaller book will accept, which is the small one. Finding them requires either paid software or constant screen time. And every arb you take moves you closer to the limit that ends the account. The strategy has a ceiling set by the bookmakers, and they lower it as you approach.

It is a legitimate way to extract a small, bounded amount of money from a market inefficiency. It is not a career, and it is not what Tofiko does — our models estimate probabilities and compare them to prices, which is the harder problem with the higher ceiling and no guarantee attached. If you want to know whether that harder approach is working, closing line value is the measure, and we publish ours.