Arbitrage betting explained

What an arbitrage is

Every bookmaker builds a margin into its prices, which is why backing all three outcomes of a match at one book always loses. Add up the implied probabilities and you get more than 100% — the excess is their cut.

But bookmakers price independently. One may be quick to move on a team-news story while another lags; one may shade a price to balance its own customers' money. Take the best available price for each outcome across several books and occasionally that total lands below 100%. When it does, the shortfall is yours, and it does not depend on the result.

That is an arbitrage, or a surebet. It is not a prediction, a system, or a read on the match. It is two companies disagreeing in public.

The maths

Bet365 has Over 2.5 goals at 2.10. Pinnacle has Under 2.5 at 2.00. Those two outcomes cover every possible result of the match.

Split a 100 stake in proportion to each outcome's implied probability — 48.78 on the over, 51.22 on the under — and both outcomes return 102.44. Nothing else about the match matters.

The arbitrage calculator does this for any number of outcomes, and reports the worst case once stakes are rounded — which is the figure that is actually guaranteed, since rounding breaks the equal-return property.

Why they exist at all

It is reasonable to be suspicious of free money. The specific inefficiencies that produce arbitrages are well understood:

• Different speeds. A sharp book repositions on news within seconds. A recreational book may take minutes, and in that window its old price can be beaten by the new one elsewhere.
• Different customers. A book heavy on one side may shade that price to attract money to the other, moving away from what it believes is true.
• Different margins. A book taking 2% and one taking 8% will diverge on the same view of a match, and on the outcome where the low-margin book is generous the two can cross.
• Different lines. Books do not always hang the same handicap or total, and the resulting price sets occasionally overlap in a way that leaves a gap.

All four are temporary by nature. Arbitrages typically last minutes and often less, which is why every leg on a scanner carries its age and why an old price is the most common reason a "surebet" turns out not to be placeable.

The seven checks

This is the part that matters, and it is where most arbitrage losses come from. Detecting a sum under 100% is arithmetic anyone can do. Not acting on a false one is the actual skill, because a bad arbitrage loses both stakes where a bad value bet loses one.

Each of these exists because it caught something real:

• 1. Matching settlement rules. Pinnacle gives a tennis moneyline action once a set is complete; several other books void the bet entirely if a player retires. Combine those and a retirement leaves you with one live bet and one refund — an open position, not a locked one. Legs may only combine within a rule class, and a book whose rules have not been researched must be excluded rather than assumed.
• 2. The same line on both sides. Over 2.5 against Under 2.75 is not a covered market. Quarter lines are worse than useless here: half the stake pushes, so a guaranteed return becomes a coin flip on the rest.
• 3. A complete outcome set. Home and away in a football match ignores the draw. The expensive version of this mistake is a basketball moneyline priced for regulation time on one book against one including overtime on another — the tell is that one of them quotes a draw.
• 4. Fresh prices. Every leg needs an age, and old legs must be discarded. A book that has stopped updating looks identical to a generous one, and ranking opportunities by margin surfaces exactly the stale ones first.
• 5. The right fixture. Books spell teams differently, and two different matches paired by a fuzzy name match produce a spectacular fake arbitrage. Anything matched with low confidence should be excluded outright. Video-game leagues carrying real club names — Esoccer, simulated-reality fixtures — are a particular trap.
• 6. At least two distinct books.Two prices from one bookmaker are not an arbitrage; they are one book's market, and it will not let you have both sides at a profit.
• 7. A believable margin. Above roughly 10%, a palpable error, a mismatched fixture or a settlement difference is far likelier than an opportunity. Bookmakers void palpable errors, so the very best-looking arbitrages are the ones least likely to pay.

Placing them in practice

The mechanics decide whether the theoretical margin survives:

1. Have both accounts funded and open. The window is minutes. Logging in and depositing while an arbitrage is live means it will be gone.
2. Place the soft leg first. The soft book is the one likelier to move, refuse the stake, or limit you. The sharp book is more likely to still be there afterwards — so take the fragile side while you know the other is available.
3. Check the bet slip price, not the listed price. Books frequently re-price at the moment of acceptance. If it has moved, recompute before confirming.
4. Round sensibly and know your worst case. Rounding breaks the equal-return property. On a 2.4% arbitrage, careless rounding can eat most of it.
5. Record everything. Both legs, both prices, the time, and the settlement. Unrecorded arbitrage is indistinguishable from gambling.

What limits it

Arbitrage does not scale indefinitely, and the reasons have nothing to do with finding opportunities.

• Account restrictions. The binding constraint. Soft books identify arbers quickly — unusual stake sizes, always the outlying price, obscure markets, no recreational betting — and cut maximum stakes to a level where the work is not worth it.
• Capital spread thin. Money has to sit at every book simultaneously, and a large share of it is tied up in unsettled positions at any moment.
• Execution risk. One missed leg can cost more than a month of completed arbitrages.
• Voided legs. A cancelled match, an abandoned game or a rule difference can void one side and leave the other standing.

Which is why many people who start with arbitrage move toward value betting: fewer guarantees, more variance, but a much higher ceiling and only one leg to place.

Frequently asked questions

Related

• Arbitrage calculatorstake split and worst-case profit
• Value betting guidethe other way to beat a price
• Live odds comparisonwhere the disagreements show up
• Arbitrage software compared