Arbitrage Betting: The Maths, the Catch and the Calculator
Two prices disagree about the same match. Back every outcome and one of them always pays, but only if every leg lands at the price you saw.
Arbitrage betting means backing every outcome of one event at different venues, at prices whose implied probabilities add up to under 100%. Add 1 divided by each decimal price; if the total sits below 1, splitting your stake in proportion to those shares returns the same amount whichever outcome wins, but only if every leg is actually placed at the price you saw. That is a bigger condition than it sounds: three different venues carry three different minimum stakes, three different ID and withdrawal clocks and, if one leg sits on an exchange, its own commission rate on top, so a gap on paper is rarely the gap you actually collect. The calculator further down does this sum for two or three outcomes at once; read the frictions on this page before you trust the number it gives you.
Two prices, one gap
Three venues price the same match differently. One calls the home win 3.05. Another calls the draw 3.15. A third calls the away win 3.30. Prices disagree constantly. What is unusual is when the disagreement across every outcome of one market is large enough that backing all of them costs less than the smallest possible payout.
Three venues. One match. A gap that, for a moment, adds up to less than itself.
The test is one division per price, added together. Turn 3.05 into a probability: 1 divided by 3.05 is 32.79%. Do the same for 3.15 and 3.30, then add the three shares. Call that total S. At 100% or more, the three prices already cover the market fairly and there is nothing to split. Below 100%, the gap between S and 100% is not itself the return, a distinction the calculator below is built to get right rather than gloss over.
This only ever works back against back, one outcome per venue, never a bet against your own bet at the same venue. Laying is different; if it is a lay leg you are pricing, the mechanics of backing and laying on an exchange is the page to read, since the calculator below has no lay leg at all.
The maths, run once and checked twice
Take a football-style match with three outcomes: home win, draw, away win. Suppose the prices you find, this second, on three different screens, are 3.05, 3.15 and 3.30.
| Outcome | Decimal odds | Implied probability | Stake of a £1,000 total | Payout if this outcome wins |
|---|---|---|---|---|
| Home win | 3.05 | 32.79% | £345.72 | £1,054.45 |
| Draw | 3.15 | 31.75% | £334.75 | £1,054.46 |
| Away win | 3.30 | 30.30% | £319.53 | £1,054.45 |
| Total | 94.84% | £1,000.00 |
Add the three implied probabilities and the total is 94.84%, not 100%. That 5.16 percentage-point gap is not the return; it is the margin against a single unit stake, and confusing the two is the most common way people misread this sum. The actual return comes from inverting the total: 1 divided by 0.94836 is 1.0545, so the return on the whole stake is about 5.45%.
Splitting £1,000 between the three outcomes in proportion to each share, rounded to the penny, gives £345.72, £334.75 and £319.53, exactly £1,000.00 once rounded. Whichever outcome wins, the payout lands at £1,054.45 or £1,054.46, a one-penny spread from rounding, not a mistake in the arithmetic. Profit, after rounding, is £54.45 or £54.46 on the £1,000 risked, against a theoretical £54.45. Nobody collects any of it until the match finishes and every operator settles its own side, and that gap between the sum and the settlement is most of what the rest of this page is about.
The tool above works two or three outcomes at once, back bets only, and does this same sum; work through your own three prices on the arbitrage calculator before you trust a gap spotted on a phone screen. It also carries the exchange-commission field the next section explains.
Four brokers, four different answers on arbitrage
Ask each of the four brokers this site's own comparison lines up side by side what they make of arbitrage, and the answers do not agree with each other, and at least one broker does not agree with itself.
BetInAsia's own homepage states plainly that "arbitrage and value bettors are not limited for their success". The same operator's terms reserve a right to refuse, restrict, cancel or limit any wager for whatever reason, and a separate right to adjust an individual account's limits without notice. Both are true of the same broker; this page states both rather than picking one. BetInAsia also turns away applicants from the UK, the USA, France, the Netherlands, Sweden and a longer list besides, so the marketing line never reaches them anyway. Check the current terms before relying on either sentence.
AsianConnect runs the same contradiction the other way. Its blog says it will "never restrict accounts for using smart strategies like arbitrage", while its terms confine use to a personal, non-professional capacity and prohibit any commercial purpose. Both sit on the operator's own site; this page prints both rather than choosing. One detail cuts against the marketing: funds must be turned over once, at odds above 1.5, before a fee-free withdrawal, and the terms exclude bets on both outcomes from that turnover, so an arbitrage round trip does not earn the free withdrawal it might look like it should.
MadMarket's terms do not mention arbitrage by name, in either direction. What they reserve is wide discretion over accounts showing irregular playing patterns, or strategies aimed at what the terms call unfair winnings, wording broad enough to catch a method without naming it. MadMarket excludes the UK, the USA and Australia among others, and its Sharp Exchange charges a flat 3% exchange commission on winning bets, none on losing ones.
Sportmarket, oddly, builds an Arb Calculator into its own betting platform, a tool for hedging a bet placed elsewhere, yet publishes no stated policy on arbitrage itself; the closest thing on record is a third-party review claiming arbitrage players are accepted, somebody else's opinion, not Sportmarket's own word. I would treat a calculator built into someone else's betslip as a convenience, not a green light for a method the operator has never written a policy on. Sportmarket also excludes the UK, the USA and France, and counts turnover as the smaller of a bet's stake or its potential profit, voids and pushes excluded, with a 2% to 5% fee if you fall short.
Sbobet takes no such middle path. Its help centre states outright that arbitrage strategies violate its own terms, no exceptions given, flatter than any of the four brokers above say about themselves; this site's fuller look at what "sharp" actually means among these operators covers Sbobet's rules in more depth.
Betfair's own terms, for the record, say nothing about arbitrage specifically. Nobody here is inferring a policy that was never written down.
Five ways the gap disappears before you collect it
"Sure bets" is the marketing name for exactly this gap, and it oversells it: a leg can be refused, a price can move before the last bet lands, and an operator's own rules can cancel a bet that already looked settled. Good in theory. Not always in practice.
Speed. Every leg has to be placed while the prices you compared are still live. Sportmarket's own platform holds an in-play bet in a "Danger" status for a few minutes around goals, cards, penalties and VAR checks, exactly when a price is likeliest to move.
Refusal and void. A venue that takes one side of your bet owes you nothing on the rest of it. AsianConnect's own rules say a placed bet cannot be cancelled: if another leg voids or is declined after you have already committed, you are left holding one side, not a settled arbitrage.
Minimum stakes that do not divide evenly. A proportional split rarely lands on round numbers, and every venue has its own floor: EUR 10 on BetInAsia's or MadMarket's Sharp Exchange, EUR 5 on Sportmarket's FairExchange, EUR 4 on AsianConnect's PIWIX. A small total stake can fail a minimum on the smallest leg before anything else goes wrong.
Exchange commission on one leg. Put one leg on an exchange charging its own commission, for example the flat 3% MadMarket's Sharp Exchange, Sportmarket's FairExchange and AsianConnect's Orbit and PIWIX each charge on winning bets (BetInAsia's current exchange commission is not stated on its live pages, so check the current terms). At 3%, the 3.30 leg is worth 3.231 in effect. Run the same sum with that change: the total rises from 94.84% to about 95.48% and the return falls from 5.45% to about 4.73%. Exchange commission always narrows the gap, and on a smaller gap to begin with it can close one completely.
Turnover rules that eat what is left. BetInAsia asks for turnover three times at odds above 1.5 before a fee-free withdrawal, or up to 5% otherwise. MadMarket asks for turnover once before any withdrawal and three times for a fee-free one. Sportmarket's own "Valid Turnover" counts the smaller of a bet's stake or its potential profit, voids and pushes excluded, with a 2% to 5% fee if short. None of the four names arbitrage in that rule; all four apply it regardless of how the money in your account was made.
| Broker | Its exchange tool, minimum stake | Turnover rule to avoid a fee | Countries it turns away (not exhaustive) |
|---|---|---|---|
| BetInAsia | Sharp Exchange, EUR 10 | 3x turnover at odds above 1.5, or up to 5% | UK, USA, France, Netherlands, Sweden and others |
| MadMarket | Sharp Exchange, EUR 10 | 1x before any withdrawal, 3x for a fee-free one | UK, USA, Australia and others |
| AsianConnect | PIWIX, EUR 4 | 1x at odds above 1.5, bets on both outcomes excluded, or up to 5% | Its own list at registration; check the current one |
| Sportmarket | FairExchange, EUR 5 | 5x Valid Turnover, voids and pushes excluded, or 2% to 5% | UK, USA, France and others |
Every figure above is a snapshot of a term that can change without notice; check the current terms of whichever broker you actually use before relying on a number here.
Doing this in the right order
- Confirm it is really one market.
Same event, same market, same outcome definitions, or the sum means nothing.
- Run the numbers before you touch a betslip.
Enter the three prices into the calculator and read the stake split and the return, including any exchange commission a leg needs.
- Check every venue's minimum stake first.
A split that works in the calculator can still fail at the counter if one leg falls under that venue's floor.
- Place the most restrictive leg first.
The smallest limit, the thinnest liquidity or the slowest confirmation goes on first, since it is likeliest to move or refuse you.
- Confirm every leg matched, not just placed.
An unmatched exchange order or a pending bookmaker slip is not a locked position yet.
- Record it, win or not.
The only way to know whether this pays over a season is a ledger, not a memory of the good ones.
Is arbitrage betting legal?
That depends on where you live and on the rules of whichever operator holds your money, not on anything this page can settle. Betfair's terms say nothing about arbitrage specifically. Sbobet's do, and call it a violation outright. AsianConnect's marketing welcomes the same practice its own terms confine to personal, non-commercial use. None of that is a legal opinion: check local law before you bet anywhere, and check the current terms before assuming yesterday's reading still holds.
If this is not your method
I would not chase this gap across brokers I had never used before. Two, at most, and only after checking what each one's own exchange tool charges and requires; the arithmetic is the easy half of arbitrage, the account-opening and the ID checks are the slow half.
This sits inside the wider set of methods professional bettors use, past the one-line summary that page gives arbitrage. If a spreadsheet-and-stopwatch method does not appeal, the other methods covered on this site include value betting and exchange trading, and the fuller, less flattering picture of what doing any of this for a living actually requires is worth reading first.
A calculator finds the gap. The terms decide whether you keep it.
Every broker above sells more than its exchange tool, and the fuller side-by-side of what each one charges and excludes is worth reading before you fund any of them for this specific purpose.