Arbitrage Betting

How to identify and calculate an arbitrage

Add up the implied probabilities (1 ÷ odds) of every possible outcome across different operators; if the sum is below 100%, a theoretical arbitrage opportunity exists.

The Dutching Calculator in the Tools section automatically works out how much to stake on each outcome to achieve the same profit regardless of the result.

This guide focuses specifically on the maths of splitting a stake across three or more outcomes, and on how the same dutching technique that makes pure arbitrage possible can also be used more broadly, including for backing several selections in a single race based on a genuine view of value, not only for locking in a guaranteed, risk-free return. The general concept of exploiting a price discrepancy for guaranteed profit (what this site calls 'sure betting' elsewhere) is covered in the Sure Betting guide, with a simpler two-outcome example.

The stake-splitting (dutching) formula

To get the same profit no matter which outcome happens, each selection's stake needs to be proportional to its implied probability, not an equal amount on each one. The formula is: stake on a selection = (that selection's implied probability ÷ sum of all implied probabilities) × total budget. Following this formula guarantees the return (stake multiplied by odds) comes out roughly equal on any winning outcome within the group of selections covered.

This contrasts with splitting the budget into equal parts across selections, which would produce different returns depending on which odds actually won: a higher-odds selection needs a proportionally smaller stake to generate the same return as a lower-odds selection. The Dutching Calculator applies this formula automatically for any number of selections, just by entering the odds and the intended total budget.

A four-outcome example

Imagine a race with four relevant runners, priced at 3.00, 4.00, 5.00, and 15.00 across four different operators. The implied probabilities are 33.3%, 25%, 20%, and 6.7%, summing to 85%, well below 100%, indicating a generous theoretical 15% arbitrage margin. For a £1,000 budget, the proportional stakes would be roughly £392 on the first, £294 on the second, £235 on the third, and £78 on the fourth (each figure divided by the 85% sum and multiplied by the total budget).

Checking the maths: if the first runner wins, the return is £392 × 3.00 = £1,176; if the second wins, £294 × 4.00 = £1,176; if the third wins, £235 × 5.00 = £1,175; if the fourth wins, £78 × 15.00 = £1,170, all four figures land very close together (the small differences come from rounding), confirming a profit of roughly £175 on £1,000, regardless of which of the four runners actually wins the race.

Value dutching: backing several outcomes without pure arbitrage

The same stake-splitting formula has a second use, separate from arbitrage: covering several selections in the same race or event where value is believed to exist in more than one outcome, even without the sum of implied probabilities falling below 100%. This is common in horse racing, where a bettor might identify two or three horses with real winning chances higher than what the odds suggest, without any price discrepancy existing between operators at all.

In this case, the goal isn't guaranteed profit (it's still entirely possible to lose the full stake if none of the covered outcomes happens), but rather spreading the budget across multiple value bets in a mathematically efficient way, instead of betting everything on one selection or splitting the budget arbitrarily among the chosen ones. The Dutching Calculator works just as well for this purpose, simply by adjusting the probabilities used in the calculation to personal estimates, rather than only to the probabilities implied by market odds.

Arbitrage in markets with many possible outcomes

The more possible outcomes a market has (a twelve-runner horse race, say, instead of a three-outcome football match), the harder it becomes to find a price discrepancy large enough across every single selection at once to create a complete arbitrage. Each additional runner is one more implied probability added to the sum, and it only takes one selection priced unfavourably to push the total above 100%, wiping out the theoretical opportunity.

For this reason, genuine arbitrages in many-outcome markets are considerably rarer than in two- or three-outcome markets, and tend to show up more often in lower-profile events, where operators dedicate fewer resources to keeping prices aligned with each other across every selection simultaneously.

Common calculation mistakes in multi-way arbitrage

With just two outcomes, a small rounding error in the stake has minimal impact on the final result. With four, six, or more outcomes, those small rounding errors on each leg add up, and the real return in each scenario can drift further than expected from the theoretically equal value the formula calculates, usually still within an acceptable margin, but it's worth being aware the return won't be mathematically identical to the last cent on every outcome, especially with minimum stakes imposed by operators that prevent very fine adjustments.

A second common mistake is confirming the odds on only some selections before calculating the stake split, and assuming the rest are still at the level seen earlier: with more legs involved, the chance that at least one odd has moved between the initial check and submitting every bet rises proportionally, making reconfirming every odd immediately before betting even more important than in a simple two-outcome arbitrage.

How many legs are actually worth covering

Every additional leg in an arbitrage operation increases execution risk: one more account to manage, one more price that can move before the bet gets confirmed, one more opportunity for human error somewhere in the process. For this reason, many experienced bettors prefer sticking to two- or three-leg operations, even when a theoretical opportunity covering more outcomes appears to offer a slightly larger margin on paper.

The decision on how many legs are worth covering comes down to weighing the extra theoretical margin gained from covering one more outcome against the extra execution risk that introduces: there's no universally correct number, but the general pattern is that complexity grows faster than the marginal benefit from around four or five simultaneous legs onward, especially in less liquid markets where confirming each price takes longer.

Arbitrage in line markets: handicaps and totals

Beyond winner markets, line markets, like Asian handicaps or goals totals (over/under), can also generate arbitrage opportunities, through a slightly different mechanism. Instead of a direct price discrepancy on the same outcome, the opportunity arises when two operators set the line at different points: one operator might offer 'over 2.5 goals' at an attractive price, while another offers 'under 3.0 goals' also at an attractive price, creating an overlap zone (exactly 3 goals) where both bets can theoretically win at once, or at least one of them wins in nearly every scenario.

This kind of line arbitrage is more complex to calculate than direct winner arbitrage, because it requires considering not just the odds but also the probability distribution of possible outcomes around each specific line: essentially, how many goals a typical match in that competition tends to produce. Because it's more complex, it's also a type of opportunity less contested by other arbitrageurs, which occasionally means slightly larger margins for anyone who can calculate it correctly.

Recording and tracking arbitrage operations

As with matched betting, arbitrage's real value over time comes from adding up many small operations, not from a single exceptional opportunity. Keeping a record of each operation (the operators involved, the odds used, the stake on each leg, the theoretical margin calculated, and the profit actually confirmed once the event finished) makes it possible to check, with real numbers rather than a general impression, whether the theoretical margin is genuinely converting into net profit over time.

This record also protects against selective-memory bias, a common risk in any repetitive, numbers-based activity: it's easy to remember the operations that went exactly to plan and forget the ones that hit an execution problem or a calculation error, skewing the perceived profitability of the whole activity over time if there's no objective record to check against.

Choosing between pure arbitrage and value dutching

The two applications of the dutching formula covered in this guide (pure arbitrage, with mathematically guaranteed profit, and value dutching, backing several selections based on a personal probability assessment) call for different skills and suit different situations. Pure arbitrage depends entirely on finding real price discrepancies between operators, a task that's become more competitive as automated detection tools have spread, and one that demands execution speed above everything else.

Value dutching, on the other hand, depends on the quality of the probability assessment itself: without a solid estimate of how likely the covered outcomes really are, the technique offers no mathematical edge over betting in a less structured way, just a more efficient way of splitting capital across choices already made. Neither approach is absolutely superior; the right choice depends on where a given bettor's actual edge lies: in detection and execution speed, or in the quality of their own analysis of a specific event. Many experienced bettors end up using both approaches at different times, switching between them depending on what the available market actually offers on any given day.