Settings

Language, shortcuts & timezone

Language

Timezone

Time zone changes apply on this device. A new device may start at GMT+8; check match times before reading.

Passer au contenu
Références

Paramètres

Personnalisez vos préférences

Correct Score Dutching: What the Coverage Misses

Par Opta Tips Team Mis à jour

Learn how a fixed stake can cover selected correct scores, what remains uncovered, and why equal returns do not imply a likely profit.

Illustration of two selected scorelines, their hypothetical returns, and a loss for other scores.

Correct score dutching means dividing a total stake across several different correct-score selections. The amounts can be chosen so that, if one of those selected scores wins, each winning outcome produces the same gross return. The important qualification is “one of those selected scores”: a score outside the chosen set can lose every stake.

This article uses invented odds and cash stakes to show the arithmetic. The figures are not actual quotes, recommendations or OptaTips results. Operator rules, market availability and costs differ. A correct-score selection normally refers to a defined match period, so check the actual market terms before comparing outcomes.

Start with a complete outcome question

Imagine a reader wants to understand two mutually exclusive selections in a fictional match: 1–0 at decimal odds of 6.00 and 2–0 at 9.00. Exactly one final score can occur under the same settlement period. Both selections therefore cannot win together, but both can lose together. A 0–0 draw, 2–1 win, 3–0 win or any other score outside this pair is uncovered.

The method allocates stakes; it does not establish whether either score is likely or whether the quoted prices are attractive. Nor is backing two scores equivalent to backing the home side to win: home wins such as 2–1 and 3–0 remain outside the pair.

A ten-unit worked example

Set a total budget of ten units. Allocate six units to 1–0 at 6.00 and four units to 2–0 at 9.00. If 1–0 wins, its gross return is 6 × 6.00 = 36 units. If 2–0 wins, its gross return is 4 × 9.00 = 36 units. The losing selection's stake has already been included in the ten units staked.

Final scoreGross returnTotal stakeNet result before costs
1–03610+26
2–03610+26
Any other score010−10

The equal return is conditional. It does not turn a partial set of scores into a guaranteed profit. The figures assume both bets are accepted in full at the stated prices, no commission or fee, and ordinary full-win/full-loss settlement. These assumptions are part of the example, not claims about an operator.

Why the stakes are unequal

To make each gross return equal, divide a target return by each selection's decimal odds. A shorter price requires a larger stake to reach the same gross return. In the example, 36 ÷ 6.00 = 6, and 36 ÷ 9.00 = 4. Their sum is the chosen ten-unit budget.

The calculation also works from a fixed budget. Add the reciprocals of the selected decimal prices: 1/6 + 1/9 = 5/18. Divide the ten-unit budget by 5/18 to obtain the common gross return of 36 units. Divide 36 by each price to obtain the two stakes. These calculations assume mutually exclusive outcomes and do not estimate their chances.

Statometrics' dutching explanation describes dividing a stake across selections to produce the same return if one of them wins. It is a commercial source; the table above independently shows both the covered results and the uncovered loss.

The same result label can conceal different coverage

The existing double-chance explanation compares two match-result selections. A match-result selection covers a family of correct scores: home win includes 1–0, 2–0, 2–1 and many others. A pair of exact-score selections covers only two members of that family. Even if the two tickets show a higher return when either named score occurs, they are not equivalent to the complete home-win event.

This difference matters when reviewing a prediction. A forecast that a team will win does not by itself distinguish 1–0 from 2–1, and cannot supply the probability of a small score set without additional information. Conversely, a predicted most likely exact score may still be unlikely in absolute terms because the remaining probability is spread across many alternatives. The price allocation tells you what would be paid under the stated terms; it does not tell you how likely that payment is.

What happens when the actual transaction differs

The equal-return table depends on both portions being accepted. If the six-unit portion is accepted at 6.00 but only two of the intended four units are accepted at 9.00, the actual matched total is eight units. Under the same no-cost assumptions, 1–0 produces a 36-unit return and 28-unit profit, while 2–0 produces an 18-unit return and 10-unit profit. Any other score loses the eight accepted units. The unaccepted two units do not belong in the stake total.

An attempted transaction may not be accepted in full. Use the actual transaction history and the applicable operator rules to identify the accepted amount. If the price changes or minimum-stake rules require rounding, recalculate each possible net result from the accepted stakes. Do not treat a calculator's requested amounts as proof of execution.

Keep a record that can be checked

For each selection, retain the fixture, score, period, operator, timestamp, accepted decimal odds and accepted cash stake. After settlement, record the winning return and any documented adjustment separately. The total profit is total actual returns minus total accepted stakes and any applicable separately recorded costs. If a leg is void, promotional credit is involved or settlement is disputed, the simple table above is no longer sufficient; consult the terms governing that specific transaction.

The allocation can be calculated without placing a bet. As an exercise, change one price and keep the budget fixed, then check the new stake amounts and all outcomes. A useful comparison exposes the cost and the uncovered scores rather than presenting the same covered return as an assurance about the match.

Coverage probability is a separate input

The return table tells you what happens for each result, but it does not say how often a covered score should occur. Suppose, solely for illustration, an independently specified model assigns 12% to 1–0 and 8% to 2–0 for the same fixture, period and information cutoff. Because these exact scores are mutually exclusive, the combined model probability is 20%. That does not follow from the 6.00 and 9.00 prices alone.

Under the earlier ten-unit example, a 20% combined probability gives an illustrative expected net result of 0.20 × 26 + 0.80 × (−10) = −2.8 units before costs. The break-even coverage probability at a 36-unit gross return is 10 ÷ 36, about 27.78%. That is a threshold implied by the assumed prices and stakes, not a measured probability or evidence that the prices are attractive. Change the prices, accepted amounts, costs or outcome definitions and the comparison must be recalculated.

A model that omits unlikely scorelines must also disclose the remaining probability. Treating the visible rows as if they sum to 100% can inflate apparent coverage. If a model supplies only a home-win percentage, it does not provide the separate 1–0 and 2–0 probabilities needed for this comparison. The responsible conclusion is that the coverage probability is unknown until suitable estimates are available, not that it equals the sum of price reciprocals.

Discussion des lecteurs

Commentaires 0

Aucun commentaire pour le moment.

Laisser un commentaire

Your email address will not be published. Required fields are marked *