
Rounding stakes in correct score dutching can change the return on each selected result. A calculator may show an ideal allocation with many decimal places, while the operator accepts stakes only in a particular increment. Once the amounts are rounded, the combined stake may differ from the intended budget and the covered outcomes may no longer return the same amount.
The practical check is to use the amounts actually accepted, add them, and calculate the return for every selected score and for results outside the selection. This guide uses invented prices and a fictional stake increment. It does not describe a current opportunity, guarantee acceptance or recommend placing a bet.
The introduction to correct score dutching explains why the ideal stakes differ. Here the narrower question is how a fixed ten-unit budget can be allocated in hundredths and how that restriction changes the ledger. The separate any-other-score guide explains why an unlisted result cannot automatically be assigned to a catch-all selection.
Start with the exact allocation
Suppose three mutually exclusive final-score selections are offered at fictional decimal odds of 7.00, 8.00 and 9.00. The intended total cash stake is ten units. Assume all three prices concern the same event and settlement period, with no commission, fees, promotions, rounding of returns, voids or price changes. These assumptions define the exercise, not an operator's terms.
Add the reciprocal prices: 1/7 + 1/8 + 1/9 = 191/504. Dividing the ten-unit budget by this sum gives the equal gross return, 5040/191, or approximately 26.3874346 units. Dividing that return by each respective price gives the ideal stakes:
| Selection | Decimal odds | Exact ideal stake | Approximate stake |
|---|---|---|---|
| A | 7.00 | 720/191 | 3.7696335 |
| B | 8.00 | 630/191 | 3.2984293 |
| C | 9.00 | 560/191 | 2.9319372 |
The exact fractions add to ten. Each fraction multiplied by its price returns 5040/191. These ideal quantities are mathematical values; they are not evidence that an operator accepts fractions of a unit or that the prices are available together.
Round the stakes and rebuild the table
Assume for this exercise that each stake must be an exact multiple of 0.01 units and that the three accepted amounts must sum to exactly ten units. Rounding each ideal stake independently to the nearest hundredth happens to produce 3.77, 3.30 and 2.93. Their sum is exactly 10.00 here. Other inputs need not have that convenient property.
| Winning selection | Accepted stake | Gross return | Net result on ten staked |
|---|---|---|---|
| A at 7.00 | 3.77 | 26.39 | +16.39 |
| B at 8.00 | 3.30 | 26.40 | +16.40 |
| C at 9.00 | 2.93 | 26.37 | +16.37 |
| None of A, B or C | 10.00 across the three | 0.00 | −10.00 |
The FootyMetrics dutching calculator illustrates that choosing a rounding step can make selection returns uneven. Its available steps are tool settings, not proof of any operator’s acceptance rules. The three winning returns differ by up to 0.03 units. The displayed net figures subtract the entire ten-unit portfolio, including the stakes on losing selections. A score outside the named set is not a fourth winner simply because the three covered returns are close.
Why independent rounding may miss the budget
A rounded stake is a change to a constraint, not merely a shorter way of displaying the same number. For a separate example, consider three ideal stakes of 3.335, 3.335 and 3.330 units. They sum to ten. Under an assumed half-up rounding rule to hundredths, the first two become 3.34 and the third 3.33, for a total of 10.01. A different tie-breaking convention can give a different result.
An allocation process has to state what to do with the extra hundredth. Reduce one stake, increase the budget if permitted, or leave a different amount unallocated: each choice creates a different set of returns. The decision cannot be recovered from the word “rounded” alone. Retain the rounding increment, tie rule and total budget with the calculated amounts.
The example does not imply any particular operator uses half-up rounding. An operator may impose a minimum stake, increment, maximum, price step or settlement convention that differs by product and account. Check the terms and accepted ticket rather than assuming the arithmetic display is executable.
Compare complete allocations, not just individual rounding errors
In the first three-price example, the full set of allowed allocations can be examined without claiming anything about a real market. Represent the ten-unit budget as 1,000 hundredths. Give A an integer number of hundredths from zero to 1,000; for each A amount, give B an integer amount from zero to the remaining budget; give C the remainder. This creates 501,501 ordered allocations, including allocations with a zero stake.
For each allocation, multiply its three stakes by 7, 8 and 9. Measure the largest gross return minus the smallest. Under the stated assumptions and this particular measure, the unique allocation with the smallest spread is 3.77, 3.30 and 2.93. Its gross returns are 26.39, 26.40 and 26.37, giving a spread of 0.03 units.
That is a result of the declared finite exercise, not a general rule that independently rounding is optimal. The possible allocations, budget, increment, prices and objective all matter. A minimum stake or a requirement to keep every selection above a specified amount would change the feasible set. A different objective—such as prioritising the smallest covered return—also asks a different question and needs a separate calculation.
A reproducible audit can use this pseudocode:
for A from 0 to 1000 hundredths:
for B from 0 to 1000 - A hundredths:
C = 1000 - A - B
returns = [7*A, 8*B, 9*C] # in hundredths
spread = max(returns) - min(returns)
retain all allocations with the smallest spreadBecause the units are integer hundredths and the example prices are whole numbers, this comparison uses exact integer arithmetic. The count of possible pairs is 1001 + 1000 + … + 1 = 501,501. The result would need a different representation if accepted prices had fractional increments that cannot be expressed as whole return-hundredths.
Spread is not the same as expected return
Minimising the gap between covered returns does not estimate the probability of any score. It also does not establish that the position has positive expected value. To estimate an expected net result, a reader would need defensible probabilities for the actual outcomes, including every uncovered score, plus costs and a complete settlement model.
For the first example, the three named returns are close, but an unselected result still loses ten units. Calling the position “balanced” should therefore mean only that the named winning returns are close under this allocation. It should not imply that the match is more predictable or that the whole position cannot lose.
The same caution applies to a calculator's combined odds. They describe the payout implied by the chosen prices and stakes under its assumptions. They do not supply an independently verified forecast or remove an operator's margin.
Reconcile the requested and accepted amounts
An ideal allocation, rounded request and final accepted ticket are three different records. If an operator rejects or changes one amount, recompute the total actually accepted before describing any outcome. If a transaction is partially matched or a quote moves, preserve the individual accepted portions and their prices. Do not replace missing execution evidence with the calculator's intended number.
For illustration, if only two of the proposed amounts, 3.77 and 3.30, were accepted at the hypothetical prices, the accepted total would be 7.07 units. A winning A returns 26.39, giving 19.32 net before costs; a winning B returns 26.40, giving 19.33 net. A result outside those two accepted selections loses 7.07. The third intended 2.93 units is not a stake if it was never accepted.
A void or refund is different again. Its treatment depends on the actual terms and settlement, and the simple full-win/full-loss table is then incomplete. Keep promotional credits, commission and exchange-specific matching rules separate from this ordinary cash illustration.
A small discrepancy needs the right explanation
A displayed return can differ from a statement because of stake rounding, price precision, per-ticket settlement rounding or costs. Change one assumption at a time when reconstructing the figures. If accepted amounts or the operator's rounding rules are unknown, report that limitation rather than selecting the convention that makes a result agree.
Save the source quote, complete market name and period, calculation timestamp, chosen stake increment, ideal stakes, rounded requests, actual accepted portions and prices. Add the final result and actual returns. This makes the calculation independently checkable without treating a screenshot as proof that the position was executed.
For the fictional 7.00, 8.00 and 9.00 prices, 3.77, 3.30 and 2.93 allocate exactly ten units and make the three covered returns as close as possible under the specified finite search. That finding remains conditional on the fictional inputs; it says nothing by itself about whether any real selection should be made.
コメント 0
まだコメントはありません。