It is tempting to treat exchange commission as a rounding step: calculate the hedge as if commission did not exist, then take a percentage off the winnings at the end. That approach is wrong, not just imprecise, because commission only applies to one side of a covered bet. Add it afterwards and you have not adjusted for a fee; you have quietly unbalanced the two outcomes you were trying to match.
Where it actually belongs
Back a bet at odds bo, lay it at odds lo, with commission c charged on the lay side's net winnings only. The lay stake that makes both outcomes equal is:
lay stake = stake × bo ÷ (lo − c)
Commission sits inside the denominator, subtracted directly from the lay price, not applied to the answer afterwards. That placement follows directly from setting "profit if the back bet wins" equal to "profit if the lay bet wins" and solving for the lay stake. Commission changes what lo is worth to you before the stake is calculated because it changes how much of a lay win you actually keep.
Why "subtract it at the end" breaks the hedge
Calculate the lay stake ignoring commission, then subtract commission from the final result, and you have sized a stake for a world where the exchange takes nothing. You then take money away from only the outcome where the lay bet won. The two outcomes stop matching by exactly the amount of commission on that side. What was supposed to be a covered position becomes a bet with a better result on one side than the other, sized as if that difference did not exist.
Worked example
Back $100 at 3.00 decimal, lay $100-equivalent at the same 3.00 on the exchange:
- No commission - lay stake $100.00, guaranteed result $0.00
- 2% commission - lay stake $100.67, guaranteed result −$1.34
- 5% commission - lay stake $101.69, guaranteed result −$3.39
Same prices on both legs, same stake, nothing else changed. The commission rate alone moves the guaranteed outcome from break-even to a $1.34 cost, or a $3.39 cost. That gap is exactly what decides whether a qualifying bet is worth placing for the free bet or bonus it unlocks: a $1.34 cost to unlock a $40 free bet is an easy yes; the same qualifying bet at a less competitive exchange, costing $3.39 or more, needs the free bet to actually be worth the difference.
It scales with the price, not just the stake
The same comparison at shorter odds shows commission having a different effect, not a smaller one. Back $100 at 2.00 decimal instead of 3.00, lay at the same 2.00:
- 2% commission - lay stake $101.01, guaranteed result −$1.01
- 5% commission - lay stake $102.56, guaranteed result −$2.56
The absolute cost is smaller than in the 3.00 example because there is less profit at 2.00 for commission to take a cut of. But the underlying pattern is the same: the guaranteed cost roughly triples when commission goes from 2% to 5%, whatever the price. Commission does not apply to the stake. It applies to the winnings, so it scales with how much the bet actually pays, not with how much was risked to place it.
Not every exchange charges it the same way
The formula here assumes commission on that single bet's net winnings, which is how most exchanges charge it. Some instead charge commission on your net profit across an entire market at settlement, meaning every bet you placed in that market is combined rather than charged separately. For a single back/lay pair, the two approaches converge, but for dutching or arbing across several selections in the same market, market-level commission changes the true cost in ways the per-bet formula only approximates. Check which model your exchange actually uses before assuming the numbers here transfer exactly to a multi-bet position.
The number that actually matters
Two percentage points, on their own, sound small. Run through the formula properly and they are the difference between a qualifying bet that costs a dollar and change and one that costs two and a half times as much, on identical odds, identical stake, and identical outcome either way. Commission is not a fee you pay at checkout. It is a term in the equation that decides what stake you need in the first place, and skipping it does not make the calculation simpler. It makes it wrong.
Both the back/lay and dutching calculators on this site take commission as an input for exactly this reason. Check your exchange's actual rate before you rely on either number.