Cash-out is presented as control. It is a trade: the bookmaker offers to buy your open bet at a price it sets, and the price is built exactly the way every other price on the site is built — from a probability, shaded by a margin. In July we compared 180 cash-out offers with the fair value of the positions. The average haircut was 6.4%.
How the offer is priced
Your £10 at 3.00 on a team now leading at half-time might be worth £22 at the live price. The bookmaker calculates that from its own in-play odds — which already carry a margin — then applies a second shading to the cash-out offer, and shows you £20.50. You give up £1.50 of expectation for certainty, and the certainty was never as valuable as it feels while the match is on.
The comparison to make
Stake × original price × current probability of winning = fair value of the position. Take the current probability from the live price at the sharpest operator you can see, not from the one making the offer. If the offer is within a couple of per cent of that figure, it is unusually fair; most are not.
The two good reasons
- Your view has changed. You backed over 2.5 goals because both midfields were weakened; the home side has gone down to ten men and dropped deep. The position is now a different bet from the one you made, and if you would not make it fresh at the live price, selling it at a modest haircut is reasonable.
- Your bankroll needs it. A large open position relative to your bankroll — an accumulator with one leg to go — is a risk you would not take as a single. Cashing out part of it is the equivalent of reducing a stake you should not have placed at that size.
The bad reasons
Nerves. A lead that looks fragile. A green number on the slip. None of these change the probability of the outcome; they change how it feels to wait for it. If you find yourself cashing out often, the size of the original stake is the problem, and the bankroll guide is the fix. Partial cash-out — selling a fraction of the position — carries the same haircut on the fraction sold and is not a compromise, only a smaller version of the same decision.
