Image by Isobel Harcastle
All eight horses on a 50p accumulator placed at William Hill’s branch in Thirsk, North Yorkshire, won on 22 February 2008. The first, Isn’t That Lucky, took the 2.55pm at Sandown, and the last, A Dream Come True, the 9.20pm at Wolverhampton’s Dunstall Park. The slip belonged to Freddie Craggs of Bedale, and reports over the following days put its odds at anything from two million to one to 2,798,000 to one.
One result like that proves only that such a win can happen. Whether long accumulators pay over years is a different question. Bookmakers’ accounts, which count millions of bets, answer it, and so does the pricing of bets with several legs: as an accumulator gets longer, the share of each pound the bookmaker expects to keep tends to rise.
A 50p slip outgrew a £1 million limit
An accumulator stakes the whole return from each leg on the next, and the prices the Express & Star printed the next day show how steeply that snowballs. Isn’t That Lucky won at 8-1, turning 50p into £4.50. Three winners at Nad Al Sheba in Dubai, at 7-2, 11-2 and 4-1, took it to about £658, and Ede’s, a 10-1 winner at Warwick, to about £7,239.
The evening card at Wolverhampton did the rest. Rio L’Oren won at 15-2 and Guadaloup at 11-2, so roughly £400,000 rode on A Dream Come True, a 5-2 chance, in the final race. At full odds the slip was worth about £1.4 million on paper.
Each of those eight prices carried the bookmaker’s margin. Because the stake rolled forward, each new margin bit into winnings that earlier margins had already trimmed, much as a loan charges interest on unpaid interest. The compounding that built £1.4 million from 50p also deepens the bookmaker’s expected edge on every long accumulator.
That £1.4 million was never payable. William Hill’s limit for winners on horse racing was £1 million, so about £400,000 of the paper value fell outside it. In effect, the limit priced the final leg at 6-4 instead of 5-2.
A limit that size rarely touches a single bet: at 10-1, returning £1 million takes a stake of about £90,909, while at this slip’s odds about 36p would do. The limit bites on long, cheap accumulators, whose appeal is the size of the return.
Big wins can bring questions
Winning and getting paid were separate events. A report the following Monday said Craggs had yet to claim, quoting a William Hill spokesman who was “baffled as to why the winner has not contacted us”.
Online, the account takes the place of the slip, and winnings can still wait in it. Writing in July 2024, Andrew Rhodes, then the Gambling Commission’s chief executive, put the complaints its contact centre was still receiving about delayed withdrawals at around 2,000 a year, more than on any other subject.
For an accumulator that lands, the questions that can hold up payment include who placed it and how the account behind it was funded. Gambling.com argues that a best payout casino should not hold such questions back until a customer withdraws if it could fairly have raised them sooner.
The argument applies just as much to an online bookmaker. The Commission’s rule on verifying a customer’s identity and age covers betting sites as well as casinos and must be met before any stake is taken, so a request to prove identity or address that arrives only with a big win concerns a check that should already have been done. Questions about funds are different: evidence of where gambling money comes from may reasonably be needed later, and the site may hold a withdrawal while it reviews that evidence.
Those later checks have their place, because the Commission expects operators to monitor accounts “on an ongoing and risk-sensitive basis”. Rhodes’s examples show how they can be misused. One operator took a new customer’s £30 deposit through a payment method it had flagged internally as a risk, possibly linked to a third party, then froze the account when the customer tried to withdraw winnings of over £1,000, paying out only after it had investigated.
Another waited until a customer asked to withdraw winnings before requesting almost four years of information on the source of their income and deposits, although nothing in those deposits appears to have raised suspicion. It paid in full once it had obtained a Defence Against Money Laundering from the National Crime Agency.
Where an operator has no regulatory concerns about a customer, Rhodes wrote, “there is no valid reason to delay the payout of the withdrawal”, and unless, for example, a suspicion of money laundering rules it out, the customer should be told why a withdrawal has been delayed. A winner given that reason can work out which question is holding the money, and whether it is one the site should have settled at the start.
Multiples in the annual reports
Listed bookmakers publish the share of stakes they keep. Flutter, whose brands in Britain include Sky Bet and Paddy Power, defines its net revenue margin as sportsbook revenue as a percentage of the amount staked. In the UK and Ireland, that margin climbed 200 basis points to 13.8% in 2024, as sportsbook revenue rose 16% to $1,691 million.
The company credited “greater adoption of higher margin same game parlay bet types”, parlay being the American word for an accumulator, and said it also benefited from 150 basis points of favourable sports results.
Ladbrokes’ parent, Entain, reported a sports margin of 13.2% on £2,460.9 million of online sports bets in the UK and Ireland in 2025. Evoke, which owns William Hill, reported online sports staking there down 12% and a flat net win margin, “albeit with a structural increase in the margin due to customer and product mix changes”.
Bet builders price dependent legs together
In Britain the same-game parlay goes by the name bet builder: several legs inside one match. Those legs lean on each other. A goal from a team’s centre-forward makes its win more likely, so the bookmaker cannot build the price from two singles and sets one for the combination.
Manchester City’s 2-2 draw with Nottingham Forest at the Etihad in March 2026 shows how. As this site’s results and highlights page for the fixture recounts, Antoine Semenyo volleyed City ahead, Morgan Gibbs-White equalised with a backheel, Rodri restored the lead and Elliot Anderson levelled again.
A builder pairing a City win with a Semenyo goal looked healthy at half-time, because the goal that settled one leg had also put City in front. Gibbs-White’s backheel threatened it, Rodri revived it and Anderson’s equaliser sank it. The price had to allow for that dependence.
Propus Partners, a sports-betting consultancy, tested such prices for SBC News, comparing British bookmakers’ odds on ten bet builders for each of the five Premier League matches of 12 to 14 January 2024 with marginless probabilities from its own model. Average margins ranged from 20% to 35% by bookmaker.
Builders with three selections averaged about 25%, and those with four, five or six about 40%. Propus found that link weaker than in “standard accumulators/parlays across different events (where margin simply compounds)”. The mix of markets among goals, corners, cards and player goals mattered more, from about 20% with one type to 40% with all four.
Because the possible combinations are too many to list, Propus wrote, “the margin that is applied is somewhat hidden”, and customers are “left to accept or reject what they are offered”. Its final chart showed margins falling as a bet’s chance of winning rose: on shorter-odds builders, “the customer is likely to receive more of their stake back over time”.
The spread pays for acca insurance and price boosts
One British bookmaker’s published acca insurance applies only to accumulators of five legs or more. If just one selection loses, the stake is refunded as a free bet, and a winning free bet returns the winnings but not its stake.
A price boost works on the odds instead, lengthening them on chosen selections. Flutter says it generates revenue by setting odds that include “a theoretical spread to be earned on each contest less winnings paid and expenses associated with promotional activity”, and Entain reports revenue net of “free bets, promotions, bonuses and other fair value adjustments”. On those definitions, the cost of either offer comes out of the spread.
Whatever the offer, the law in Great Britain bars anyone under 18 from these bets. Gamstop, which is free, excludes people who register with it from online gambling with any company the Commission licenses. The exclusion cannot be cancelled before the term a person picked has run, and it stays in place after that until they ask for it to be removed. Someone in England or Scotland can phone the National Gambling Helpline free via 0808 8020 133, and someone in Wales the NHS helpline via 0808 2819 265.
Bookmakers ride out bad quarters
Results go against bookmakers too, and they report it. In the final quarter of 2025, Flutter’s revenue in the UK and Ireland fell 9%, and its sportsbook revenue there reflected “a 500bps adverse swing in sporting results”: customers staked about as much as a year earlier and won more of it back.
A bookmaker’s targets look past results like that. When Evoke tested its businesses for impairment, it assumed it would achieve its target sports betting margins, which it bases on its experience of results “over the long term, given the tendency for sports results to vary in the short term but revert to a norm over a longer term”.
A bookmaker takes enough bets for that norm to show. One customer’s bets are far fewer, so a lucky run can hide the margin for a long time without removing it.
Accumulators also look better than they are. In a 2021 paper on the structural characteristics of fixed-odds sports betting products, Philip Newall, Alex Russell and Nerilee Hing wrote that their win probabilities are “often small, and overestimated by casual observers”.
Small stakes that win “eye-watering amounts”, they added, are “often reported in the news”. Craggs’s slip became one of those stories, and the accounts above record what the same structure earns bookmakers across millions of bets no newspaper covers.
Related stories
Edward Davies Bennett is Editor-in-Chief and Responsible Publisher at Insight Britain, overseeing editorial standards, publication decisions and the corrections process.