Odds, margin, liability and why the most visible part of the industry is also the most volatile.
In this lesson:
- Read odds in decimal, fractional and American formats and convert them into implied probability
- Calculate the theoretical margin built into a betting market and explain how overround works
- Distinguish between turnover, gross gaming revenue and actual hold, and explain why they diverge
- Describe how a trading function manages liability, and why in-play betting changed the economics of sports betting
The visible business
Sports betting is the face of the gambling industry. It carries the shirt sponsorships, the television advertising, the stadium branding and the public debate. In some mature online markets, Great Britain among them, it also generates less revenue than casino, and with far greater volatility: in Great Britain in 2025/26, online casino games produced £5.7 billion of gross gambling yield against £2.4 billion for remote betting.
That combination is not a contradiction. Sportsbook acquires customers, and casino monetises them. Understanding why requires understanding how a sportsbook actually makes money.
Odds as prices
The starting point is that odds are prices rather than forecasts. A sportsbook is not primarily trying to predict the result of a football match. It is trying to set prices at which it can accept bets on all outcomes and retain a margin regardless of which one occurs.
Odds appear in three formats, and anyone working in this vertical needs to move between them fluently.
Decimal odds express the total return per unit staked, including the stake. Odds of 3.50 return £3.50 for every £1 staked, comprising £2.50 profit and the £1 back. Decimal is standard in Europe and is the easiest format for calculation.
Fractional odds express profit relative to stake. Odds of 5/2 return £5 profit for every £2 staked, so the total return is £7. Fractional remains conventional in the United Kingdom and Ireland, particularly in horse racing.
American odds use a positive or negative number relative to 100. A price of +250 returns $250 profit on a $100 stake. A price of -150 requires a $150 stake to win $100 profit. Negative numbers indicate favourites, positive numbers indicate underdogs.
The concept that unifies all three is implied probability. Any price converts into the probability the sportsbook is implicitly assigning to that outcome, and for decimal odds the conversion is simply 1 divided by the odds. Decimal odds of 4.00 imply a 25% probability. Odds of 1.25 imply 80%.
Overround and the margin
Here is where the commercial mechanism becomes visible.
Consider a tennis match with two possible outcomes. If the sportsbook genuinely assessed both players as evenly matched and priced accordingly, it would offer 2.00 on each, implying 50% and 50%, totalling exactly 100%. It would then expect to break even over time, which is not a business.
Instead it offers something like 1.90 on each. Each price implies a probability of roughly 52.6%, and the two together total approximately 105.3%. That excess above 100% is the overround, and in this case the theoretical margin is around 5%.
The mechanism generalises to markets with any number of outcomes. A football match has three, a horse race may have twenty, and a first goalscorer market may have forty. In each case the implied probabilities are summed, the excess above 100% is the overround, and the margin per outcome is the overround distributed across the market.
Margin varies enormously by market type, and the variation follows a clear logic. Highly liquid markets on major events carry thin margins, sometimes only a few percent, because competition is intense and sharp customers will move immediately to a better price. Obscure markets, exotic bet types and multi-leg accumulators carry much higher margins, because prices are harder to compare and because the margin on each leg of an accumulator compounds. A four-fold accumulator built from legs each carrying a 5% margin does not carry a 5% margin overall; it carries considerably more.
This is why accumulators, parlays and bet builders are commercially important out of proportion to their share of turnover. They are the highest-margin products a sportsbook offers, and they are also the products customers find most appealing, because a small stake can return a large sum.
Turnover, revenue and hold
Three numbers describe a sportsbook's performance, and confusing them is the most common error made by people new to the vertical.
Turnover, also called handle, is the total amount customers staked. It is a volume measure. It is the number most often quoted in press coverage, and it dramatically overstates the size of the business, because the overwhelming majority of turnover is returned to customers as winnings.
Gross gaming revenue is what the sportsbook retained: turnover minus winnings paid out. This is the actual revenue figure.
Hold is GGR expressed as a percentage of turnover. It varies by market mix, product mix and results, and in Great Britain it has risen over the past decade: remote betting held about 7.6% of turnover in 2015/16, and in 2025/26 operators reported turnover of about £25.0 billion and gross gambling yield of about £2.4 billion, a hold of roughly 9.8%. A sportsbook holding around 10% of turnover is converting £100 million staked into roughly £10 million of revenue.
The relationship between theoretical margin and actual hold is where sportsbook volatility lives. Theoretical margin is what the prices were designed to deliver. Actual hold is what results produced. They differ because sporting outcomes are not controlled by the operator.
When favourites win consistently across a weekend of major football, sportsbooks pay out heavily, because customers bet disproportionately on favourites and on popular teams. When results are unpredictable, sportsbooks hold above theoretical margin. Over a full year these deviations largely offset one another, but over a month or a quarter they can be severe enough to reshape reported earnings, and this is why sportsbook operators routinely explain results with reference to sporting outcomes. In the same Gambling Commission data, British remote betting held about 9.4% of turnover in July to September 2024 and about 12.6% in October to December 2024.
The trading function
Managing this is the work of the trading desk, and it is a genuine skill rather than an administrative one.
Trading begins with pricing. Modern sportsbooks price using models built on historical data, current form, team news and market signals, supplemented by human judgement on markets where models perform poorly. Many operators buy pricing feeds from specialist suppliers for the long tail of events and apply their own expertise selectively to markets where they believe they hold an edge.
Trading then moves to liability management. Once a market is live, money arrives unevenly. If a disproportionate volume backs one outcome, the sportsbook's exposure to that outcome grows, and it faces a decision: accept the exposure and rely on its assessment of the probability, adjust the price to attract money to other outcomes, reduce the maximum stake it will accept, or lay off part of the risk elsewhere.
Balancing a book perfectly, so that the same profit results whichever outcome occurs, is theoretically possible but rarely achieved and often not desirable. A sportsbook confident in its pricing will accept unbalanced positions, because refusing exposure means refusing the margin that comes with it.
Trading also involves customer management, which is the most contentious aspect of the vertical. A small number of customers bet in ways that consistently beat the prices offered, whether through superior information, faster reaction to news, arbitrage between operators or exploitation of promotional terms. Sportsbooks respond by restricting the stakes these customers can place or by closing their accounts.
This practice attracts sustained criticism, and the arguments on both sides are worth understanding. Operators argue that accepting unlimited stakes from customers with a demonstrated edge would make recreational pricing impossible, and that the alternative would be wider margins for everyone. Critics argue that a business advertising itself to the public should not be permitted to exclude customers simply for winning, and that the practice sits awkwardly with how the product is marketed. Several regulators have examined the issue. The Massachusetts Gaming Commission, for example, held public sessions with sportsbook operators on wager limitations in 2024, and from June 2026 requires an operator that limits an account to give the customer a specific reason within 48 hours. It remains one of the sector's genuinely unresolved conduct questions.
In-play and the transformation of the vertical
The single largest structural change in sports betting was the shift to in-play betting, and its significance is easy to underestimate.
Before in-play, a football match presented one commercial opportunity. Customers bet before kick-off, the match was played, and the bets settled. In-play converts that single opportunity into a continuous stream of them. Next goal, next corner, next card, the score at half time, whether a specific player will score in the next ten minutes, and dozens more, all repriced continuously as the match unfolds.
The commercial effect is substantial. In its September 2016 position paper, the Gambling Commission reported that in-play betting accounted for over one-third of online betting gross gambling yield in Great Britain. Operators value it because prices move quickly, comparison between operators is harder, and decisions are made under time pressure.
The operational requirements are correspondingly demanding. In-play pricing depends on knowing what is happening in the event faster than the customer does, which makes latency a competitive issue. Sportsbooks buy real-time data from specialist sports data companies, which they use to revise prices, suspend markets and settle bets. A delay between an event occurring and prices updating creates an exploitable window, so sportsbooks suspend markets around significant moments, build a delay of a few seconds into bet acceptance, and manage the risk of stale prices continuously.
In-play also has responsible gambling implications that regulators have taken seriously. Continuous betting opportunities on a live event compress the decision-making cycle in a way that resembles casino play more than traditional sports betting. The Gambling Commission's position paper noted that its survey data indicated online gamblers who bet in-play were more likely to be classified as problem gamblers than those who did not. Jurisdictions restrict in-play in different ways: Australia's Interactive Gambling Act 2001, for example, prohibits online in-play sports betting.
Bet builders, cash out and the modern product
Two further product developments define the contemporary sportsbook.
Bet builders, which allow a customer to combine multiple outcomes within a single event into one bet, have become central to the product. They are engaging, highly personalised and mathematically demanding to price, because the outcomes within a single match are correlated rather than independent. A bet combining a team to win and its striker to score is not two independent probabilities multiplied together, and pricing correlated combinations correctly is a specialist capability.
Cash out allows a customer to settle a bet before the event concludes, at a price reflecting its current standing. It is popular because it gives customers a sense of control, and it is commercially advantageous because the operator sets the cash out price from its own live prices, which carry its margin, and can reduce or withdraw the offer as the event moves. It also reduces the sportsbook's liability on outstanding positions, which trading teams value.
Horse racing and the specialist markets
Sports betting is not a single homogeneous product, and horse racing deserves separate treatment because it operates differently from team sports in several respects.
Racing has historically been the foundation of the betting industry in markets such as the United Kingdom, Ireland, Australia, Japan and Hong Kong, and in several jurisdictions it is subject to its own funding arrangements, where a levy or a share of betting revenue is directed back into the sport. In Great Britain, bookmakers pay the Horserace Betting Levy at 10% of their gross profits on British horseracing above £500,000, and in March 2026 the government confirmed it would not change the rate. This creates a structural relationship between betting operators and the racing industry that has no direct equivalent in football or basketball.
Racing also uses pool betting alongside fixed odds in many markets. In a pool, or pari-mutuel system, all stakes go into a common fund, the operator takes a fixed deduction, and the remainder is divided among winning tickets; British law defines pool betting as betting where winnings are determined by reference to the aggregate of stakes or divided among the winners. The operator therefore carries no liability on the result, because payouts are determined by the size of the pool rather than by prices set in advance. This is a fundamentally different risk model from fixed odds betting and is worth recognising when comparing operators.
Beyond racing, sportsbooks increasingly offer markets on esports, on politics and current events where permitted, and on virtual sports, which are computer-generated events with outcomes determined by a random number generator. Virtual sports are an interesting hybrid: presented as sports betting, but with outcomes generated by software in the way casino games are. Regulatory treatment varies by jurisdiction; in Great Britain they are licensed as betting, under a separate general betting (virtual events) operating licence.
Free bets, price boosts and acquisition economics
Sportsbook promotion works differently from casino bonusing, and the mechanisms are worth knowing.
Free bets grant a stake that, if it wins, returns the profit but not the stake itself. Price boosts enhance the odds on selected outcomes above the sportsbook's normal price, deliberately sacrificing margin on that market to generate attention and volume. Money-back offers refund a stake if a specified near-miss condition occurs. Acca insurance refunds an accumulator that fails by a single leg.
These are expensive, and in the most competitive markets the cost of acquiring a sports betting customer can be very high indeed. During the opening phase of state-by-state expansion in the United States, acquisition spending reached levels that proved unsustainable, and the resulting correction reshaped the competitive landscape considerably.
Promotional generosity also attracts customers who are there for the promotion rather than the product. Bonus arbitrage, where players extract guaranteed value by combining promotional offers with hedged positions elsewhere, is an established practice, and identifying it is a routine part of sportsbook risk management.
Regulatory pressure specific to sportsbook
Because sports betting is the industry's public face, it absorbs a disproportionate share of regulatory and political attention.
Advertising is the most visible pressure point. Several jurisdictions have restricted or banned gambling advertising around sporting events, prohibited shirt sponsorship, limited the use of sporting figures in marketing, or confined advertising to specific time windows. Spain, for example, prohibits gambling sponsorship on sports shirts and confines most gambling advertising on television and radio, including during live sport, to between 1am and 5am; in Britain, advertising rules since October 2022 bar gambling adverts of strong appeal to under-18s, a test under which, the advertising regulator said, ads can no longer feature top-flight footballers, and Premier League clubs agreed to withdraw gambling sponsorship from the front of matchday shirts from the end of the 2025/26 season; and in August 2026 Australia's Parliament passed reforms restricting gambling advertising, including in sports venues, most of which commence on 1 January 2027. The rationale concerns exposure of children and of vulnerable adults to gambling messaging in a context they are watching for other reasons.
Sporting integrity is the second. Betting markets create an incentive to influence events, and the industry accordingly participates in integrity monitoring arrangements that flag unusual betting patterns to sports governing bodies and law enforcement. Certain markets are restricted for this reason, particularly those on discrete events that a single participant could control, and betting by participants on their own sport is generally prohibited.
The third is product restriction. In-play limits, restrictions on betting on youth or amateur competition, and prohibitions on specific market types all reflect judgements about where betting creates unacceptable risk.
Where sportsbook sits in the business
Pulling this together, the vertical has a consistent profile. Revenue is smaller than casino's in some mature online markets, Great Britain among them, even though a sportsbook keeps a larger share of each pound staked: in Great Britain in 2025/26, remote betting retained about 9.8% of turnover and online casino about 4.5%, because casino players re-stake their winnings many times. Revenue is volatile month to month and highly seasonal across the year, concentrated around major tournaments and league calendars. Operating costs are higher, because trading teams, data feeds and pricing infrastructure are expensive. Marketing costs are higher still, because sports betting is the most competitive customer acquisition battleground in the industry.
What sportsbook delivers in return is customers. Sports betting has mainstream cultural presence in a way casino does not. It is socially acceptable, it is tied to events people already follow, and it reaches a broad audience through channels casino cannot use as effectively.
This is why the cross-sell relationship matters so much. An operator acquiring a customer through sports betting and subsequently introducing them to casino products can find that the casino revenue exceeds the sportsbook revenue from the same customer. The sportsbook is the acquisition engine; the casino is a large part of the monetisation. Recognising that relationship explains a great deal about how the industry allocates its marketing budgets, and why so many companies known for sports betting also offer casino.
Key takeaways
- Odds are prices, not predictions. They reflect where money is going as much as what the sportsbook believes will happen.
- Overround is the theoretical margin, but actual hold is what the business earns, and results determine the gap between them over any short period.
- Turnover is a volume measure and revenue is an outcome measure. Confusing them is the most common error in reading sports betting numbers.
- In-play betting transformed sportsbook economics by multiplying the number of betting opportunities per event, and it made data speed a competitive necessity.
- Sportsbook is structurally volatile, structurally seasonal and, in some mature online markets, smaller than casino by revenue, which is why cross-sell into casino matters so much to sportsbook-led operators.