Bookmakers are not trying to predict the result
That is the first thing to unlearn. A sportsbook trading team does not sit around forecasting winners; it prices events so the book makes a margin whatever happens, then manages the money that arrives against those prices. Here is sports betting trading explained in one line: odds are prices, not predictions, and a trading desk is the department that sets and defends them.
The reason to get into this now is a hiring story. Pythia Sports, a supplier of horse and greyhound racing pricing, trading and risk management services, has appointed Ben Hutton as Head of Trading after two decades leading international racing at Entain and its predecessor businesses. The job description reported alongside it is unusually revealing about what the role really covers, so it works as a way into the mechanics that sit behind every number you see on a betting slip.
What is a sportsbook trading team?
A sportsbook trading team is the group responsible for creating betting markets, setting the prices in them, and controlling the operator’s financial exposure once bets start landing. Everything else in a sportsbook, the app, the promotions, the payments, is downstream of the numbers this team produces.
In a small operation one person might do all of it. In a large one the work splits into distinct jobs that are easy to confuse from the outside.
Odds compiler, trader, risk manager: who does what
| Role | Core job | Typical decision |
|---|---|---|
| Odds compiler | Turns data and sport knowledge into a probability estimate, then into an opening price | What should this horse, or this team, actually be? |
| Trader | Runs live markets: moves prices, suspends and reopens, prices in-play | Do I shorten this price now, and by how much? |
| Risk or liability manager | Watches exposure across selections, events and customers | How much can we lose on this outcome, and is that acceptable? |
| Head of Trading | Owns margin, limits, process and the tools the desk uses | What are the rules everyone above operates inside? |
The compiler is closest to a modeller. The trader is closest to a market maker. The risk manager is closest to an accountant with a stopwatch. They overlap constantly, which is why the fourth role exists.
How are sports betting odds set?
Sportsbooks set odds by estimating each outcome’s probability from statistical models and sport-specific data, converting those probabilities into decimal prices, then shaving the prices so the implied probabilities add up to more than 100%. That excess is the margin, or overround. The result is the opening line, which is then adjusted as money and information arrive.
From raw data to a priced market
- Gather inputs: form, ratings, going and track data for racing, lineups and schedule for team sport, plus prices from other markets where they exist.
- Run the pricing model to produce a probability for every outcome in the market. These should sum to 100% before any margin is applied.
- Apply human overlay. A compiler adjusts for things a model handles badly: a jockey booking, a stable’s recent form, a late change in conditions.
- Add margin by reducing every price proportionally, or by loading more of it onto outcomes where the model is least confident.
- Set stake limits per selection, then publish. The opening price is a hypothesis about where the market will settle, not a final answer.
Why bookmakers build in a margin
Take a genuine coin-flip market. True odds are 2.00 on each side, and 1 divided by 2.00 gives an implied probability of 50% each, totalling 100%. Price both sides at 1.91 instead and each implies 52.36%, for a book total of about 104.7%. That 4.7% overround is roughly a 4.5% margin on turnover if the money splits evenly.
This is the sportsbook equivalent of the house edge, and it does not disappear because you picked well on a given afternoon. Racing markets with many runners typically carry a wider overround than a two-way football market, because there are more outcomes to load it across. Comparing the total implied probability of a market across two operators tells you more about value than any tipster ever will.
What a Head of Trading does day to day
A Head of Trading sets and polices the framework the desk works inside: target margins by market, stake and liability limits, which prices a trader can move alone and which need sign-off, staffing for big race days or fixture pile-ups, and the tooling the team uses to do all of it. Less pricing, more governance.
Risk and liability management
The daily rhythm is exposure. If one runner has taken heavy support and would cost the book a six-figure sum, the Head of Trading decides whether to shorten the price, cut the limit, lay some of it off, or simply carry the position. Carrying it is a real option, and this is where the second myth dies: sportsbooks rarely run a perfectly balanced book. Most manage to a target margin over thousands of markets and accept that individual events will lose money.
Why operators are hiring senior traders
Pythia’s remit for the role is specific. Hutton will lead the trading operation with responsibility for trading margins, how the racing product is presented to customers, and the tools the trading team uses. He joined after two decades on the bookmaker side, starting with Gala Coral Group and staying through its transitions into Ladbrokes Coral, GVC and Entain, where he led the group’s UK-facing international racing product and dealt with racing suppliers and rights holders including ARC, RMG, SIS, PMU and 1/ST CONTENT. He has also recently built AI-based trading tools and models, which he brings to the new role.
Notice that margin, presentation and tooling sit in the same job. That is not an accident. A price is a commercial decision and a product decision at once.
Model or trader: the trade-off no desk has solved
Here is the argument worth having. One camp says pricing is now an engineering problem: feed enough data into a model, let it price thousands of markets a day, and let automated risk rules handle the exposure. The other says racing in particular is full of information that never reaches a dataset in time, and that a trader who knows a yard, a track and a market’s habits still prices it better.
Both readings are defensible. Automation genuinely wins on breadth and speed, and no human desk can hand-price a full day of global racing and in-play football at once. But I think the second camp has the better of it on the decisions that actually cost money. Models are confident in exactly the spots where they should not be, because the unusual event is by definition thin on training data, and that is where a book takes its biggest hits.
The reported remit hints at where the industry has landed: someone who has built AI trading tools and spent twenty years running a racing product by hand. The honest position is that the models do the volume and the humans set the boundaries, and nobody can yet tell you where that line should sit. Whether senior trading hires look like this in five years, or whether the role collapses into model supervision, is genuinely unsettled.
Why do odds move after they’re published?
Odds move for two reasons: new information about the event, and new money in the market. Team news, a non-runner, a change in ground conditions and a withdrawn starter all shift the underlying probability. Betting volume shifts the price regardless of whether the probability changed, because the desk is managing liability as well as accuracy.
Sharp money versus public money
Not all stakes carry equal weight. A desk treats a large bet from an account with a long record of beating the closing price as information, and will often move the price immediately while cutting the limit. The same amount spread across hundreds of casual accounts on a popular favourite is treated as liability to be managed, not as a signal. Prices can therefore drift in the opposite direction to where most of the tickets are, which looks perverse until you know whose money is moving it.
In-play is a different job entirely
In-play trading recalculates prices continuously as the state of the event changes, with suspensions around significant moments while the model and the trader catch up. The margin tends to be wider than pre-match, because the pricing risk is higher and the feed delay is a real cost. If you bet in-play, you are paying for that uncertainty in the price.
Reading a market without kidding yourself
What all this buys you is literacy, not an edge. A line that moves is telling you that information or money arrived, and knowing which is which changes how you interpret it. Three habits follow from that, and none of them involve picking winners: add up the implied probabilities before you judge a price, check more than one operator on the same market, and treat a sudden move against the public side as information rather than a trap.
What it will not do is overturn the margin. The overround is the structural reason the operator profits over time, and understanding how it is built does not remove it from the price you take. Bet only money you can lose without it mattering, set deposit and loss limits before you need them, and use cool-off or self-exclusion tools if betting stops being entertainment. If it feels like income, step away.
Who is Pythia Sports?
Pythia Sports supplies pricing, trading and risk management services for horse and greyhound racing, meaning it provides the numbers and systems that bookmakers build their racing products on. Announcing the appointment, MD and CTO Adrien Tillier said the company was bringing Hutton’s bookmaker-side experience in to shape its services around what operators need from suppliers, whether in pricing, managing risk or improving the product.
That is a supplier hire rather than a sportsbook hire, which is the part worth keeping in mind. The prices you see on a racing market increasingly originate a step removed from the brand on the betting slip, and the trading decisions that matter are being made by people whose names you will never see on the app.