Reading the Overround

Every race you have ever bet on was priced against you before you looked at it. Not by conspiracy, and not by any judgement about your ability. By arithmetic. A bookmaker's prices, added together, always claim more than 100% of the race. The excess is called the overround, and it is the quiet toll collected on every book, every day, at every meeting from Ascot to Ayr.

Most punters never learn to read it. They will argue for twenty minutes about whether a horse should be 7/2 or 4/1, then hand over their money without asking how much the whole market is charging them for the privilege. That is like haggling over the price of a pint without noticing the pub has added a 15% service charge to the bill.

This piece is the full explainer. What the overround is, how to estimate it from a racecard in your head, what a normal margin looks like in UK and Irish racing, why the margin is not spread evenly across the field, and — because we don't dress things up here — what it means for your realistic long-run expectation. It isn't cheerful. It is the single most useful piece of arithmetic in betting.

What the overround actually is

Every price implies a probability. A horse at evens (2.00) is being offered as a 50% chance. A horse at 3/1 (4.00) is offered as a 25% chance. If a bookmaker priced a race fairly — matching the true chances exactly — the implied probabilities of every runner would sum to 100%, because one of them has to win.

They never sum to 100%. A real book might sum to 110%, 118%, 130%. That surplus above 100 is the overround: the theoretical margin built into the prices. A book at 110% is often described as "10% over"; a fair one would be "round", hence the name.

The margin exists for a reason. A bookmaker laying every horse takes on risk, pays staff, pays data fees and levy, and needs the arithmetic to favour them across thousands of races. Whether the margin on a given race is reasonable or extortionate is the question, and the only way to answer it is to add the book up.

To feel the mechanism: imagine backing every horse in a race so that you collect £100 whichever wins. With a fair 100% book, that costs exactly £100. With a 110% book it costs £110 — you have guaranteed yourself a loss of about 9% of your outlay before a stall opens. That loss doesn't fall on any one punter. It falls on all of them, collectively, race after race.

Converting a price to a probability in your head

The conversion is one line. For a fractional price of a/b, the implied probability is b ÷ (a + b).

So 7/2 is 2 ÷ 9 ≈ 22%. 9/4 is 4 ÷ 13 ≈ 31%. 100/30 is 30 ÷ 130 ≈ 23%. For decimal odds it is even simpler: 100 divided by the decimal price, so 5.00 implies 20%.

You do not need precision on a racecourse. You need anchors:

• Evens = 50%. 6/4 = 40%. 2/1 = 33%. 3/1 = 25%. 4/1 = 20%.
• 5/1 ≈ 17%. 7/1 ≈ 12%. 9/1 = 10%. 12/1 ≈ 8%. 19/1 = 5%. 33/1 ≈ 3%.
• Odds-on mirrors: 4/6 = 60%, 1/2 ≈ 67%, 1/4 = 80%.

With those in your head, you can run down a six- or seven-runner field in about thirty seconds, add the rough percentages, and know whether you are looking at a book of 108% or 125%. Nobody at the rail will know you are doing it. It is the cheapest skill in racing.

A worked example: adding up a six-runner book

Here is a realistic six-runner race — the sort of novice hurdle you'd find on a midweek card — priced by a single firm. Convert each price, sum the column, and the margin shows itself.

The book is 110.2% — about 10 points over, or roughly 1.7% per runner, which as we'll see is close to par for a modern industry starting price. To strip the margin out, divide each implied percentage by the total (1.102). The 5/4 favourite's honest share of the book is 40.3%, which corresponds to a "fair" price of 2.48 — nearly 6/4. The 20/1 shot's honest share is 4.3%, a fair price of about 22/1.

Two things to notice. First, the favourite at 5/4 loses about a tenth of a point of decimal price to the margin, while the outsider loses more than two full points — the margin, applied proportionally, costs the big prices far more. Second, this proportional method assumes the margin is spread evenly across the field. It usually isn't, which is where things get interesting. Hold that thought.

What a normal overround looks like

The most useful yardstick is overround per runner: the total margin divided by the field size. The Horseracing Bettors Forum, which has tracked this closely, reports that the UK industry starting price has run at roughly 1.7% to 1.9% per runner since the SP system was overhauled in mid-2020 — since when the SP has been compiled largely from a sample of off-course prices rather than the on-course ring alone. Before that change, the on-course SP overround hovered around 1.85% per horse for years, peaking at about 2.1% in the summer of 2019; Irish on-course books ran higher still, at around 2.2% per horse. (We cover how the SP is actually formed in SP, board price and BSP.)

Multiply the per-runner rate by the field and you get a feel for what a normal book should sum to. The table below is derived from HBF's published per-runner figures at roughly 1.8% per horse:

This is why "the overround was 125%" means nothing on its own. In a five-runner match it would be daylight robbery; in a 20-runner handicap it is about standard. Judge the margin per runner, not the headline number.

The number also moves with the market. Big, heavily-traded races — a Saturday feature handicap, a Cheltenham Festival contest — tend to be priced more keenly early and competed down further, while a weak midweek card can carry a fatter margin because nobody is fighting for the business. Ante-post markets, priced months out with no guarantee your horse even runs, carry substantially larger overrounds again. And early-morning prices at a single firm can be sharper or fatter than the eventual SP depending on how aggressively that firm wants your custom.

Where the margin sits: it is not spread evenly

Here is the part most explainers skip. The overround is not a flat tax. Bookmakers load a disproportionate share of the margin onto the outsiders, and the prices at the front of the market are usually much closer to fair.

The evidence for this is one of the oldest and most replicated findings in the economics of betting: the favourite–longshot bias, first documented by Richard Griffith in 1949 and confirmed across UK, US and Australian racing in decades of studies since (the survey by Ottaviani and Sørensen is the standard academic overview). The consistent result is that money backed at short prices loses at a far slower rate than money backed at long prices — losses on extreme outsiders run several times greater, per pound staked, than losses on favourites.

You can see why from the bookmaker's side. Clipping a 50/1 shot to 33/1 barely registers with the punter who fancied it, but it adds a chunk of implied probability to the book — 33/1 implies nearly 3%, 50/1 under 2%. Shaving the same margin off a 6/4 favourite would be obvious to everyone and would send the business elsewhere. So the outsiders absorb the margin, quietly.

The practical consequence: our proportional "fair price" column above is a decent first pass, but it flatters the outsiders. The true fair price of the 20/1 shot in a 110% book is often longer than the calculated 22/1, and the favourite's true fair price is often nearer its board price than the naive method suggests. When a big price looks generous, the honest starting assumption is that it isn't — a theme we return to in spotting fake favourites, and one that matters doubly for place terms, covered in each-way value.

Exchanges, commission and Best Odds Guaranteed

Betting exchanges change the arithmetic because there is no bookmaker building a margin into the prices. On a liquid Betfair win market close to the off, the backers' book routinely sums to within a point or so of 100% — the competition between thousands of opposing opinions squeezes the overround almost flat. The catch is that the exchange charges commission on net winnings instead: under Betfair's current My Betfair Rewards structure that is between 2% and 5% depending on the package chosen, per Betfair's own published charges. A 100.5% book minus a few per cent commission on winnings still usually works out a materially thinner total cost than a 114% fixed-odds book — especially at the bigger prices, where the fixed-odds margin bites hardest — but it is not zero, and thin markets on minor races can be far from tight.

Best Odds Guaranteed pulls in the other direction, in the punter's favour. Where a firm offers BOG — typically on win bets, on the day of the race, with conditions varying between firms — you receive the larger of the price you took and the SP. That converts the early price from a gamble on market movement into a one-way option: if the horse drifts, you get the bigger SP; if it shortens, you keep your price. Mathematically it hands a slice of the margin back. It is also precisely why heavy users of BOG tend to find their accounts restricted; the concession costs bookmakers real money, and they withdraw it from people who use it well.

None of this makes any individual bet good. It changes the size of the toll, not the direction of it.

What the margin means for your long run

Be honest with yourself about what all this arithmetic adds up to. If the books you bet into average, say, 115%, then the market as a whole is structured so that punters collectively lose around 13% of everything staked at those prices over time. You are not the average punter, you will say — nobody is, apparently — but the margin is the reason that most punters lose over the long run. Not bad luck. Not fixed races. Compounding percentages. To come out in front over hundreds of bets, your judgement doesn't need to be merely good; it needs to be better than the market by more than the margin, consistently, and the evidence of a century of betting turnover is that very few people manage it.

The overround is why the punter's default long-run outcome is a loss. Reading it doesn't remove the toll — nothing does — but it tells you exactly what you're being charged, race by race, and lets you refuse the worst of it.

So use it the way it deserves to be used: as an instrument, not a system. Add up the book before you look at any horse. Judge it per runner against the 1.7–1.9% benchmark. Expect the margin to be hiding in the big prices. Prefer the tighter market or the BOG concession when the choice exists. And size your betting on the assumption that the arithmetic is against you — which is a bankroll question, and we deal with it properly in bankroll and staking.

Racing is decided in lengths. Betting is decided in percentages, and the overround is the first one to learn. There is plenty more where this came from on the home page.