Betfair Exchange vs UK Bookmakers: The Overround, the Commission and Where a Free Bet Really Pays

Editorial Team «Free Horse Racing Betting» June 18, 2026 Reading time: 25 min

Two punters back the same horse in the same race. One uses a fixed-odds bookmaker at a board price of 5.0. The other uses the Betfair Exchange and takes back odds of 5.4. Both win. The fixed-odds punter collects £50 on a £10 stake. The exchange punter collects £54, loses £2.20 to commission on the £44 profit, and walks with £51.80. A four-pound edge from the exchange on a single bet. Over a thousand bets at £10 stakes in a season, that four-pound edge compounds into several hundred pounds of additional return — and the reason it exists is the gap between 110-130% overround on fixed-odds racing books and 102-105% on exchange markets.

This is the piece that deals with the numbers directly. Not “the exchange is better” — that is too simple and ignores the real trade-offs of liquidity, commission, and the fact that exchanges do not offer free bets in the traditional sense. And not “the bookmaker is simpler” — that is equally lazy and ignores the structural overround cost that compounds across a season. The honest answer is that each model has a specific role, and a serious UK racing punter benefits from understanding the mathematics of both. The UK remote racing market handled £766.7m in gross gaming yield across fixed-odds books in the most recent reporting year; the exchange handle sits at a different scale entirely, but the per-bet value calculation works in the punter’s favour on exchange for any bet where liquidity is adequate.

What follows is the mechanics: how bookmakers construct an overround, how exchanges construct a market peer-to-peer, what the commission math actually costs you on a winning exchange bet, the difference between back and lay bets with worked liability examples, the conversion between fractional and decimal odds, how to calculate implied probability by hand, and where Tote pools sit as a third pricing model. At the end, a practical view of where free bets fit across all three channels.

Overround mechanics: how a fixed-odds bookmaker builds in margin

A fair-odds market, in the economic sense, would have probabilities that sum to exactly 100%. If one horse was truly a 50% shot, another 30%, another 20%, a fair book would price them at 2.0, 3.33, and 5.0 decimal respectively. Convert those prices to implied probabilities — 1 divided by decimal odds — and you get 50 + 30 + 20 = 100%. That is the fair-odds baseline that never actually appears in a real market.

Overround is what a bookmaker adds to that baseline. A typical UK racing fixed-odds book runs at 110-130% total implied probability. In other words, across all horses, the prices add up to 110-130 pence of probability for every pound of stake notionally returned. That 10-30 pence per pound is the bookmaker’s margin — the structural return they earn over the long run regardless of which horses win. On ante-post markets for Grand National, overround can run up to 180%, because the bookmaker is pricing a field of 60+ potential runners that will eventually reduce to 40 declarations, and absorbing both the outcome uncertainty and the non-runner risk into the prices.

The arithmetic on a small field: a three-horse race with true probabilities of 50%, 30%, and 20%. At 120% overround, the bookmaker shortens each price proportionally. Horse A’s 50% becomes 60% implied — price of 1.67 decimal instead of 2.0. Horse B’s 30% becomes 36% implied — 2.78 instead of 3.33. Horse C’s 20% becomes 24% implied — 4.17 instead of 5.0. Bet £10 at the fair prices on any horse with matching real probability and your expected value is zero. Bet at the overround prices and your expected value on each horse is -£1.67 for every £10 staked — exactly the 16.7% margin spread the bookmaker extracted.

The same logic applies to larger fields with proportionally different price distributions. A 16-runner handicap with overround of 125% spreads the margin across all 16 horses. The favourite at an implied probability of 25% becomes 31.25% — price of 3.2 instead of 4.0. The mid-field 8/1 shot becomes roughly 9.5/1 at fair — 10.5/1 after the overround adjustment. The long-shot 33/1 becomes roughly 40/1 at fair odds. Each horse carries a slice of the margin.

Overround is not uniform across markets even within the same race. Bookmakers typically apply a higher margin on outsiders — a 100/1 shot shortened to 80/1 is a bigger absolute probability adjustment than a 2.0 favourite shortened to 1.67. The margin skew is why “skinner” each-way bets on short-priced horses are poor value: the overround cost is concentrated in the part of the book where you are betting. Market structure and field size both drive the overall overround percentage, with bigger fields generally running tighter books (110-115%) and smaller fields running wider (120-130%).

The exchange model: peer-to-peer pricing at 102-105%

The exchange is a different structural beast. Betfair Exchange, the dominant UK exchange, acts as a matching platform rather than a bookmaker. Punters place back bets — “this horse will win” — and lay bets — “this horse will lose” — against each other. The exchange does not set the price; the price emerges from the matching of back and lay offers, with the exchange earning commission on net winnings rather than building margin into the odds.

Because the exchange earns through commission rather than overround, the prices on the exchange approximate fair odds far more closely than a fixed-odds book does. Implied probabilities on liquid exchange markets sum to 102-105% — the 2-5% “book” represents the tightest part of the spread between the best available back and lay prices. Compare that 102-105% to the 110-130% on a fixed-odds book and the structural price improvement on the exchange runs at 5-25 basis points per bet. That is the raw price edge, before commission is applied.

The liquidity question matters. Exchange markets work because there is a counterparty willing to lay your back at a given price. On high-handle markets — Group 1 races, championship races at festivals, Grand National, Gold Cup — liquidity is deep. You can get a £1,000 back matched at the market price with minimal slippage. On low-handle markets — midweek maidens, selling platers, obscure international meetings — liquidity is thin. You may get £50 matched at the best price and then wait several minutes for the rest to fill at progressively worse prices. On the worst-liquid markets, you may not get matched at all. Liquidity is the exchange’s real constraint, not price.

Market depth varies by horse within a race as well. The favourite and second favourite usually have deep books on both back and lay sides. Mid-price horses have thinner books. Long-shot outsiders may have virtually no lay action at all, which means you can sometimes get back prices on outsiders that are substantially longer than the fixed-odds equivalent. A 50/1 shot with no active lay interest may sit at 100/1 on the exchange’s “back” side of the book — someone would need to offer to lay at that price for the bet to match.

The exchange charges commission on net profits, not on stakes. If you back a horse at 5.0 for £10 and it wins, gross profit is £40. Commission at the standard rate (between 2% and 5% depending on points-based discounts) applies to that £40, not to the £50 return. This is structurally different from a bookmaker’s margin, which is priced into the odds and therefore effectively charged on every stake, win or lose. Commission only on net winnings means that a losing exchange bet pays zero commission — you simply lose the stake.

One structural difference worth flagging: exchanges do not traditionally offer free bets in the token sense. Some exchange operators run commission rebates, deposit bonuses, or referral credits, but “free bet” in the tokenised sense is primarily a fixed-odds product. A punter optimising for free bet value works on the fixed-odds side; a punter optimising for price efficiency works on the exchange side. The two are complements, not substitutes.

Commission math: what 5% actually costs you

Commission on an exchange bet is simpler than it looks but gets misunderstood often. The formula: net return = (stake × decimal odds) – stake – commission × (stake × (decimal odds – 1)). In plain language: gross profit is the stake times (odds minus 1); commission applies to that gross profit; your net return is gross profit minus commission, plus stake back.

A £10 back bet at 5.0 decimal that wins: gross profit = £10 × 4 = £40. Commission at 5% = £40 × 0.05 = £2. Net profit = £40 – £2 = £38. Total return including stake = £48. Compare that to a fixed-odds £10 bet at a slightly shorter price of 4.7 (which is roughly where the fixed-odds equivalent would sit under a 115% overround): return of £47, profit of £37. The exchange punter, even after commission, collects £1 more on the winning bet than the fixed-odds punter. Over a large enough sample of wins, that £1 compounds.

The commission rate varies by volume. Betfair operates a points-based discount scheme where heavy users pay commission as low as 2% instead of the standard 5%. A regular exchange user with a year of activity typically pays 3-4%. A very active professional user pays closer to 2%. The difference between 5% and 2% on a £40 profit is £1.20 of commission — 3% of the gross profit — which matters enormously at scale. A punter placing £100,000 of winning stakes in a year at 3.0 average price is earning commission discounts worth thousands of pounds purely from volume tiering.

Commission also interacts with “premium charges” on Betfair Exchange for certain heavy users who are consistently profitable. The premium charge is a separate levy on very large net winnings over the lifetime of an account, typically applied to users whose winning ratio significantly exceeds industry averages. For a standard recreational punter this is not a concern; for a semi-professional it is a material factor. The headline commission rate does not capture the full cost structure for heavy users.

One behavioural note on commission. Because it applies only to net profit, the psychological shape of an exchange account is different from a fixed-odds one. Losing bets cost zero commission, so the total cost of a losing bet is just the stake. This is mathematically identical to fixed-odds on losses, but the framing is different — on fixed-odds, every bet carries the overround whether it wins or loses, because the margin is priced into the odds. On exchange, you pay commission only when you win. That structure rewards selectivity: fewer, better bets cost less proportionally than more, worse bets.

Back and lay bets: structure, liability, and the opposite side of the exchange

A back bet on the exchange is conceptually identical to a fixed-odds bet: you stake £10 on Horse A at 5.0, the horse wins, you collect £40 profit (less commission). The difference from fixed-odds is how that bet is constructed — your £10 is matched against another user’s lay bet of £10 at 5.0. That counterparty is committing to pay you £40 if Horse A wins.

A lay bet is the inverse. You are offering to pay out if a horse wins; in return, you collect the other party’s stake if the horse loses. If you lay Horse A at 5.0 for £10, you are saying: “I will accept £10 from someone who backs Horse A, and I will pay out £40 to them if Horse A wins.” Your liability — the amount at risk — is £40, not £10. Lay bets require you to have the full liability covered in your account balance, not just the nominal stake.

Worked example of liability: you lay a horse at 4.0 for £20 stake. Liability = stake × (odds – 1) = £20 × 3 = £60. Your account must have £60 free to cover this bet. If the horse loses, you collect £20 (less commission on the £20 net profit). If the horse wins, you pay out £60. The profit ceiling on a lay bet is the stake; the loss ceiling is the liability. This is structurally the opposite of a back bet, where the loss ceiling is the stake and the profit ceiling is the liability-equivalent.

Lay betting has specific use cases. Laying a favourite in a race where you believe the favourite is overpriced — because of recent poor form, unsuitable ground, or a stronger-looking challenger — lets you profit from the horse losing without needing to pick the winner. On handicap races with 16+ runners, laying a 3.0 favourite often pays because the favourite’s implied 33% probability overstates genuine win chance in a competitive handicap; a horse priced at 3.0 realistically wins closer to 25-28% of the time in that context. Lay that bet consistently across the right race types and the expected value is positive.

The risk management around lay bets is more demanding than back bets. Multiple simultaneous lays compound liability. A £10 lay at 5.0 carries £40 liability; a £20 lay at 4.0 adds £60 more; together, if both horses win the same race (impossible in a single-horse win market but possible across races or across markets like forecasts), you are down £100 on £30 of stakes. Experienced lay bettors track aggregate liability across their active positions carefully, because the downside shape is steeper than it appears.

Fractional and decimal odds: converting between the two formats

UK racing traditionally quotes in fractional odds: 5/2, 11/4, 9/1. Most of the rest of the betting world — and most modern apps — prefer decimal: 3.5, 3.75, 10.0. They describe the same information but with different framing. Fractional odds quote net profit relative to stake; decimal odds quote total return relative to stake including the stake.

The conversion is straightforward. Decimal = (fractional numerator / fractional denominator) + 1. So 5/2 = 2.5 + 1 = 3.5 decimal. 11/4 = 2.75 + 1 = 3.75. 9/1 = 9 + 1 = 10.0. 1/3 = 0.333 + 1 = 1.333. The “+1” component is the stake return, which is always implicit in decimal but explicit in fractional. Going the other way, fractional = (decimal – 1) as a fraction. 3.5 decimal = 2.5/1 = 5/2. 4.5 decimal = 3.5/1 = 7/2.

Short fractional odds — odds-on — are quoted with the denominator first. 4/5 means you risk 5 to win 4; your return on a £5 stake is £9 (£5 stake + £4 profit). In decimal, that is (4/5) + 1 = 1.8. The 1/2 (1.5), 4/6 (1.67), 4/5 (1.8), 5/6 (1.83), 9/10 (1.9) range is the odds-on territory where welcome offer minimum odds typically sit. Every minimum-odds clause you read in a UK welcome offer is denominated in one or both of these formats, so knowing the conversion by sight saves a calculator trip.

The format matters because it changes the perception of value. A 5/2 favourite and a 3.5 decimal favourite are identical bets, but the fractional version emphasises the profit multiple (2.5 times your stake) while the decimal version emphasises the total payout multiple. On longer-shot prices, decimals become more intuitive: 25.0 decimal is easier to multiply by a stake than 24/1. On shorter prices, fractional is arguably cleaner: 1/9 (decimal 1.11) makes the tight odds-on nature immediately visible.

The Betfair Exchange displays prices in decimal by default but allows switching to fractional. Most UK fixed-odds operators allow the customer to switch. Picking one format and staying consistent is easier than flipping between them. My own preference is decimal for all calculation work and fractional only for quick mental shorthand on familiar price ranges.

Implied probability by hand: the calculation every punter should do

Implied probability is the single most useful calculation in betting. It tells you what the price says about the bookmaker’s view of a horse’s chance of winning, and it lets you compare that to your own view. The formula could not be simpler: implied probability = 1 / decimal odds, expressed as a percentage.

Examples. 2.0 decimal = 1/2.0 = 0.50 = 50% implied probability. 5.0 = 1/5 = 0.20 = 20%. 11.0 = 1/11 = 0.091 = 9.1%. 26.0 (25/1) = 1/26 = 0.038 = 3.8%. Every price converts to an implied probability in one step, and those implied probabilities should sum across the field to roughly the overround figure — 110-130% on a fixed-odds book, 102-105% on a liquid exchange market.

The practical use is value assessment. If you believe a horse has a genuine 20% chance of winning, the fair price is 5.0 decimal. Anything above 5.0 is value; anything below is negative expectation. A bookmaker offering 6.0 on the same horse is offering an implied probability of 16.7% — below your 20% estimate, meaning your expected value on a £10 bet is £10 × (0.20 × 6.0 – 1) = £10 × 0.20 = £2 of positive expectation per bet. Over a long sample, you would expect to win £2 per £10 staked.

The complication is that your own probability estimate might be wrong. Sophisticated value betting requires honest calibration against actual outcomes. A punter who believes they are picking 20% shots but is actually picking 15% shots is losing expected value across the season. Tracking your own estimates against realised outcomes — a simple spreadsheet logging claimed probability and actual result over 100+ bets — is the only way to calibrate honestly. Most recreational punters overestimate their edge by 3-5 percentage points, which is enough to turn a claimed positive EV into realised negative EV.

Summing implied probabilities across a race gives you the overround directly. Take a 6-horse race with prices of 2.5, 4.0, 6.0, 8.0, 12.0, 20.0 decimal. Implied probabilities: 40%, 25%, 16.7%, 12.5%, 8.3%, 5%. Sum = 107.5%. That is the overround — 7.5 percentage points of margin. This is a relatively tight book, consistent with exchange pricing or a very competitive fixed-odds market on a Group race. A 125% book on the same race would shorten each horse’s price proportionally, extracting roughly 20 percentage points of margin.

Tote pools: the third pricing model

Pool betting — the Tote, in UK context — is structurally different from both fixed-odds and exchange. The operator collects all stakes into a pool, takes a fixed percentage as takeout (typically 17-25%), and divides the remainder among winning bets. The final price is only known after betting closes, which means you are betting blind against the eventual odds.

The Placepot is the flagship UK Tote product — pick one horse to place in each of the first six races at a specific meeting, divide the pool among successful tickets. Jackpot is the harder sibling: pick winners of six races. Quadpot is smaller scope (four races). Scoop6 is a weekend flagship product on a Saturday card. These are multi-race products where the pool structure generates the prices, and successful tickets can pay out many multiples of the nominal cost because of the compounding difficulty.

The attraction of pool betting for a small-stakes punter is the payout asymmetry. A £2 Placepot ticket can return several hundred pounds on a complicated card; a £10 investment in fixed-odds multiples would pay out at a similar level only if the individual prices lined up perfectly. But the takeout is brutal: 25-30% on some pools. Over many Placepots, the expected return is negative even if your selection skill is neutral. The pool product is best understood as a lottery-adjacent product with a skill component, not as a value-betting mechanism.

Single-race Tote betting — the traditional Win pool, Place pool, Dual Forecast — sits differently. Takeout is lower (around 17%), and the payout is based on pari-mutuel allocation of the pool among winning tickets. For races with unusual public-favour distribution, where heavy money has distorted the fixed-odds market, Tote prices can occasionally exceed fixed-odds prices on outsiders. This is a niche opportunity but a real one.

Free bets and Tote pools rarely align. Most operators’ Tote partnerships are separate from the standard fixed-odds book, and free bet tokens are usually excluded from Tote markets — the operator has no margin on pool bets to fund the promotion. Pool betting is a cash-stake product for the most part, with the partial exception of Placepot-specific promotions that occasionally appear around festival weeks.

Where free bets fit across fixed-odds, exchange, and pool channels

Free bet tokens are a fixed-odds product. That is the starting point for thinking about channel allocation. The £766.7m UK remote racing GGY generated last year came overwhelmingly from fixed-odds operators, and the promotional spend that funds welcome offers, Extra Places, BOG extensions, and enhanced prices comes out of fixed-odds margin. Sky Bet’s £10m Extra Places liability at Cheltenham 2026 and bet365’s £50m BOG exposure are fixed-odds spend categories. None of that promotional money flows to exchange users or Tote users in the same token form.

This creates a three-channel optimisation problem for the serious UK racing punter. Channel one: fixed-odds bookmakers, for welcome offer extraction, Extra Places deployment, BOG protection, and promotional stacking. Channel two: exchange, for price-efficient pure cash-stake betting on high-liquidity markets where the 5-25 basis point price edge compounds across a season. Channel three: Tote pools, for the occasional pool-bet with favourable takeout structure or multi-race product with asymmetric payout profile.

The practical split I use is straightforward. All welcome offers, reload bonuses, and tokenised promotions go through fixed-odds operators. Any cash-stake bet on a high-liquidity exchange market — Group 1s, Grand National, Cheltenham championship races — goes through the exchange if the price is demonstrably better than the fixed-odds equivalent. Marginal markets and festival midweek handicaps get bet where the promotional value is best, which usually means fixed-odds. Multi-race speculation with small stakes and asymmetric payout goes through Tote pools occasionally.

The black-market landscape is the silent alternative hanging over all three channels. Affordability pressure on licensed UK operators has pushed some customers toward offshore providers that lack the regulatory overhead. As Lewis, former CSO of evoke, framed the sector’s direction at an illegal gambling panel: the sweeping changes of the last five years — affordability, credit card bans, VIP limits, auto-spin caps, tax changes — have made a huge difference to the industry, and if the goal were to make the black market more attractive, there is not much more that could be done. That comment sets the context for why the licensed fixed-odds promotional architecture exists in its current form. Operators are competing not only against each other but against the structural pull of the unregulated sector.

For the broader picture — how Levy economics, regulatory framework, and festival-week promotional waves shape the overall UK racing landscape — the pillar on UK free horse racing betting maps the full territory.

The mathematical punter’s channel framework for 2026

Each channel has a specific role. Fixed-odds is where promotional value concentrates — welcome tokens, Extra Places, BOG, enhanced prices — at the cost of 10-25 basis points of structural overround per bet. Exchange is where pure-price efficiency sits, at 102-105% books minus commission on net profit, at the cost of no token-based promotions. Tote is the niche channel for multi-race speculation with asymmetric payout and relatively poor takeout economics.

A serious UK racing punter operates across all three with intent. Free bet tokens and Extra Places extraction on fixed-odds; high-liquidity cash bets on exchange; occasional pool speculation on Tote. The £4 edge on a winning exchange bet at 5.4 versus fixed-odds at 5.0 is not cosmetic — it is the difference between a break-even season and a modestly profitable one for a disciplined player. The maths, once you see it, cannot be un-seen.

Created by the ”Free Horse Racing Betting” editorial team.