Best Horse Racing Betting Sites – Bet on Horse Racing in 2026
Loading...
What 380,000 Race Starts Tell Us About SP
Opinion is cheap in horse racing. Everybody has a theory about which prices offer value, where the bookmaker’s edge is sharpest, and whether favourites or outsiders represent the smarter long-term play. Data is more useful — and in the case of SP returns, the data is both extensive and unforgiving.
Research compiled through the NBER, drawing on approximately 380,000 starts in Great Britain, provides one of the deepest available analyses of how SP returns vary across the full price spectrum. The picture that emerges is consistent, well-documented, and directly actionable for any punter who bets at starting price. The numbers don’t lie — but they do surprise, particularly at the extremes of the market where the gap between what the price implies and what actually happens is widest.
ROI Table: SP Evens Through to 100/1+
The most comprehensive publicly available breakdown of SP returns by price comes from FlatStats, which analysed a multi-year dataset of UK turf racing. The data is presented by SP bracket, with each row showing the implied probability, the actual strike rate, and the resulting ROI.
| SP Bracket | Implied Probability | Actual Strike Rate | ROI |
|---|---|---|---|
| Evens (1/1) | 50.0% | ~48.0% | −3.87% |
| 2/1 | 33.3% | ~32–34% | ~−6% to −8% |
| 4/1 | 20.0% | ~17.15% | −14.23% |
| 10/1 | 9.1% | ~7% | ~−30% |
| 20/1 | 4.8% | ~3% | ~−45% |
| 33/1 | 2.9% | ~1.24% | −57.67% |
| 50/1+ | <2% | <1% | <−60% |
The pattern is unmistakable and monotonic: the longer the price, the worse the return. At Evens, the punter loses roughly £3.87 for every £100 staked — a margin that is tight enough for a skilled selector to overcome. At 4/1, the loss widens to over £14 per £100. By 33/1, the market is extracting nearly 58p from every pound wagered. Backing horses at 50/1 and above at SP is, in aggregate, the most expensive bet in the market.
These are not cherry-picked figures. They reflect the structural reality of how bookmakers construct their books. The overround — the built-in margin — is not distributed evenly across the field. It is concentrated at the long end of the market, where recreational punters are most willing to accept poor value in exchange for the excitement of a large potential payout. The bookmaker knows this and prices accordingly.
For the data-literate punter, the table is a cost sheet. It tells you exactly how much the market charges at each price point. Any profitable strategy must generate enough edge to overcome these costs, and the table makes clear that the hurdle at 4/1 is four times as high as the hurdle at Evens. Selecting the right horse is necessary. Selecting the right price range is equally important.
Strike Rate vs Implied Probability: Where the Gap Widens
The ROI figures above are the end product. The mechanism that produces them is the gap between implied probability and actual strike rate — the favourite-longshot bias expressed in percentages.
At Evens, the implied probability is 50% and the actual strike rate is approximately 48%. The gap is two percentage points — small, and consistent with a modest bookmaker margin. The market is pricing these runners with reasonable accuracy. The overround is present but not punitive.
At 4/1, the implied probability is 20% but the actual strike rate drops to approximately 17.15%. The gap has widened to nearly three percentage points. The market is overestimating the chance of these runners by a meaningful margin, and the punter is paying the difference with every bet.
At 33/1, the gap becomes a chasm. The implied probability is 2.9%, but horses at this price actually win only about 1.24% of the time. The market says these horses have a roughly one-in-34 chance; the data says it is closer to one in 80. The punter who backs 33/1 shots at SP is paying for a horse with a 3% chance when the true chance is barely 1%. The difference — nearly two percentage points on a base of 3% — represents an enormous proportional error, and it compounds into the brutal −57.67% ROI.
The widening gap is not a conspiracy. It reflects the aggregate behaviour of the betting public. Recreational punters overbet longshots because the potential payout is exciting. This demand keeps longshot prices shorter than they should be, inflating the implied probability beyond reality. Bookmakers have no incentive to correct this mispricing — it is profitable for them — and so the bias persists, year after year, dataset after dataset.
Trends Over Time: Is SP Becoming Fairer?
The favourite-longshot bias is persistent, but it is not static. Research by Smith and Vaughan Williams, published in the International Journal of Forecasting, found that the bias in UK Flat racing was measurably smaller in the decade after 2000 compared with the preceding period.
Several structural changes explain the improvement. The arrival of betting exchanges in 2000 introduced a price-discovery mechanism free of the bookmaker’s overround. Exchange prices — particularly at longer odds — tend to be more accurate than bookmaker prices, and their visibility exerts a corrective force on the traditional market. When the exchange says a horse is 40/1 and the bookmaker has it at 25/1, sharp money bridges the gap.
The professionalisation of punting syndicates, armed with computational models and large bankrolls, has also pushed the market toward efficiency. These operators identify and exploit the bias systematically, laying overbet longshots on the exchange and backing underbet favourites. Their activity narrows the gap between implied and actual probability at both ends of the market.
Greater public access to data — form databases, race replays, statistical tools — has flattened the information asymmetry that once separated insiders from the public. The average punter in 2026 has more analytical firepower than a professional handicapper had in 1996. This democratisation does not eliminate the bias, but it reduces the degree to which uninformed betting distorts prices.
The trend is encouraging but not transformative. The bias is shrinking, not disappearing. The ROI at each price point has improved modestly over time, but the fundamental structure — favourites returning more than outsiders — remains intact. The market is fairer than it was, but it is still not fair.
What This Means for Your Betting
The data supports several practical conclusions for SP bettors.
First, concentrate your SP activity in the price ranges where the market’s drag is lowest. The Evens to 3/1 band is the most cost-efficient segment of the market. You need less edge to generate a positive return, because the structural penalty for being there is smaller. This does not mean you should only bet favourites — it means that the same level of selection skill produces a better financial outcome at shorter prices.
Second, if you bet at longer prices, take a fixed price rather than SP whenever possible. The FlatStats data shows that the overround on outsiders is where the bookmaker extracts the most margin. A fixed price taken in the morning — particularly with Best Odds Guaranteed — can sometimes capture a number longer than the eventual SP, partially offsetting the structural disadvantage. The exchange is an even better alternative at longer prices, where BSP typically offers a tighter margin than bookmaker SP.
Third, evaluate your record by price range, not in aggregate. A betting record that shows +5% ROI overall might be hiding −25% at longer prices, cross-subsidised by strong performance at the short end. Breaking your results into price bands reveals where your edge actually lives — and where the market is eating you alive. The numbers don’t lie. Listening to them is optional. Profiting from them is not.