The Thursday Take is our weekly column: one racing talking point, put to the test against the Proform database — every British and Irish runner, every result. This week it is a talking point nobody has raised: a number in the in-running record that has quietly fallen by a quarter.
The short version
- In 2018, 42.8% of runners traded at half their Betfair SP or shorter once the race had started. In 2025 it was 31.8%. Trading at a third of the price fell from 28.5% to 20.6%.
- It was a slide, not a jump. Flat until the middle of 2021, falling until the middle of 2025, and level since. 2026 is running at 31.9%.
- The pre-race market did not move at all. Measured against Betfair SP, winners came in at exactly the rate the prices said every single year (an A/E of 0.999 to 1.000). Only what happens after the off has changed.
- It is everywhere. Every price band, every race type, Britain and Ireland. It is biggest for horses priced 4/1 to 28/1, and biggest of all for the horses that finish second or third: three in five used to trade at half their price; now two in five do.
- We cannot tell you why, and we say so below. The top of the in-running range has narrowed too, which fits a quieter in-running market — or a change in how in-running highs and lows are recorded. Our data cannot separate the two.
- Why it matters: any in-running idea back-tested on 2018–2021 looks about ten points better than the same idea on what the record shows today.
The trade, in one paragraph
A popular way to bet on the exchanges is Double Or Bust. You back a horse at Betfair SP, then try to lay it once the race is running at half the price — back at 10.0, lay at 5.0. If the price halves at any point, you have locked in a profit whatever happens next. If it never does, you lose your stake. Its bigger brother, Treble Or Bust, lays at a third of the price for twice the reward. Neither needs a winner, only a horse that looks like a winner for a few strides. We explained it in full, with how to test it, in Double Or Bust: back-test any system as an in-running trade.
Whether either trade works comes down to one number: how often a runner's in-running price falls to half (or a third) of its Betfair SP. Because the record holds every runner's lowest and highest in-running price, we can count it for every race since 2018.
The number nobody mentioned
| Year | Traded at half its SP | Traded at a third | Double Or Bust return | Treble Or Bust return | Pre-race A/E |
|---|---|---|---|---|---|
| 2018 | 42.79% | 28.46% | −15.3% | −15.7% | 0.999 |
| 2019 | 42.17% | 28.21% | −16.5% | −16.5% | 0.999 |
| 2020 | 41.75% | 27.85% | −17.3% | −17.5% | 0.999 |
| 2021 | 40.49% | 27.23% | −19.8% | −19.4% | 1.000 |
| 2022 | 37.98% | 25.10% | −24.8% | −25.7% | 1.000 |
| 2023 | 36.76% | 24.09% | −27.2% | −28.7% | 1.000 |
| 2024 | 33.43% | 21.85% | −33.8% | −35.3% | 1.000 |
| 2025 | 31.77% | 20.60% | −37.1% | −39.0% | 1.000 |
| 2026 (to 12 Oct) | 31.94% | 20.56% | −36.8% | −39.1% | 0.999 |
That is every eligible runner in Britain and Ireland: about 120,000 a year, 1,039,022 in all. Eleven points off the half-price rate in seven years. To break even at Double Or Bust you need the price to halve on 50.5% of your bets once Betfair's 2% commission is paid, so backing every runner blind was never going to work — but it went from costing about 15p in the pound to 37p.
How to read the numbers. A runner only counts if a half-price lay was possible: Betfair's lowest price is 1.01, so a horse that started at 2.0 or shorter cannot be halved and is left out. The returns are level stakes at 2% commission. A/E (actual over expected) is winners divided by the winners the Betfair SPs implied: 1.00 means the prices were exactly right.
Look at the last column. The pre-race market is as sharp as it ever was — the Betfair SP called the winners to the third decimal place in every year. Whatever changed, it changed after the stalls opened.
A slide, and it has stopped
Split by quarter, the picture is clear. For beaten horses, the half-price rate sat at 35–37% every quarter from 2018 to the spring of 2021. It started falling in the third quarter of 2021, kept falling until the middle of 2025, and has been level since: every quarter of 2026 is within about a point of the same quarter of 2025. So this is not a recent shock, and it is not still getting worse. It is a new normal that arrived over four years.
It is not a quirk of which horses ran
The obvious objection is that the runners changed: shorter-priced fields, more favourites, gaps in the data. So we checked inside each price band.
| Betfair SP (decimal) | Half-price rate 2018 | 2025 | Change |
|---|---|---|---|
| 2.02 to 3 | 47.7% | 44.6% | −3.1 |
| 3 to 5 | 45.8% | 37.9% | −7.9 |
| 5 to 8 | 42.6% | 30.9% | −11.7 |
| 8 to 15 | 40.7% | 26.6% | −14.1 |
| 15 to 30 | 41.0% | 26.0% | −15.0 |
| 30 and bigger | 43.8% | 35.8% | −8.0 |
Every band fell. The short-priced horses barely moved; the fall is concentrated in the mid-priced runners from about 4/1 to 28/1, which lost 12 to 15 points each. It is the same story by race type and country:
| Half-price rate, 2018–19 to 2025–26 | Britain | Ireland |
|---|---|---|
| Flat turf | 42.5% → 31.2% | 38.0% → 26.0% |
| All-weather | 40.8% → 29.5% | 39.7% → 26.6% |
| Chase | 45.9% → 38.1% | 44.0% → 33.9% |
| Hurdle | 44.6% → 37.0% | 42.6% → 33.4% |
| Bumper | 46.3% → 37.3% | 47.2% → 36.1% |
The Flat lost 11 to 12 points, jumps 8 to 10. And two data checks: in races where we know the in-running record is complete (the winner traded at 1.05 or shorter), the series is identical to a twentieth of a point. And runners that never traded below Betfair's maximum price of 1,000 have become more common, but they are only about 3% of the runners and take out about a point at most.
Where you would feel it: the horse that had every chance
| Traded at half its SP, by where it finished | 2018 | 2025 |
|---|---|---|
| Winners | 100.0% | 99.8% |
| Second or third | 59.4% | 40.2% |
| Fourth or worse | 30.3% | 19.2% |
| Did not finish | 22.3% | 18.3% |
A winner nearly always halves its price, by definition. The change is in the horse that hit the front a furlong out and was collared, or came to win its race and found one better. In 2018 that horse traded at half its SP three times in five. Now it is two in five. If you trade in-running, that is the bet you used to get matched on and now often do not.
What we cannot tell you
The top of the range narrowed too. A beaten horse used to touch Betfair's maximum price of 1,000 at some point in the race almost every time: 99.9% of them in 2018, and still 99% in 2022. In 2023 it was 90%, in 2025 67%, and so far in 2026 63%. The share of beaten horses that traded below their SP at all fell from 87.7% to 71.3%.
When the lows rise and the highs fall together, there are two possible explanations. One is that the in-running market itself has become quieter, with fewer small bets matched at the extremes. The other is that the way in-running highs and lows are captured and recorded has changed. From our data alone we cannot tell those apart, and we will not pretend we can. Our figures match the published in-running prices exactly where we have checked them; that shows we record the source faithfully, not that the source has measured the same way since 2018.
Either way, the practical point for anyone who uses this data stands. What the in-running record shows today is not what it showed in 2018 to 2021, and anything tested on those years will look better than it would now.
Why it matters if you back-test
Suppose you built an in-running angle in 2022 on the four years before. A system that halved its price on 52% of its runners would have looked like a winner against the 50.5% break-even. Run the same selections on what the record shows since mid-2025 and, if the angle followed the market-wide fall, it would land somewhere around 41–42%: a clear loser. That is the size of the trap.
Three rules follow from it:
- Judge any in-running result on recent seasons first. The earlier years are not wrong, but they describe a different-looking market.
- Compare against the base rate for the same years. A system hitting 36% in 2025 is beating the field by four points; the same 36% in 2019 was six points worse than doing nothing.
- Watch the mid-priced runners. They lost the most; a strategy built on 4/1 to 28/1 shots has had the ground move most under it.
Build it yourself
None of this needs a single rule. Every figure above can be reproduced in the System Builder on any system, including one with no rules at all.
- Staking tab: choose Double Or Bust (DOB) or Treble Or Bust (TROB) as the strategy, and the System Builder settles every selection as that trade against the real in-running lows. It is back-test only, because in-running prices exist only after a race.
- Breakdowns tab: open the Year table and tick the In-running (DOB / TROB) columns. You will see your own system's half-price rate year by year, and whether it fell with the market.
- Grid tab: put Year against BSP Band and choose DOB hit % to see the price-band table above for your own selections.
- For the full mechanics, including break-even rates, read our Double Or Bust guide.
The take
The betting market before the off is as accurate as it has ever been. What happens to prices once the race is running is not the same. Since 2021 the share of runners trading at half their starting price has fallen from about 43% to about 32%, in every price band and every code, and it has stayed there for 18 months. We do not know whether the market got quieter or the way it is recorded changed. What we do know is that an in-running back-test that leans on 2018 to 2021 is flattered by about ten points, and that is worth knowing before you trust one.
Last week we asked whether jumps yards have an edge in the Cesarewitch — the trend was real, and the market already had it priced. Next week: Billy Loughnane, and whether the market has caught up with him.
Questions we get asked about this
Has the Betfair in-running market changed?
What the in-running record shows has changed. In 2018, 42.8% of British and Irish runners traded at half their Betfair SP or shorter during the race; in 2025, 31.8%. The fall ran from mid-2021 to mid-2025 and has been level since. The top of the range narrowed at the same time, which fits either a quieter in-play market or a change in how highs and lows are recorded; the data alone cannot separate the two.
Is Double Or Bust profitable?
Not by backing every runner. You need the price to halve on 50.5% of bets after 2% commission, and market-wide it happened 42.8% of the time in 2018 and 31.8% in 2025. A blind Double Or Bust lost 15.3% of stakes in 2018 and 37.1% in 2025.
Did the pre-race Betfair market get less accurate?
No. Winners divided by the winners implied by Betfair SP came out at 0.999 to 1.000 in every year from 2018 to 2026.
Which horses lost the most in-running?
Mid-priced runners and placed horses. Runners at decimal 8 to 30 fell 14 to 15 points, and horses that finished second or third fell from 59.4% trading at half their price to 40.2%.
Can I test my own system as an in-running trade?
Yes. In the System Builder, the Staking tab settles any system as Double Or Bust or Treble Or Bust, and the Breakdowns and Grid tabs show the half-price rate by year and price band.
