The Thursday Take is our weekly column: one racing talking point, put to the test against the Proform database — two decades of British racing, every runner, every result. This week the talking point came with a date on it.
The short version
- Two in five British races now pay two each-way places at most. The share of races with fewer than eight runners has gone from roughly one in five in 2004–08 to 42.5% in 2026. Within that, 7.5% of races are win-only affairs of fewer than five runners, against 2–3% two decades ago. This is the punter-facing price of the fixture list, and as far as we can tell nobody has published it.
- The cause is division, not mystery. Average field = runs ÷ races. Since 2004 British racing has added 1,548 races — an 18% expansion — while the runs available to fill them fell 7%. Average field size fell 21%, from 10.82 to 8.52.
- The horses were never there to fill them. The running population peaked in 2008 at 20,231 and is 11% below that today. Nor is the remaining herd being run harder to compensate — runs per horse has gently fallen, 5.12 to 4.78.
- The BHA's own number checks out. Its 2027 policy blog says horses in training have fallen "between one and two per cent each year" since the pandemic. Measured on horses actually running: −1.7% a year since 2021. They have the number right.
- It is a jumps crisis first. One British jumps horse in five has gone since 2008 while jumps races increased — costing nearly three runners a race, from 10.70 down to 7.85.
- Ireland is the control group, and it went the other way. Every Irish code has grown its running population over the last decade, and every Irish code fields between 11.2 and 11.6 runners. An Irish jumps race carries 3.6 more runners than a British one.
- The 2027 policy pulls the same lever that caused it. Eighteen fewer fixtures, the same total races, achieved by redistributing them onto the cards that remain. Fatter cards are one of the two things that shrank the fields in the first place.
- Build it: System Builder — Course Nation = England, Scotland, Wales, then put Field Size (exact) on the breakdown to see where your own bets actually land. Switch Course Nation to Ireland and watch the whole distribution move right.
This week's talking point: the fixture list
On Thursday 27 August — six days before this column — the BHA published its fixture list policy for 2027. Richard Wayman, its Director of Racing, set out roughly 1,440 fixtures for 2027 against about 1,458 for 2026, with the total number of races held broadly flat: courses that give up a fixture get its races redistributed across the ones they keep. He noted that horse numbers have been falling by between one and two per cent a year since the pandemic. And then he wrote the sentence that made this week's column write itself:
“…even though that means field sizes are likely to remain under pressure at times of the year.”
Richard Wayman, BHA Director of Racing, Fixture List Policy 2027, 27 August 2026
That is the regulator conceding the finding. What it does not do — what nobody has done publicly, as far as we can find — is put a number on it. Ed Grimshaw made the same argument from the other direction in his Cutting Comment column, that the fixture list is being drawn as though the horses were still there. He is right, and he is roughly right about the scale. We have the database, so we measured it properly: 218,951 British races and 2.04 million runs, from 2004 to July 2026.
Start with what it costs you
Most of the field-size argument is conducted in the language of owners, racecourses and media rights. Here is the version that lands on a betting slip.
Under eight runners, each-way pays two places instead of three. Under five, it pays nothing but the win. That threshold is not a matter of opinion — it is the same in every shop and on every exchange — and British racing has walked steadily across it.
| Period | Races under 8 runners | Races under 5 (win only) |
|---|---|---|
| 2004–2008 | about 1 in 5 | 2–3% |
| 2025 (full year) | 39.7% | 6.4% |
| 2026 (January–July) | 42.5% | 7.5% |
Two in five. If you are an each-way punter, the fixture list has quietly removed a third of your places from two races in every five you look at, and the process took twenty years, which is exactly why nobody noticed it happening.
The identity that ends the argument
Everything else is commentary. You can argue for a week about incentives, ownership costs, prize money and breeding — and those arguments matter — but you cannot argue with a division. There are only two terms on the right-hand side, and we can measure both.
Since 2004, British racing has added 1,548 races, an 18% expansion. Over the same period the runs available to fill them fell 7%. Divide the second by the first and average field size falls 21%, which is precisely what it did: 10.82 to 8.52.
Note carefully what is not being claimed. Measured from 2004, the number of horses running is essentially unchanged — 18,091 then, 18,048 in 2025. The population did not collapse from a 2004 baseline. It rose to a peak of 20,231 in 2008 — by which point the fixture list had already grown by a tenth — and has fallen 11% from that peak since, while the cards kept getting fatter. And the remaining horses are not being flogged harder to cover the gap: runs per horse has drifted down, 5.12 to 4.78. More races, the same or fewer horses, each running slightly less often. The fields had nowhere to go.
The core series
| Year | Races | Avg field | % under 8 runners | Horses running | Runs per horse |
|---|---|---|---|---|---|
| 2004 | 8,570 | 10.82 | 21.7 | 18,091 | 5.12 |
| 2005 | 8,587 | 11.02 | 18.1 | 18,674 | 5.07 |
| 2008 | 9,493 | 10.32 | 22.9 | 20,231 | 4.84 |
| 2011 | 10,143 | 9.30 | 33.8 | 19,329 | 4.88 |
| 2014 | 10,104 | 8.66 | 40.8 | 18,077 | 4.84 |
| 2019 | 10,082 | 9.06 | 35.9 | 19,272 | 4.74 |
| 2022 | 10,214 | 8.46 | 42.3 | 18,561 | 4.65 |
| 2025 | 10,118 | 8.52 | 39.7 | 18,048 | 4.78 |
| 2026 (Jan–Jul) | 5,856 | 8.43 | 42.5 | 14,256 | — |
Actual runners, non-runners excluded. The 2026 row is January to July and must only be read like-for-like. On the same seven-month window, 2026 (8.43, 42.5% under eight) is the second-worst on record, behind 2022 (8.19, 46.6%). The back half of the year runs bigger fields, so full-year 2026 will settle a little above 8.43. It is not a record and we are not going to call it one.
The BHA's own number verifies. The 2027 policy says horses in training have declined "between one and two per cent each year" since the pandemic. Measured independently, on horses actually running rather than horses in training: −1.7% a year since 2021 on full years, −1.6% a year on the January–July window. They have the number right. What follows is the argument about what to do with it.
Where the horses went: it is a jumps crisis first
| Code | Horses 2008 | Horses 2025 | Change | Races 2008 → 25 | Avg field 2008 → 25 |
|---|---|---|---|---|---|
| Jumps | 10,063 | 8,018 | −20% | 3,367 → 3,579 | 10.70 → 7.85 |
| Flat turf | 9,881 | 8,754 | −11% | 3,641 → 3,747 | 10.56 → 8.81 |
| All-weather | 7,579 | 7,842 | +3% | 2,485 → 2,792 | 9.47 → 9.01 |
One jumps horse in five has gone since 2008, and the number of jumps races went up over the same span. The result is a code that has lost nearly three runners from the average race and bottomed out at 7.73 in 2022. Flat turf lost one horse in nine and nearly two runners a race.
The all-weather is the instructive one. Its population has essentially held since 2008 — and its fields still fell, from 9.47 to 9.01, because the all-weather programme grew fastest of the lot, up 80% in races since 2004. Go back to 2005 and an all-weather race averaged over eleven. A stable pool of horses, diluted purely by being asked to fill more races: that is the whole mechanism in one code, with the population variable held still.
Ireland: the control group
| Code | Horses 2015 | Horses 2025 | Change | Avg field, Ireland | Avg field, Britain |
|---|---|---|---|---|---|
| Jumps | 4,305 | 5,003 | +16% | 11.49 | 7.85 |
| Flat turf | 2,550 | 3,373 | +32% | 11.61 | 8.81 |
| All-weather | 1,261 | 1,570 | +25% | 11.23 | 9.01 |
Every Irish code has grown its running population over the last decade. Every Irish code fields between 11.2 and 11.6 runners — numbers Britain last saw in the mid-2000s. An Irish jumps race today carries 3.6 more runners than a British one.
Same breeding industry, same island group, same weather, a restrained calendar — and the opposite result. This is why the decline is not something that simply happens to a sport. It is worth saying plainly what this comparison can and cannot support: it is strong evidence that a shrinking field is not inevitable, and it is not proof of which Irish policy did it.
When it bites: February and March
The 2027 policy says fields will stay under pressure "at times of the year". The times have names.
| Month | Avg field 2004–08 | Avg field 2022–26 | Under 8: then | Under 8: now |
|---|---|---|---|---|
| February | 10.24 | 8.10 | 23.2% | 44.1% |
| March | 10.50 | 8.35 | 24.2% | 44.7% |
| May | 10.46 | 8.46 | 23.5% | 41.3% |
| July | 10.32 | 8.44 | 23.3% | 41.2% |
| October | 11.58 | 9.13 | 14.3% | 31.7% |
February and March — the heart of the jumps season, the months in which British racing is meant to be at its best — now send nearly half of all races to post with fewer than eight runners. And the damage is not confined to the winter: October, historically the month of the 11.6-runner average, has lost two and a half runners a race.
The base of the pyramid is dissolving
| Class | Horses 2008 | Horses 2025 | Change | Avg field 2008 → 25 |
|---|---|---|---|---|
| Class 1 | 2,321 | 2,290 | −1% | 10.53 → 9.05 |
| Class 2 | 2,925 | 3,999 | +37% | 11.39 → 10.19 |
| Class 3 | 5,185 | 4,064 | −22% | 9.26 → 7.78 |
| Class 4 | 11,472 | 9,532 | −17% | 10.24 → 8.01 |
| Class 5 | 11,583 | 10,252 | −11% | 10.38 → 8.39 |
| Class 6 | 7,176 | 3,847 | −46% | 10.60 → 9.14 |
The top of the sport is intact, and Class 2 has grown by more than a third — the "premierisation" of the programme, doing exactly what it was designed to do. The damage is underneath. The competitive middle, Classes 3 and 4, now averages 7.8 and 8.0 runners. And the Class 6 population has nearly halved.
Class 6 fields only look survivable because the survivors are being asked to do far more: 2.8 Class 6 runs per Class 6 horse in 2008, against 4.4 now. The base of the pyramid is being recycled rather than replenished. One honest caveat on that table — class boundaries and race classifications have shifted over two decades, so read the direction of travel, which is robust, and not the second decimal place, which is not.
The mechanism — and why 2027 repeats it
| Year | Fixtures | Races per card | Cards of 6 or fewer | Cards of 8 or more |
|---|---|---|---|---|
| 2004 | 1,299 | 6.60 | 50.0% | 9.9% |
| 2008 | 1,423 | 6.67 | 41.0% | 7.9% |
| 2013 | 1,430 | 7.10 | 13.4% | 22.9% |
| 2019 | 1,444 | 6.98 | 19.4% | 19.0% |
| 2025 | 1,427 | 7.09 | 25.7% | 26.6% |
Two levers were pulled, one after the other. First ten per cent more fixtures, nearly all of it in place by 2008; then fatter cards — 6.60 races per fixture in 2004 against about 7.1 now, with half of all cards running six races or fewer twenty years ago against a quarter today. Multiply the two and you have the 18% race growth that ate the field sizes.
Now read the 2027 policy against that table. Eighteen fewer fixtures. The same total number of races. Achieved by redistributing the races from the fixtures given up onto the cards that remain — which the policy describes, reasonably enough, as reducing the number of six-race cards.
That is the second lever. Pulled again, in the same direction, and presented as reform. We would rather be wrong about this, and it is a genuinely difficult decision made by people who can read their own data — Wayman says as much in the blog, that it was not an easy decision and that some will disagree. But the arithmetic does not care how hard the decision was: if the number of races is held flat and the horse population keeps falling one to two per cent a year, the average field falls again. That is not a forecast. It is the same division as before.
An idea, not a solution: the six-race card
Criticism is cheap, so here is something to shoot at. It is offered as a stopgap while the ownership incentives do their much slower work — a bridge, not a fix for the herd.
The rule: turf cards capped at six races. All-weather exempt. Championship cards exempt. Modelled on what actually happened in 2025:
| 2025, British turf | As it ran | Under the cap |
|---|---|---|
| Turf fixtures | 1,076 | 1,076 (none lost) |
| Turf races | 7,326 | about 800 fewer |
| Turf average field | 8.34 | about 9.4 |
| British average field, all codes | 8.52 | about 9.3 |
| Races under 8 runners | 39.7% | about 34% |
Not one fixture is lost — which matters, because fixtures are what racecourses actually defend. A fixture is the attendance-and-media unit; the seventh race on a Tuesday card is not. And the gain is not a summer-flat artefact. It bites in both codes:
| Card type | Fixtures | Cards of 7+ races | Races cut | Field now | Field capped |
|---|---|---|---|---|---|
| Jumps | 533 | 64% | 360 | 7.85 | about 8.7 |
| Flat turf | 543 | 73% | 436 | 8.81 | about 10.0 |
| All-weather (exempt) | 351 | 92% | — | 9.01 | 9.01 |
Forty-five per cent of the races cut are on jumps cards, which is where the crisis is. Capped, the jumps average climbs back to about 8.7 and flat turf to about ten — levels neither code has reached in a normal season for more than a decade. Even February — where 74% of turf cards still run seven or more races — would go from 8.00 to roughly 9.1.
The all-weather exemption is a choice with a visible price: all-weather cards average eight races and would reach about twelve if capped. We would exempt it anyway. It is the one code whose horse population has held, and winter evening racing has its own economics.
The championship exemption costs less than people assume, because the excess does not live where they think. Cards with three or more Class 1 races — the Cheltenham Festival, Royal Ascot, the Ebor meeting and their kind, 64 cards in 2025 — hold just 76 of the 872 excess races. Six hundred and forty-seven of them sit on ordinary cards with no Class 1 race at all. Exempt every festival and the cap still removes about 800 races.
And the precedent is the point: this is the 2004 card. That season averaged 6.60 races per fixture, half of all cards were six or fewer, and fields averaged 10.82. The cap does not invent a new model of British racing. It restores the card shape the sport had before the expansion, without surrendering a single fixture.
What it does to prize money
The obvious objection is that eight hundred races is eight hundred prize funds. It is not, because the money does not follow the races — it follows the idea.
Applying the championship exemption to the money as well as to the race count, the 796 races actually cut carried £6,950,471 of prize fund in 2025, or 4.2% of the £164.7m British turf-and-jumps total. That is valued conservatively, by assuming each capped card gives up its cheapest races. Redistribute that fund across the races that remain and the average prize fund per turf race rises from £22,488 to £25,229 — up 12.2%. Which is precisely the owner-economics incentive the population problem needs, funded from within the sport rather than asked for from outside it.
Or retain a slice of it. Roughly £7m is the same order of magnitude as the "several million" the BHA fears losing in media-rights and levy income from fewer races — so the policy's own cost line could be self-insured by the races it removes.
The exemption spares the dear races. The 76 races spared by the championship exemption are the dearest on the whole list: they sit on Royal Ascot, Cheltenham and Ebor-class cards, where even the cheapest race on the card averages £43,743 against £8,731 everywhere else. Sparing fewer than one excess race in ten therefore spares nearly a third of the money.
The honest ledger, stated before anyone else states it. One: those 800 races carry race-level picture payments, and that is exactly where the "several million" objection attaches; the counterweight — bigger fields make viable each-way markets, bigger pools and more levy per race — is real but not quantifiable from our data, so we are not going to pretend to quantify it. Two: fewer races means more eliminations in oversubscribed handicaps, and horses that cannot get a run. Three: a cap is a blunt instrument. Which six races survive matters enormously, and thoughtful programme design would choose better than arithmetic would. Which is why it is an idea, and not a solution.
Build it yourself
The whole of this article is one measurement — how many horses are in the race — and it is one you can put on your own screen in about a minute. It is also the single most under-used breakdown in the System Builder, because field size is not a filter you set out to test; it is the thing quietly deciding what your bets are worth.
Recipe: where your bets actually land
| Setting | Value |
|---|---|
| Course Nation | England, Scotland, Wales |
| Everything else | leave alone |
| Breakdown | Field Size (exact) |
That gives you every exact runner count as its own row, which is the view the banded split cannot give you — and exact is what matters here, because the each-way terms change at exact numbers. The fifth runner adds the second place. The eighth adds the third. Look at how much of your action sits at five, six and seven, and you are looking at the part of your betting the fixture list has been quietly taxing.
Then change Course Nation to Ireland (inc NI) and run it again. The entire distribution slides to the right. That is the control group, on your own screen, in one click.
If you would rather see the drift over time, put the two together on the Grid tab — rows = Year, columns = Field Size — and read the small-field columns fattening year on year.
One honest limit. The System Builder's data starts in 2018, so it will show you the last eight years of this, not the full twenty-two. The long series above comes from the same database going back further than the back-tester is allowed to reach. The eight years it does cover carry the trend perfectly well — and unlike our tables, they run against your own rules rather than ours.
Not a Proform Premium member? The System Builder answers questions like this one in about ninety seconds, and runs your saved systems against every declared runner each morning. See what's included.
FAQ
How much have British field sizes actually fallen?
The average British field peaked at 11.02 runners in 2005 and stood at 8.52 in 2025, a fall of 23% from the peak (21% measured from 2004). In the first seven months of 2026 it was 8.43. Measured across 218,951 British races since 2004, that is roughly two and a half runners removed from the average race.
Why are field sizes falling in British racing?
It is arithmetic before it is anything else: average field equals total runs divided by races run. Since 2004 Britain has staged 1,548 more races — an 18% expansion, delivered by 10% more fixtures and fatter cards — while the runs available to fill them fell 7%. The horse population peaked in 2008 and is 11% below that peak, and runs per horse has fallen slightly rather than risen. Why the ownership base shrank is a separate argument the data does not settle.
What does a smaller field mean for each-way betting?
Under eight runners, each-way pays two places instead of three; under five runners it pays win only. The share of British races with fewer than eight runners has gone from about one in five in 2004–08 to 42.5% in 2026, and the win-only share from 2–3% to 7.5%. Two races in every five now pay at least one place fewer than they would have twenty years ago.
Is it worse over jumps or on the Flat?
Jumps, clearly. The British jumps population fell 20% between 2008 and 2025 while the number of jumps races rose, taking the average jumps field from 10.70 to 7.85. Flat turf fell 11% in horses and from 10.56 to 8.81 in fields. The all-weather population actually grew 3% — and its fields still fell, from 9.47 to 9.01, because the all-weather race programme grew 80% since 2004.
Do Irish races have bigger fields than British ones?
Substantially. Every Irish code fields between 11.2 and 11.6 runners against 7.9 to 9.0 in Britain, and an Irish jumps race carries 3.6 more runners than a British one. Ireland also grew its running population in every code over the last decade — jumps +16%, flat turf +32%, all-weather +25% — while Britain's fell. (These count horses running in each jurisdiction rather than horses trained there.)
Which months have the smallest fields in British racing?
February and March. February now averages 8.10 runners with 44.1% of races under eight; March averages 8.35 with 44.7%. Twenty years ago both months averaged over ten runners with under a quarter of races beneath the eight-runner mark. They are the heart of the jumps season, which is also the code that has lost the most horses.
Will the BHA's 2027 fixture list improve field sizes?
On its own arithmetic, no. The 2027 policy reduces fixtures by about 18 while holding the total number of races broadly flat, redistributing races onto the cards that remain. Fatter cards are one of the two levers that shrank fields in the first place. The policy itself concedes that field sizes are "likely to remain under pressure at times of the year".
What is the six-race card idea?
A stopgap we have modelled, not a policy anyone has adopted: cap turf cards at six races, exempt the all-weather and exempt championship cards. On 2025 actuals it removes about 800 races while losing no fixtures at all, lifts the turf average field from 8.34 to about 9.4 and the British average to about 9.3, and cuts the share of races under eight runners from 39.7% to about 34%. Redistributing the £6.95m of prize fund attached to the cut races would raise the average fund per turf race 12.2%. It is an idea offered to be shot at, with its costs stated.
The Thursday Take runs every Thursday morning, with a special whenever the big meetings come around. All figures from the Proform database, queried 26 August 2026 and independently re-verified. Scope: British races that were run and resulted, 2004 to July 2026 — 218,951 races and 2.04 million runs. Irish coverage begins in 2009; "Ireland" means courses in the Republic and Northern Ireland, which race under one jurisdiction. "Horses" means distinct horses actually running in that jurisdiction in that period, which is a different and slightly smaller measure than horses in training. 2020 is distorted by the COVID suspension throughout and should not be trended through. 2026 figures are January to July only and are compared only against the same seven-month window in earlier years.
Two basis notes, because they are the two places this subject trips people up. Fixtures: our counts are fixtures that ran and were resulted — about 1,427 in 2025 — and will always read lower than the BHA's published fixture list, which counts fixtures scheduled, abandonments included. The two are never comparable and we have not compared them. Thirsk on 28 August 2026 is the live illustration: carded for six races, stopped after two, and the four that never ran are simply absent from the results data. Prize money: every figure here is the published total fund recorded against each race — the round advertised numbers, complete for every race from 2009, totalling £194.8m across British racing in 2025. It is deliberately not built from the individual place payments, which under-record sixth to eighth place from 2023 onward.
One footnote for completeness: British hunter chases stopped entirely on 20 May 2026 under an equine influenza restriction and had not resumed by the end of August. That is a disease measure rather than structural decline, and at roughly 11 races against a British year of about 10,000 — 0.11% — it does not move any figure in this article.
Prompted by the BHA's Fixture List Policy 2027 blog (Richard Wayman, 27 August 2026) and by Ed Grimshaw's Cutting Comment, both of which got there first. We just had the database.
