The Thursday Take: Back the Jumps Yard in the Cesarewitch? The Trend Is Real. The Edge Isn't.
This is The Thursday Take — our weekly column on RaceMetrics, where we take the topic racing is talking about and put it to the test against the Proform database. No hunches, no hype. Just what the data says. This week, with the Cesarewitch on Saturday, the trend every preview of the race repeats: back the jumps horses.
This week's talking point
The Cesarewitch is a big-field handicap over two and a quarter miles at Newmarket every October, and it has a reputation: it is the Flat race the jumping yards come to win. The previews say it every year in much the same words. Back horses from yards known for jumping. Back horses that ran over hurdles or fences earlier in the year. Look for a 4 to 7-year-old carrying 9st 2lb or less, drawn 13 or lower.
Recent history makes the case for them. Willie Mullins won it three years running from 2018 to 2020; Irish yards won six of the eight runnings from 2018 to 2025. Then last year Beylerbeyi won it for a mainly Flat yard. A fortnight ago Pierre Royal won the Cambridgeshire for an Irish trainer at a Betfair price of 45, and the story was told again.
Every one of those claims is checkable, so we checked them. We start with the race itself, then ask the question that matters to a punter: not "do jumpers win the Cesarewitch", but "does the betting market already know they do?"
How to read the numbers. Every figure is measured against Betfair SP, which carries no bookmaker margin, so it is the fairest price to judge a group against. A/E (actual over expected) is winners divided by the winners the prices implied: 1.00 is par, 1.05 means 5% more winners than the market expected. The ± figure is one standard error — roughly, how far the number could wander by luck alone. A result inside two standard errors of 1.00 is not a finding.
The trends, checked against the winners
The Proform database holds every Cesarewitch since 1997. Here is how the claims in circulation stand up against the winners themselves:
| The claim | What the winners say |
|---|---|
| 13 of the last 23 winners came from jumps yards | 12, or 13 counting one yard that ran 44% of its horses over jumps |
| 11 of the last 12 were aged 4 to 7 | 11 — confirmed |
| ...and carried 9st 2lb or less | Age and weight together: 10, not 11 |
| ...and had run over jumps earlier that year | 7 in the same calendar year; 10 in the previous twelve months |
| 14 of the last 23 drawn 13 or lower | 13 |
Near enough, give or take a definition. The trends are not made up. But a trend tells you who won; it does not tell you whether the winners were a surprise. A horse can fit every box and still have been priced for it.
One thing the trends never mention: the race is shrinking. For twenty years the Cesarewitch had 30 to 36 runners. It had 21 in 2022, 24 in 2024 and 19 last year.
Inside the Cesarewitch: the market saw them coming
| Cesarewitch, 2008–2025 (18 runnings, 554 runners) | Share of runners | Share of expected winners | Winners |
|---|---|---|---|
| Ran over jumps in the previous 12 months | 39% | 46% | 11 (8.4 expected) |
| No jumps run in the previous 12 months | 61% | 54% | 7 (9.9 expected) |
| Drawn 1 to 13 | 42% | 47% | 9 (8.7 expected) |
The jumpers did win more than their share: 11 races from 39% of the runners. But the exchange was already giving them 46% of the chance. Eleven winners against 8.4 expected looks like a 31% edge, and with 18 races it is under one standard error — luck alone produces a gap that size roughly one time in five. And the low draw, the other famous trend, won exactly what its prices said it would: 9 against 8.7.
Eighteen races cannot tell an edge from a lucky decade. So we widened the question to every race that looks like it.
Every staying Flat handicap: the jumpers win less often
| Flat handicaps, 14 furlongs and over, Britain and Ireland, 2018 to Oct 2026 | Runners | Strike rate | A/E | Level stakes |
|---|---|---|---|---|
| No jumps run in the previous 12 months | 17,858 | 11.62% | 0.991 ±0.022 | −1.7% |
| Ran over jumps in the previous 12 months | 9,284 | 9.70% | 1.022 ±0.034 | +1.3% |
Across 27,142 runners, horses with recent jumps form win less often than the rest, not more. Measured against their prices they come out at 1.022 — under one standard error from par. The previews also talk about jumps yards, which is not the same thing, so we sorted every trainer by the share of their runners that went over jumps the year before:
| Same races, by the trainer's runs the previous year | Runners | Strike rate | A/E |
|---|---|---|---|
| Flat yard (under 10% jumps runners) | 12,932 | 12.59% | 0.990 |
| Dual-purpose (10–49%) | 6,433 | 9.51% | 1.008 |
| Jumps yard (50% or more) | 6,888 | 9.45% | 1.005 |
Jumps yards win less often than Flat yards in these races, and the market prices that to the second decimal place.
Where the jumpers do win more: big fields and long trips
The trend is not invented, and the place it lives is exactly the Cesarewitch's shape. In big fields and at the longest trips, horses with jumps form do win more often than the rest:
| Staying Flat handicaps, 2018 on | No jumps form: strike rate / A/E | Jumps form: strike rate / A/E |
|---|---|---|
| Fewer than 14 runners | 12.80% / 0.994 | 10.92% / 1.017 |
| 14 to 19 runners | 6.32% / 0.957 | 6.79% / 1.055 |
| 20 or more runners | 3.99% / 0.972 | 4.81% / 1.009 |
| 2m2f and further | 7.56% / 0.907 | 9.32% / 1.106 |
In fields of 20 or more, jumps-form horses are 48% of the runners, 52% of the expected winners and 53% of the actual winners. That is the trend and the market's answer to it in one line: they win more, and they are priced to. At two and a quarter miles and beyond the gap looks wider, 1.106 against 0.907, but that is 52 expected winners — under one standard error. No figure in this table clears the bar.
The full trends table, tested on races it was not built on
Trends are usually quoted together, as a checklist. So we took the Cesarewitch's own checklist — aged 4 to 7, carrying 9st 2lb or less, a jumps run in the previous year, drawn 1 to 13 — and applied it to every other staying Flat handicap of 14 runners or more since 2018, leaving the Cesarewitch itself out.
| Big-field staying handicaps (not the Cesarewitch) | Runners | Winners | Expected | A/E | Level stakes |
|---|---|---|---|---|---|
| Every runner | 4,914 | 305 | 305.3 | 0.999 | −1.9% |
| All four trends | 479 | 37 | 29.1 | 1.270 ±0.185 | −0.1% |
| of which 2018–2021 | 232 | 14 | 14.2 | 0.988 | −7.2% |
| of which 2022–2026 | 247 | 23 | 15.0 | 1.538 | +6.6% |
The headline is tempting: 37 winners where the prices expected 29. But it returned nothing at level stakes, because the extra winners came at shorter prices. And the split is the giveaway. In the first four years — the years you would have used to decide the checklist worked — it did nothing at all, 0.988. All of the effect is in the second half, which nobody could have chosen in advance. Take the draw out of the checklist and the second half flips the other way. That is what a pattern found by looking backwards usually looks like.
Why this is not a system. A checklist that scores 1.27 overall, 0.99 in the half you would have built it on and nothing at level stakes is a loose thread to keep watching, not something to back. Trends tables are a way of describing past winners. They are not, on this evidence, a way of finding future ones.
And the Irish raiders?
Pierre Royal's Cambridgeshire revived another favourite line: the Irish target Britain's big handicaps. Across British Flat heritage handicaps since 2018, Irish-trained runners have won 20 races from 329 runs against 21.5 expected — an A/E of 0.932 ±0.22. Over staying trips it is 11 winners from 10.7 expected. Priced right, the same answer we found for Irish raiders in general last month.
Build it yourself
As of this week both rules are in the System Builder, so you can check every figure above and test the ideas on any race you like.
- Jumps Runs (last 365 days) on the Record tab: the horse's runs over hurdles, fences and in bumpers in the year before the race. Set it to 1 or more on Flat handicaps of 14 furlongs and over, from 2019, and the System Builder gives 8,220 runs, 800 winners against 782 expected: 1.023, about level at Betfair SP. Its breakdown splits it 0, 1, 2, 3–4 and 5 or more runs.
- Trainer Jumps % (previous year) on the Connections tab: the share of the trainer's runners last calendar year that went over jumps. 50 or more is a jumps yard. On the same races: jumps yards 1.006, Flat yards 0.989. Its breakdown, Trainer Code Mix, puts every yard into Flat, Dual-purpose or Jumps.
- The full trends checklist is four rules: Age 4 to 7, Weight Carried 128lb or less, Draw 1 to 13 and Jumps Runs (last 365 days) 1 or more, plus Field Size 14 or more. From 2019 the System Builder shows 37 winners against 28 expected. Then open the Holdout view, which splits it into an early period and a recent one, and watch what happens.
- Open the Grid and put Trainer Code Mix against Field Size, or Jumps Runs (last 365 days) against Distance, to see where the jumpers do win more and what the market made of it.
- Both are known before the race, so they work in today's qualifiers, and both are columns in Data Export.
The take
The Cesarewitch trend is real about winners and wrong about edges. Horses with jumps form do win the big staying handicaps out of proportion to their numbers, and the betting market has noticed: it prices them up by almost exactly the amount they win. Across every staying Flat handicap the jumpers win less often than the rest and return par, and the famous checklist only works in the years after you would have needed it.
None of that says don't back a jumper on Saturday. It says back one because you like the horse at the price, not because it ticks a box every preview has already ticked for you — and the market with it.
Last week we asked whether the betting market underestimates women riders — it prices the professionals right, and lags only on the riders still claiming. Next week: the in-running market, and a change in it nobody has mentioned.
The full article — with five common questions answered — is on the RaceMetrics blog: Back the jumps yard in the Cesarewitch? The trend is real. The edge isn’t..