Before you build anything on UK sectional times data, ask a plainer question than “what does it contain”. Ask whether the races you care about have it at all. We measured that for every runner since 2017, and the answer depends on the year, the country and the course far more than most people expect.
Coverage by year
The share of runner records that carry sectional times, split between Britain and Ireland:
| Year | Britain | Ireland |
|---|---|---|
| 2017 | 18% | 0% |
| 2018 | 22% | 0% |
| 2019 | 31% | 0% |
| 2020 | 42% | 0% |
| 2021 | 42% | 0% |
| 2022 | 44% | 0% |
| 2023 | 45% | 0% |
| 2024 | 91% | 0% |
| 2025 | 91% | 0% |
| 2026 to 30 September | 91% | under 1% |
Nothing exists before 2017. If your study starts in 2011, the first six years have no timing at all.
The figure for Britain doubles between 2023 and 2024, from 45% to 91%. That is a step in the data, not a gradual climb. Anything you compare across that boundary is comparing two different populations of races.
And Ireland has none. The figure is 0.0% in every year from 2017 to 2025, and in 2026 there are 26 Irish races with some. Irish courses are 25 of the 86 in our UK and Ireland archive, so any number that mixes the two hides a very uneven picture. Blended, the coverage for 2024 and 2025 is about 68% of runners, which describes neither country.
It depends on the course
Even within Britain, coverage follows the course. In 2025 the busiest courses with at least 1,500 runners that year looked like this:
| Course | Runners with sectionals |
|---|---|
| Wolverhampton | 99.5% |
| Newmarket | 99.3% |
| Newcastle | 98.6% |
| York | 98.5% |
| Kempton | 98.1% |
| Windsor | 97.9% |
| Haydock | 97.7% |
| Lingfield | 97.7% |
| Southwell | 97.1% |
| Doncaster | 97.1% |
Ayr, the lowest British course in that group, is at 91.3%. By contrast every Irish course in the same group sits at exactly zero: Cork, Punchestown, Navan, Gowran Park, Dundalk, Leopardstown, Curragh, Naas and Fairyhouse.
Overall, 57 of the 83 courses that held races in 2025 have any sectional times, and 9,903 of the 12,842 races that year do, about 77%.
What missing means
The missing data is not random, and that matters. Take 2025. Flat all-weather runners have the highest coverage at 87%, flat turf 70%, chase 54% and hurdle 51%, and those figures include Irish racing, which drags every group down. A model trained only on races with sectionals learns about those races. It will say little about a Thursday jump card in Ireland.
Three habits help:
- Keep a flag for “has sectionals” and report results separately for each group.
- Split your validation by year, and be careful around 2023 to 2024.
- Do not impute the splits. A guess at a sectional is worse than a gap.
The check takes two lines of pandas:
df["has_splits"] = df["sectional_time_1"].notna()
df.groupby([df["date"].dt.year, "country"])["has_splits"].mean()
What you get when it is there
Each runner with sectionals has up to 18 splits. Each is a time in seconds with a label for the stretch it covers, written like Start-5f or 2f-1f. In the races we checked they sum to the finishing time, which is itself a text value such as 1m 12.45s. Alongside the splits are speed fields for the early, middle and closing parts of the race.
Our sectional times guide walks through a real six furlong race to show how to read them. For the wider picture of the UK archive, see the UK horse racing data overview.
Run that coverage check on your own sample before you decide anything. The splits are in the UK datasets and the API, and the numbers above should not surprise you once you have seen your own.
Comments are closed