Most racing fans look at the last race and stop there. The horse won, or it did not, and they move on. But the last race is one frame from a moving picture. The more interesting question is whether the horse in front of you is on the way up or on the way down, and that is a question you can answer in seconds.
So how do you spot an improving horse, and how much should you lean on it? Race Advisor pulled its master file to find out, using runs from January 2020 to March 2026 for the main figures, and the answer is a useful one: improvement is easy to spot, the market already notices it, and that tells you exactly where to use it.
How do you spot an improving horse in UK racing? Compare a horse’s recent peak against its longer-term peak. If the best figure from the last 30 days is sharper than the best from the previous 90, the horse has just set a fresh peak. In an earlier Race Advisor analysis of 341,801 UK runners from January 2020 to March 2026, that “fresh peak” group won 11.8% of the time with a Betfair SP A/E of 1.02. A/E means actual winners divided by the winners the odds implied, so 1.02 says the market had already priced the improvement in. Use it as a supporting clue rather than a stand-alone reason.
What does “improving” mean in racing?
Improvement is not a feeling. It is a measurable shift in a horse’s recent figures relative to its longer trajectory. The cleanest way to read it is to put two windows side by side: a short one and a longer one.
Race Advisor’s data carries this directly. Race Advisor groups speed figures into bands, and a lower band number is the sharper figure. The ShorBE columns hold the horse’s best speed band over the last 30, 60, 90 and 365 days. The ShorAV columns hold its average speed band across the same windows. If ShorBE30 is sharper than ShorBE90, the horse has set a new peak inside the last month. If ShorAV30 is sharper than ShorAV90, the recent average is running ahead of the longer-term average. That is the textbook definition of an improver.
The reverse case matters too. A horse whose recent average has slipped behind its longer-term average is on a declining path. In the same file that group was the largest of the three: 135,285 runners on a declining recent average, against 99,996 improving and 106,520 flat. Knowing which bucket a horse sits in gives you a quick, fair read of where it is in its cycle.
What is the quick way to check?
Before the data was in the racecard, finding an improver was a manual job. You would open a ratings page, scan the figures for steady upward movement, then work out an average distance behind the winner in similar races from another site. It worked, but it took around fifteen minutes a race.
The shortcut now is to look at ShorBE30, ShorBE90, ShorAV30 and ShorAV90 on Race Advisor’s racecard. Where ShorBE30 is below ShorBE90, the horse has set a new peak inside the last month. Where ShorAV30 is below ShorAV90, it has been averaging better than its 90-day baseline. Those are your improvers, found in seconds rather than minutes. The next question is what to do with the answer.
Does improvement beat the market on its own?
On its own, improvement is priced in, which makes it a helpful guide to how the market thinks. The January 2020 to March 2026 sample in the Race Advisor master file shows it clearly:
- Fresh peak (ShorBE30 below ShorBE90): 125,579 runners, 11.8% strike rate, A/E 1.02.
- Peak older than 30 days (ShorBE30 equal to ShorBE90): 216,222 runners, 12.5% strike rate, A/E 1.02.
- Recent average improving (ShorAV30 below ShorAV90): 99,996 runners, 11.6% strike rate, A/E 1.02.
- Recent average declining (ShorAV30 above ShorAV90): 135,285 runners, 13.3% strike rate, A/E 1.02.
Read the list slowly. The “declining” group has a higher raw win rate than the “improving” group, and every group settles at the same A/E. The market has, on average, built the trajectory into the price, so a horse’s direction of travel is already reflected in the odds you see.
The same picture holds over a longer window. Compared with the full 365-day average, horses whose recent 30-day average is sharper won 10.9% of the time at an A/E of 1.02, and those whose recent average had slipped won 13.5% at 1.02. Size of the move does not change much either. Among improvers, a peak that jumped by more than 100 points (10,829 runners) returned an A/E of 1.04, and a small 1 to 5 point shift (9,928 runners) returned 1.02.
What does that tell you? Both groups have been priced by the market for what the figures show. That is useful to know, because it means the trajectory is best used as a supporting clue, alongside the price and the race.
What happens when you combine improvement with the ratings?
This is the question readers ask next, so we sliced the file on it. In the same period, adding an improvement filter on top of Race Advisor’s top-ranked horses did not lift the A/E, and the strike rate on the filtered group was a little lower. Ratings that are already strong do not need a trajectory layer to work. If you already lean on the findings in our single-rating profit test, the improvement flag is best treated as a tie-breaker rather than an upgrade.
Race type matters more than trajectory. In turf chases, the 9,680 improver runners had an A/E of 1.14, and the 17,441 non-improver runners had 1.16, so the discipline is doing the work, not the improvement flag. On the all-weather, the 44,438 improver runners came in at an A/E of 0.98. Those are the figures for those slices, so check the slice you are looking at rather than assuming the pattern carries over.
How do you use improvement signals well?
The improvement signal is real, and the best uses are practical ones.
- Confidence in staking. Two horses look equally backable on your main ratings, and one has a fresh peak while the other has slipped on the longer-term average. The improver is a perfectly sensible tie-breaker.
- A watch list. A horse that is improving on the trajectory bands but ran a flat last race is exactly the profile the bare result can mislead you on. Put it on your tracker and wait until conditions line up.
- Reading the other side. A horse with a declining recent average at a tight price is one to approach with extra care, especially when the market has not yet adjusted.
What this gives you is a quick read on form that no single result can show.
What else is worth checking alongside it?
Improvement works best as one clue among several. Useful questions to ask when you see an improver are whether a step up in class explains a last-time-out defeat, whether a course or surface switch could be behind a change in figures, whether there has been a jockey change, and whether the yard’s 14-day strike rate is moving the right way. This analysis did not test those combinations, so treat them as questions to ask rather than rules.
This is why a single rating, however well built, works best alongside others. Our work on the SHorBES speed rating made the same point from a different angle: the measure showed its best results when it was paired with the race conditions it suits. The improvement layer is a confidence input on a stack. Pair it with the price and the race context, and it becomes much more useful. For more on keeping a steady head over time, our guide to losing runs is a good companion.
What is the takeaway?
Spotting an improver is straightforward on Race Advisor. ShorBE30 below ShorBE90 means a fresh recent peak. ShorAV30 below ShorAV90 means the recent average is running ahead of the 90-day baseline. Both flags describe something true about the horse.
Across 341,801 UK runners from 2020 to 2026 in the earlier analysis, the A/E for every improvement band sat at 1.02, which tells you the market reads the same data you do. That is a helpful thing to know, because it points you towards the part of the process where you can add something of your own: the price and the context around the horse.
To see how today’s improvers line up against the market, open today’s race cards and check the ShorBE and ShorAV bands next to the ratings for any race you are interested in. Compare the trajectory signal with the price and the race context before deciding how much weight it deserves. It works best as the second thing you look at, not the first.
