Research · 15 Oct 2026
The excuse test: do unlucky runs win next time?
Every punter has backed a horse because it was “desperately unlucky last time”. The HKJC writes a trip note for every runner, so for once the oldest excuse in racing can be tested instead of argued. We flagged every blocked, checked, shuffled-back and wide-without-cover run in 33,857 official comments and asked one question: do those horses beat the base rate at their next start?
They don't. Not even slightly.
6,591 next starts followed a trouble-flagged run. Those horses won 7.9% of them. We then shuffled the trouble flags within each race 20 times (so field quality and race shape stay fixed) — randomly flagged horses “won” 8.1%. The real figure sits below chance (z = -1.19); first-three finishes read the same (z = -1.02). An excused run carries no bonus at all.
We also rebuilt our sealed rating with the obvious fix — a trouble-flagged run may not count below mid-field — and re-ran 2,759 races walk-forward. Top-pick strike rate moved +0.11 points, winner-in-our-top-three -0.29 points: noise in both directions. The rating stays at v0.3; we do not ship changes the data refuses.
On one season the strongest phrases (“denied a clear run”, “held up”) showed a spark: +1.3 points of win rate on 601 cases. On three and a half seasons the same bucket holds 2,080 cases and the spark shrank to +0.7 points — about one standard error. Real signals grow with data; noise shrinks. This one shrank.
The one comment that does mean something
A veterinary line — lame, bled, distressed — is the exception, and it points the other way: those 611 horses won 5.2% next start against a 8.2% base, and made the frame 20.6% against 24.6%. A health problem carries forward; bad luck does not. It is two per cent of runners and the market sees the same reports, so it changes no rating either — but it is worth knowing when you read a form line.
Why Hong Kong kills this angle
These comments are public within days and Hong Kong hosts the most heavily studied pools in racing. Whatever a trip note is worth is in the price before the horse runs again. That is also why we publish a negative result: our product is the measurement, not the myth — sealed before the race, scored after it. Method and code: scripts/measure-trouble.ts in our repository; phrase meanings and per-phrase numbers live in the comments glossary.