Speed figures and speed ratings work better in Hong Kong than in any other major racing jurisdiction because almost every variable that corrupts them elsewhere has been removed. There's around 1,200 horses, two racetracks (soon to be three with Conghua!), one training centre, no domestic breeding industry, and a rating-driven class system that keeps the same horses meeting each other with weight-based handicaps. When conditions are this controlled, a difference in adjusted time is far more likely to be a real difference in ability.
When I built the Hong Kong Speed Report, one of the things that frustrated me about American racing was the sheer amount of variables you could not account for. Inconsistent stewards, mid-tier racetracks with inconsistent timing, unpublished workouts — the list goes on. Hong Kong is the opposite, and that's exactly why the numbers hold up here.
What makes Hong Kong different?
Most racing jurisdictions are — well, they're sprawling. Britain has around sixty racecourses with wildly different topographies. My homeland, the good ol' US of A, spreads a huge horse population across dozens of tracks on two surfaces, with horses shipping constantly between them. In that environment, a time recorded at one track on one day is a noisy signal.
Hong Kong is the opposite. It is the most tightly controlled racing environment in the world, and the numbers behind that claim are worth stating plainly.
| Feature | Hong Kong | Why it matters for figures |
|---|---|---|
| Horses in training | ~1,200 | A closed pool; the same horses meet repeatedly, so form lines interlock |
| Racetracks | 2 (Sha Tin, Happy Valley) | Only two sets of track variables to model, not sixty |
| Training base | Effectively one (Sha Tin) | No shipping variance; every horse prepares under similar conditions |
| Breeding industry | None — all horses imported | No unraced homebreds appearing with unknown ability |
| Geldings | ~96% of the population | Far less of the temperament and seasonal variance that entires bring |
| Class system | Rating-driven, narrow bands | Fields are consistently matched; figures aren't distorted by mismatches |
Population figures reflect widely reported HKJC numbers for horses in training (approximately 1,213 at the time of the cited reporting, roughly 96% geldings). Sources: South China Morning Post; Thoroughbred Daily News. Verify against current HKJC figures before relying on exact counts.
Why does a closed population matter so much?
Because speed figures are ultimately a comparison exercise, and comparisons need overlap.
With roughly 1,200 horses racing at two tracks, any given runner has usually met — or met horses that met! — most of the field it faces today. Think about how many times Ka Ying Rising has faced, and bested, Raging Blizzard. Or how many times Telecom Power has lost in Class 5 (sorry, Telecom Power).
Hong Kong also has no domestic breeding industry — every horse is imported. That means no lightly raced homebreds turning up with an unknown ceiling. Almost everything in the field has a racing record you can measure.
If I haven't convinced you of just how cool all this is, watch a few races one night for yourself. I think you'll see what I'm talking about, and why the Hong Kong Speed Report can take your game to the next level.
Doesn't the handicap system already level the field?
It tries to — and that's exactly why measurement matters.
The HKJC's handicappers adjust each horse's official rating after every run, and those ratings dictate class and weight. The stated aim is competitive, bunched-up fields. And guess what? It works! Hong Kong racing is famously tight — which is exactly why a horse with the top HKSR late kick rating can make such a difference.
But a handicapper is making a judgement about merit, and hey — sometimes those judgements get it wrong. Ratings are revised after the fact, in increments that aren't fully published, and a horse improving quickly can spend several runs racing off a mark that no longer reflects what it is. An independent speed figure like HKSR's notices that before the official rating does.
Put simply: the official rating tells you what class a horse is permitted to run in. The speed figure tells you how fast it actually is. When those two disagree, that's a betting opportunity.
Why does pace matter more here than raw speed?
Happy Valley is a tight circuit hemmed in by the city. Sha Tin is a far more spacious track. On both, position and tempo do an enormous amount of work in deciding outcomes — often more than raw ability in the tightly matched lower classes.
That's why a figure without pace context is only half the picture. A horse that ran a modest final time while grinding into a hot early pace may have produced a better effort than a faster-timed rival who was left alone in front. Splitting a race into early, middle, late and late-kick segments is how you see the difference.
It also means the same horse can be a strong bet in one race shape and unplayable in another. Identifying which of today's runners will be advantaged by the projected tempo is a separate skill from identifying which is fastest, and both are needed.
What about the betting pools?
Hong Kong's pools are among the deepest in world racing, which cuts both ways.
Deep pools mean an efficient win market. The favourite is usually the right favourite, and there is little free money in straight win betting. But depth also means the exotic pools — quinellas, exactas, trios, and the multi-race pools — are large enough to absorb serious money without collapsing the odds. That's where a structural read on pace and figures pays, because it lets you construct combinations the crowd hasn't priced properly.
So why have you made the Hong Kong Speed Report public instead of winning millions of dollars yourself?
Fair question. This project originally started on several large Excel files which routinely crashed, so I upgraded to tracking everything in an Oracle SQL database, which turned into running a few Python scripts to generate some basic past performance data. The goal was to find a big edge for me and my friends who had become obsessed with Hong Kong racing.
In the end, I realised that data and algorithms alone weren't enough to find a guaranteed angle — but that the advanced metrics of the HKSR combined with basic analysis, handicapping and punting was enough to make me some money.
Everyone associated with the HKSR is vehemently anti-CAW (Computer Assisted Wagering). We're for the little guy, because we are the little guy. Pick up a monthly subscription and join us on the HKSR Discord to talk with us.
So what's the practical takeaway?
Hong Kong rewards analytical work more reliably than jurisdictions where the noise swamps the signal. The controlled environment means:
- An adjusted time is genuinely comparable to another adjusted time
- Form cycles are visible because horses race regularly against a stable population
- Handicapper lag creates recurring, findable mispricings
- Pace projections are tractable, because there are only two track configurations to learn
None of that makes it easy. The market is sharp and the crowd is sophisticated. But it does mean the work actually holds — which is more than can be said for most places you might point a stopwatch.
See it applied to a real card
The free sample reports show speed figures, four-segment pace ratings and the pace forecaster applied to an actual Hong Kong meeting. Full card, unedited. Subscribers get every card at Sha Tin and Happy Valley, plus the private Discord.