ABOUT ME
I’m a racing fan who handicaps on the side, and by day I work in e-commerce. I’ve always felt horse racing and e-commerce are the same kind of game: both are about causality, probabilities, and finding a repeatable edge.
Right now I’m using AI tools to build a faster, more repeatable handicapping workflow. At this stage I’m mainly studying past races, writing trip notes, and sharing what I learn. Toward the end of the season, if my predictions keep lining up with the actual result, I may also publish race-day analysis.
I’m also open to collaborating with other racing media. If you’d like to work together, feel free to email aa@aa.com.
Hope you enjoy the content here.
ABOUT THIS BLOG
On any given race day in Hong Kong, walk a lap around the track and you’ll see fans hunched over the form and racing pages, scribbling notes like they’re cramming for an exam. And it makes sense—the information density is huge: pedigree, same-distance results, barrier trials, trainer–jockey combos, recent speed, and comments that hint at how a horse actually ran.
You can also find plenty of YouTube videos where hosts break down race replays—what happened on the backstretch, whether a horse got a clean trip or ran into traffic, and how much finish it really showed in the stretch run.
The problem is time: do you really need hours of research to play this game well?
This blog is my attempt to answer that with a practical system. I’m building a quick-pick handicapping process and stress-testing it by simulating bets on historical races. The goal is simple: confirm whether the approach produces an edge over a meaningful sample size—rather than a short run of good luck.
ABOUT THE SYSTEM
With the help of AI tools (I use Julius.ai), I only need 10–15 minutes to produce the quick-pick horse selection table below.
All factors in the table are calculated by an algorithm.
At this stage, I have no plan to offer paid information, so there’s naturally no incentive for me to interfere with the algorithm or inflate the simulated forecast hit rate just to prove anything.
If this is your first time visiting the site, spend a few minutes reading the notes below—it’ll make it much easier to understand the selection logic used in each race case study.
 | The strongest horses in the advantage analysis (a race usually has no more than two). Strong horses often treat the race as a prime opportunity and show higher intent, so they fall within the banker / key-horse range. |
 | Horses that previously ran close to the strongest horse in the advantage analysis in the same race (generally within 3 lengths). This type may not have perfectly consistent results, but if it performs to its ability it can contend for the win. Sometimes it also comes with unexpectedly attractive odds, so it can be used as a banker / key horse as well. |
 | Horses that previously faced the weaker horses in the advantage analysis (in Class 4, there can be up to three) and at least ran close to them. Since these horses have shown they can beat at least one opponent, they typically have reasonable intent and are often considered first-choice supporting selections. |
 | Mid-strength horses in the advantage analysis (in Class 4, up to three). On paper these horses often look “competitive” or “thereabouts,” but without a clear edge. They may also be tactically compromised versus stronger rivals and not be ridden in the most optimal way. It’s not recommended to include too many Mid horses among your supporting selections. |
 | Horses that previously raced against the mid-strength horses in the advantage analysis (in Class 4, up to three) but either dead-heated with them or were beaten by them. If a horse can’t reliably beat mid-tier runners, it’s naturally even harder for it to beat stronger rivals in the same field. Unless the race is extremely thin and there are too few options, it’s not recommended as a supporting selection. |
 | Lower-strength horses in the advantage analysis (in Class 4, up to three). Unless there are simply too few viable options, these are not recommended as supporting selections. |
 | Horses that have a setup advantage in the scenario analysis due to running style and/or post position. If two horses are strong enough to be your banker, this tag can help decide which one is more likely to get a favorable race shape. |
 | Horses whose running style/post position clashes with the rest of the field in the scenario analysis. They generally won’t be used as a banker, and it’s not recommended to include too many Style- horses among your supporting selections. |
 | Besides advantage ratings and race shape, you should also factor in other items to adjust a horse’s tier. Negatives that can downgrade a horse’s tier include: 1) A top-performing jockey (e.g., Zac Purton) taking off the mount 2) A stable change that requires the horse to re-find its optimal running style 3) An equipment change after a placed effort 4） A significant change in carried weight Positives that can upgrade a horse’s tier include: The horse was checked/blocked in the straight last start and couldn’t fully run out its race. |
HOW TO FORM BET SLIPS WITH THE SYSTEM:
Regardless of the analysis, the simulation will always output a fixed structure for tracking performance:
one banker (B),
two primary supporting picks (S1),
and two secondary supporting picks (S2).
It also provides suggested bet structures for Quinella and Quinella Place (Place Q) so results can be quantified consistently.