Finding the criteria for systems

A bit more on software, categories and criteria

This is the fourth blog in a little series about how I build my systems. You can access the others in the link below.

---

Read more of my blogs here.

---

This blog says a bit more about the software I use, as it’s fundamental to the building the systems I use.

I stumbled across Horseracebase a few years ago and I quickly became a fan. There are a few different tools available to system builders. Horseracebase, Proform and GeeGeez are three of the most commonly used. All have their strengths and their little niggles. Some are more limited, others are more expensive. They all in their own way have a vast database of race results and key information, statistics and run comments about every horse that has run in them. From this, you can build systems around some of these categories.

Horseracebase breaks down its data into dozens of individual categories, which can be summarised into a few high level sets of questions about anything to do with:

• The race: think things the number of runners, the racecourse, the race distance, the class.

• The horse: think things like a horse's age, sex, headgear, wind ops, number of previous runs at the track, for the trainer, for the jockey, on the ground, over the trip, whether it generally leads or is held up.

• The trainer: think things like where the trainer is based, their recent performance, their performance at the track, how many horses they have racing that day etc.

• The jockey: same as with the trainer.

• Pedigree: various things to do with the stallion or dam.

There are other categories too, like the specifics of recent races, but you get the idea. The screenshot below shows the trainer and jockey categories available.

In my second blog, I broke down an example tip from another tipster into a range of criteria that could be considered more generally in a systematic way. Most of these ideas, except probably the head carriage, can be reviewed as categories in Horseracebase.

And in my most recent blog, I banged on about the importance of ideas. Implicit in that blog was my hesitancy to focus on individual trainers and jockeys as criteria, and the reasons why.

Very few of my systems are built around trainers and jockeys. The simple reason is that I usually don't know enough about why a certain trainer has done well in a certain context (and why the market might have missed it).

• Maybe a trainer loves a racecourse because the configuration is very similar to their gallops at home (e.g. see here). Great idea if so, but I don't know it to be true.

• Maybe a racecourse has a special place in a trainer's heart because the owner has dinner there each meeting and wants winners. Great! But I don't know it to be true.

• Or maybe the trainer just coincidentally sent some well handicapped horses there in the past and it's totally irrelevant to any future horses sent there.

I’m not looking to criticise those who use trainer- or jockey-based angles. Each to their own. I’m just generally not a fan. Where I have a specific trainer-based system (or some other system with criteria relating to the people involved), I like to know a bit more context beyond the data to give me some confidence that this might be a genuine goer of an idea.

I've contacted some trainers before to see if some past edge in the data has some theoretical foundation. If they give me confidence that I've found something, I'll use it again. Or it might be a little nugget that a trainer mentions in an interview triggers an idea. If that idea has good data behind it, I'll probably use it again.

Otherwise, I tend to leave the human-based (e.g. trainer or jockey) ideas alone. I don't (for instance) know if a trainer is sending their horse 100s of miles because they really fancy it to outrun its odds, or because they're visiting the in-laws nearby and the racecourse happens to be on route. Nor do I know that a trainer is particularly adept at finding the horses most likely to benefit from a tongue tie, or whether they've just got lucky by doing it before.

As a general rule, I don't feel that I know enough about the humans involved in horse racing to build systems based on their thinking. If other bettors know more than me to trust the human-based angles, good luck to them. If other bettors don't know more than me but they trust the past back fitted data enough to crack on in the future, equally good luck to them too.

They just don't really work for me.

Of the five categories above, that then leaves three:

• categories about the race;

• categories about the horse; and

• categories the pedigree.

Most of my systems revolve somewhat around these three. If you enjoy reading the form, the Racing Post, Timeform and watching the races of my system bets, you might have already picked up on some of the things I tend to focus on.

You might have noticed that most of my selections are in handicaps. Or that I pick across all three codes of racing. Or that I generally focus more on the UK than Ireland. You might also have picked up that I like risky propositions. The sort of horses that the Racing Post comment might say something like:

This horse *could* run well *if* a series of events occur.

I’m generally trying to find some of those unlikely events, and identify those that might be a bit more likely than the market might expect.

I currently have around 40-50 active systems saved in Horseracebase that produce the qualifiers I share with you each morning. Some systems produce qualifiers quite regularly (the busiest one produced 108 qualifiers in 2024) while others are much more occasional (the quietest one produced just 2 qualifiers).

If you’re interested in having a play around with Horseracebase, you can enjoy a free trial here. Or if you have an idea and you’d like to know how it’s performed in the past, feel free to get in touch and if the categories exist I’ll happily run the data for you.