Micro-Angles A Quick Look At How They Work Part III
Last week we covered what a profitable system looked like nine years ago! However, upon implementing it, it failed to make the grade.
I started to use the system and within the first three months, I had lost 50 points 🙁
So, what went wrong?
The system I had built was based on logic (or so I thought!) and worked on paper. As it happens and part of my learning curve was that I'd actually “back fitted” the whole system and it was doomed to fail from the outset.
First of all, it had too many rules. It's quite possible to make a profitable system from only one or two rules, as long as they are based on logic.
Second of all, I had simply added a rule that had worked in “hindsight”.
Let's look at the rules again and see how this worked (or in this case, went wrong!)
- Top weights in handicaps
- Jockey Claimers
- Left/right handed tracks
- Horse fitness i.e had run in the last 31 days
Looking at the logic of the systems first two rules is as follows –
A horse receives top weight in a handicap, because it's technically the best horse in the race, having received more weight than it's competitors, based on previous form.
Assuming that we have the best horse in the race (a top weight) backing said horse that has a jockey that can “lose a few lbs” would put that horse, back on a competitive weight?
It all makes sense, so far, right?
No matter what the data, states, the initial two rules make sense and are based on logic.
However, the only problem with a claiming jockey is that he/she has less experience when riding a superior horse!
In hindsight, my next rule should have been to look at how the claiming jockey had been riding in recent weeks.
Alas, that's not what I did and thought it would be better to focus on left/right-handed tracks. The reason I did this, was that the data worked and made a profit.
Of course, this was back-fitting!
I simply chose a rule that worked previously. It wasn't a rule that would work ongoing, because generally (a more experienced jockey, might take into account the course angles) a claiming jockey won't consider the turn of a track.
Having used the left/right-handed track rule, the fitness one was completely void!
It took me nearly 18 months to realise what I'd done, having tried and tried again, using different staking and only backing favourites, etc. They all failed and it was only when I discovered “bootstrapping” that I realised how it was possible to make a profitable system, without repeating the previous mistakes!
Stay tuned next week, as I go through how I actually made a model that used a sample set of results and turned them into profitable result without back-fitting 🙂
Have a great week,