Brought to you courtesy of Michael Wilding of Puntology.com
Whenever you look at a race it is almost always about numbers. There are people who paddock watch, and this information is very valuable if you have a strong knowledge of horses. The majority of us do not do this and use data to help us determine which horse we think will win the race. This data is almost always in the form of a number. Speed figures, official ratings, weight carried, finishing positions are a few examples of numbers that we use when assessing a horse race.
Most of the numbers that are used in handicapping are rating an aspect of the horse’s performance. Usually the higher the number the better the performance but sometimes the lower the number the better the performance is. It is common for people to look for horses that are good performers in certain areas that they have discovered have value and can make a profit.
What we must not forget in this process is that the numbers are just that. Numbers!
However accurate a number is, it is never going to be 100% correct. If they were then we would always select the winning horse. This means that we can look for different ways of analysing the numbers in order to find a profit. One such method is to use horse and race shapes.
Let us imagine that you have ten factors for every horse in a race and between all of these factors every aspect of the horse’s performance is rated. Every single horse in each race is going to have a different set of numbers. It is very unlikely that any horse in a race is going to have exactly the same number for each factor. In fact it is very unlikely that any horse will have exactly the same number as another horse over multiple years of racing.
If we were to take each rating number and plot them on a graph we would get a bendy line going higher when the horse has a higher rating and lower when it has a lower rating. Effectively this is the shape that the horse’s factors are taking. It is flat and simple but it allows us to visualise the numbers.
When you repeat this process for every horse in the race you will find that different horses are better in different areas and you can see at a glance which horse is better at what. In its simplest form this is a horse shape. It shows the ability of the horse before the race and we are representing it visually in a graph.
Let’s take this one step further. Every race is going to be made up of different horses with different shapes. Each race also has its own conditions which can be entry conditions, going condition, distance and all the factors that are race specific. All of this information combined gives us the shape of the race. We could of course plot this on a graph as well but we may have a lot of items to put on.
Not every horse is going to be suited to every type of race. I am sure you have heard many people say that it is important to find a horse that likes the conditions. An all weather sprinter is unlikely to win if he was put into a National Hunt race.
The conditions won’t suit him. With this knowledge it is then logical to assume that certain horse shapes are going to perform better when they are taking part in race of particular shapes. Surely this means that we don’t need to analyse the figures and look for the horses with the highest ratings but we need to look for patterns in the past where horses of a certain shape seem to win races of a certain shape with more regularity than they should.
The problem is that we don’t have enough data to do this. As we have already discussed a horse is going to have a different shape in every race and it is very unlikely that any two horses are going to have exactly the same shape. This means that effectively we only have one horse per shape and we cannot analyse anything with that! Although the race shapes are not going to be quite so severe as the horse’s, we are likely to have similar problems with them as well. This means is that in order to find out which horses are likely to perform well in which races we need to limit the amount of shapes that we have.
You can do this in many ways but one possible way is to use fuzzy logic. Fuzzy logic is about being approximate rather than precise. A speed rating of 65 is specific although it is impossible to know exactly how fast the horse ran with the information available to us. This means that we cannot be specific and there will always be a margin of error. Instead we could choose to say that a horse with a speed rating between 55 and 70 is ‘Good’. If we did the same for a number of ratings we would get a list for each horse that may look like:
Speed – Good
Distance – Average
Class – Poor
Weight – Excellent
A quick glance shows us where the horse is going to perform well and where it is going to be let down. If the race is a sprint then the speed is going to be high up on your list as needing a minimum of ‘good’ because without ‘good’ speed a horse cannot win a sprint.
We can use fuzzy logic to help us group our shapes together using the above method to change our ratings. One shape of a horse may simply require that it is ‘excellent’ in speed and ‘good’ at the distance.
Others may require all ten of the factors to be specific. It is important to remember that a horse shape does not have to be made up of good factors. A horse that has ‘bad’ speed and ‘poor’ class is also a valid shape. You are not looking for winning horses but just grouping them into different shapes. Perform the same task with the race shapes and you will find that you now have large quantities of horses that are categorised under the same shape. You can now start to look for patterns of which horses win which races. When you have found some then in the future you are looking for a winning horse shape to be racing in the necessary race shape in order for you to have a bet.
Using fuzzy logic is not the only way of grouping the shapes together. You may choose to use standard deviation. Horses with a distance rating within two standard deviations of the average may be put into one group, three standard deviations into another group etc… You are looking to reduce the number of different shapes that you have so that there is enough data for you to do a meaningful analysis on it. There are almost endless ways of doing this and although these will start you down the right path, I am sure that you will come up with some better ones than I have mentioned here.
