Horse Race Ratings… How Many Is Enough

There is no such thing as analysing a horse race without using horse race ratings. Whatever paper or website you open that horse racing information there will be ratings.

The reason for that is…

In order to assess something we need to make it tangible. Although we think we’re given a lot of information about horse racing, in fact we’re given very little information.

We know information like how long since it last ran, who it’s trainer is, the jockey riding, how much weight is being carried, what races it has won/lost in the past etc…

But this isn’t, what I call, solid information.

Solid information would be to do with scientific information on the physical ability of the horse. If we had that information we would be able to predict, with a very high accuracy, which of the horses were most likely to win the race.

The information we do have is more like guidelines that we can use to make an assessment. The problem being… it’s provided to us in a format that makes it almost impossible to compare one horse with another in the same race. Or compare one horse’s past performance in a race to another horse’s past performance in a completely different race. The race conditions will be completely different which means that the information, while put under the same name, will mean completely different things.

Let me give you an example…

We have two runners, A and B, and A came 1st in his last race and B came 10th in his last race.

Your immediate assumption would be that A is better than B.

But if I then told you that A was racing in a Class 5 race and B was racing in a Class 1 race, would you still have the same opinion?

If I then told you that A was racing on good ground and B was racing on heavy ground, how would that change your opinion?

As you can see we need to turn this information into something tangible that allows us to easily compare horses with each other across different areas of their performance and in different conditions.

And we do this by…

Using Ratings

Ratings allow us to put a numerical figure to a specific part of a horse’s performance. Once we have a rating we can then adjust this number up or down for whatever reasons we want.

At it’s simplest we can say that:

We want to decrease a horse’s rating if it was in an easy race and increase it if it was in a difficult race.

Doing this we will close the gap between a horse racing in an easy race and a horse racing in a difficult race so that the numbers can be compared directly.

In fact all modifications we make to a rating are doing exactly this, it’s just how we define what’s difficult and easy that changes.

A higher class may be considered more difficult so the rating goes up a bit, but good ground may be considered a bit easier so the rating decreases a bit etc…

The definition of what is difficult and easy is made by the creator of the ratings and it can contain lots of different reasons to increase and decrease a horse’s rating by different amounts.

Once we’ve done this, the question becomes…

How Many Horse Ratings Should We Use?

This is personal. There are some people who say that they only need five ratings and others who use twenty or more.

My answer is… we want lots and not very many.

Let me explain…

Our software, The Racing Dossier, contains hundreds of horse race ratings and I believe that you need that many.

However… I also say that you should be able to build a profitable strategy with 10 or less ratings.

In fact, generally speaking, the less ratings you use then the stronger the strategy.


You MUST use the RIGHT ratings.

And to make sure you have the right rating for the set of conditions you’re looking at the race in, you need to have a big choice.

Because all ratings are not made equal. The same as races.

You could be using a rating that measures of the form class of a horse and discover that rating is completely ineffective in the race conditions you’re focusing on. But if you measure form class in a different way you suddenly find that it’s incredibly powerful in the race conditions you’re focusing on.

It’s because the information we’re given isn’t tangible that allows us to convert it into ratings in a variety of different ways. And we want access to ratings created from as many different ways as possible so that we can find the ones that work in the race conditions we’re focusing on.

What To Take Away…

If you use ratings and are struggling to make a profit then start by considering how those ratings are made. Ask the creator of the ratings if you’re unsure. This will allow you to see exactly what the rating is measuring and when you know that you’ll know if it’s likely to be suitable for the race conditions you’re using it on.

Get access to as many ratings as possible. Of course I believe that the Racing Dossier is the best ratings tool around, but if you have another preference then that’s fine as long as it works for you.

Spend some time trialing out the ratings to find which are the best measurements for the race conditions you’re focusing on.

Choose just five to begin with and run some paper tests. Then add in one more and see if it makes a difference, if it does keep it. Remove one and see if that makes a difference, if it does leave it out, if it doesn’t put it back in.

Repeat this process and be prepared to spend some time investigating the best way to use your ratings before you start betting. All the important work is done before you ever place a bet and like a house, if you make solid foundations here then you shouldn’t have any problems further down the line.

Michael Wilding

Michael started the Race Advisor in 2009 to help bettors become long-term profitable. After writing hundreds of articles I started to build software that contained my personal ratings. The Race Advisor has more factors for UK horse racing than any other site, and we pride ourselves on creating tools and strategies that are unique, and allow you to make a long-term profit without the need for tipsters. You can also check out my personal blog or my personal Instagram account.


  1. Hi Michael

    That all makes sense but I wonder if you could advise on this.

    Let’s say I consider a large number of ratings with a view to distilling it down to ten and I’m looking at one type of race, for now let’s say H’cap sprints. Let’s assume the full range of each rating is 1 – 20 (I know it’s not)

    Taking the first rating, I look at the range of past winners and find that all winners of the past 200 races have a rating between say 5 and 10.

    With the second rating, all winners of the same 200 races have a rating between 2 and 19.

    And so on…….

    Two questions:

    1) Disregarding implied probability by SP, which of the two ratings would you use (if any) and why?
    2) How would you apply implied probability to the data above?

    Huge thanks in advance.

    1. Great question Keith. The first thing I would do is run a statistical test to make sure the two ratings weren’t measuring the same thing. Then I would use impact values to determine which rating has the highest impact in predicting winners in these race and consider this as the strongest one of the two and then weight this one more as more important before combining both the ratings together. You can use statistical techniques to do this but I’ve also found it very effective to use simpler approaches. Once you’ve combined the ratings you’ve effectively got an odds line, albeit unlikely to be an accurate one. I would then use the market odds implied probability to adjust our ratings up/down to bring them closer into line with the odds and then base my bets on odds available at this point.

        1. Sorry – should have been more specific.

          Using my two example ratings:

          Let’s say 200 races each with 10 runners (unlikely I know).
          Let’s also assume there is an equal number of horses rated 5-10 in each race (3 in this scenario).

          Rating 1:
          All runners (2,000) rated from 1-20
          All winners (200) rated 5-10
          Runners rated 5-10 let’s say 600

          Rating :
          All runners (2,000) rated from 1-20
          All winners (200) rated 2-19
          Runners rated 2-19 let’s say 1,800

          So my simple deductions are:

          Rating 1:
          Probability of a horse rated 5-10 winning is 600/2000 = 0.30
          Actual winning strike rate of horses rated 5-10 is 200/200 = 1.0
          So Actual/Expected is 1/0.3 or 3.33.
          Therefore this group is 3.33 times more likely to win than expected.
          Is that what you call ‘impact value’?

          Rating 2:
          Probability of a horse rated 5-10 winning is 1800/2000 = 0.90
          Actual winning strike rate of horses rated 5-10 is 200/200 = 1.0
          So Actual/Expected is 1/0.9 or 1.11.
          Therefore this group is 1.11 times more likely to win than expected.

          If the above is true, then I might deduce that Rating 1 is 3.33/1.11 times more significant than rating 2. Would you agree?

          If this is in fact the impact value, at what point do you consider it to be significant? Positively at 1.25? Negatively at 0.75?

          Hope that makes sense – I’m no stats expert (clearly)

          1. The calculations aren’t quite right. Here’s an article showing how to calculate:


            However if you want to make it more accurate them you should use this approach:


            Assuming the right calculations then your summary:

            “If the above is true, then I might deduce that Rating 1 is 3.33/1.11 times more significant than rating 2. Would you agree?”

            Yes that would be a fair assumption and can be used as a base to assign importance weights to your factors.

            If using PIV (second method) then I would say 1.10 and 0.50 are significant. We need more on the negative to make it significant than on the positive.

Back to top button