# Testing Your Factors To Find The Best (in the same way as the pros)

A couple of weeks ago I wrote a post about how to choose factors that predict winners. If you haven’t had a chance to read that then you can here.

One of the ways that you can determine the most important factors to use for a set of race conditions is by using a chi score.

The question is…

How do you use a chi score?

That’s what this post is going to answer!

In order to calculate this I recommend that you use some kind of spreadsheet program to do your calculations in as most of them contain a formula that will allow you to do the calculation without having to understand the complex maths behind it.

I have always been a fan of Excel for spreadsheets, but recently I’ve been using Google Sheets more and more which is a free online spreadsheet software provided by Google.

There are still some things that it’s not great at, but you should check it out as a possible free alternative to Excel.

In order to calculate the chi score you’re going to need a list of the selections and their odds of winning and their finish position.

**The first step is to…**

*Calculate the probability of each selections chance of winning*

This is an easy calculation to do:

*1 divided by the decimal odds*

I always prefer to use decimal odds because this makes it much easier to perform calculations in spreadsheets.

Perform this sum for every selection and you will get a figure for each selection between 0 and 1. This is the probability of the horse winning the race based on its odds.

You can use SP odds or Betfair SP odds, it doesn’t matter which as both are accurate representations of the horses true chance of winning the race.

**Once you have done this we move on to… step 2.**

*Calculating the average probability of your selections winning a race*

Again you can do this very easily in your spreadsheet. If your probabilities were in column A then you would choose a cell and type:

*=average(A:A)*

The formula above should work in pretty much every spreadsheet program out there and will give you the average probability of your selection winning. If your probabilities are in column B then you would change the *A:A* part to *B:B*.

**With this done you can move on to… step 3.**

Step three is calculating the average odds that we’ve been betting at and this is simply done by taking the average probability you calculated in step 2 and doing:

*1 divided by the average probability*

The result of this will give you the average odds, decimal format, that you’ve been betting at.

So far you have discovered:

- Probability of each selection winning the race
- Average probability of your selections winning
- Average odds you’ve been betting at

This is all good information to have anyway and something you should add to your record keeping if you haven’t already got it.

**Step 4 is… **

Where you discover how many of your selections would have won if you’d been betting random horses at your average odds.

To do this you…

*Take the number of selections and multiply it by your average probability*

The number that you get as a result of this is the number of winners you could have *expected* to have had if you’d been randomly betting all horses at your average odds.

Subtract this number from the number of selections you have to get the number of *expected *losers as well.

**Finding the actual number of winners and losers is step 5.**

Once you’ve found the actual number of winners and losers that your selections have had you now have all the information you need to be able to get the chi score.

The final step is performing the chi score calculation. In Excel this is simply done by using the formula:

*=CHITEST(Actual Winners : Actual Losers, Expected Winners, Expected Losers)*

You can type that into any cell, replacing the Actual Winners, Actual Losers, Expected Winners and Expected Losers with the numbers that you’ve discovered through the steps above.

The result of this formula will always be between 0 and 1.

The closer the results are to 1 then more likely the results are due to chance, the closer they are to 0 the less chance they are to chance.

That means that… *you want the score to be as close to 0 as possible*.

In terms of choosing factors that predict winners you want to choose the factors whose chi scores are closest to 0.

Below you can download an example CSV file that contains racing data for a speed factor since 1st July. The factor is a ranking factor so the strongest horse is 1, second strongest is 2 etc…

You can use this file to practice the process above and if you’ve got any questions then please leave me a comment.

hi michael, thanks for the article, seems very interesting, i cant find where to download the spreadsheet.

thanks

george

Thanks for letting me know George, please refresh the page and the button is now there.

where is the link for the example CSV file?????

Thanks for letting me know Ray, it is there now. Please refresh the page.

Thanks Michael – I can understand that nearer o is better than 0.5 but what figure would identify a completely useless rating – 6 .7 0.8 or higher?

Good question Peter. There isn’t really a fixed answer, think of 0 as 0% likely to be to chance and 1 as 100% likely to be chance.

Your decision of which to use should be based on what the different factor scores are. If most of your factors are 0 to 0.01 then you probably wouldn’t want to go higher, but if most are between 0 and 0.1 that gives you the scale you should use.

Personally I try to stick to factors that are 0.05 or less if I can.