Lowest on Yield – 2

Good News and...

Well, I'm afraid that I've got some bad news for you.

I've just realised that the data set that I shall be using for this analysis only contains the top three rated horses.

That means that I can't analyse all the horses in the race.

That means that I need to write some code to generate a new database table that contains all the rated horses with all the detail that I show in my normal results table.

That means that I'll be able to offer a new spreadsheet (on the sidebar of the site) with Betfair prices (and some other good stuff) for you to use in your own analysis.

I'll only be able to update that spreadsheet once a month but that should be sufficiently accurate for most needs.

So, the bad news wasn't so bad, after all.

Introduction

As I mentioned in Lowest on Yield - 1 I have decided to test a number of possible strategies at the same time.

I shall be calling these strategies Lowest on Yield Strategy or LOYS and they will be numbered as LOYS1, LOYS2 etc.

The Strategies...

1). LOYS1 - Lowest on yield horse from the top three rated horses on the first page of the ratings in non-handicap races.

This is the current strategy and is a result of my observations.

2). LOYS2 - Lowest on yield horse from all horses in the race on the first page of the ratings in non-handicap races.

I have been informed that there is more value to be had by looking at all the horses in the race rather than just the top three.

3). LOYS3 - Lowest on yield horse from the top three rated horses in the race for all rated races in non-handicap races.

I have based my observations on just the first page of the ratings but there may be a better cut off point.

4). LOYS4 - Lowest on yield horse from all horses in the race for all rated races in non-handicap races.

Who know what this might reveal?

Process

In order to investigate LOYS1 and LOYS3 I created a spreadsheet. This is based upon exactly 2 years of data (to make the calculations easier for me) from (and including) the 1st of September 2013 to the 31st of August 2015.

That spreadsheet of raw data may be found at http://focusratings.com/data/LOYS.xls

I then made a copy of that spreadsheet and removed all races that weren't on the first page of the races.

I then took out all horses that weren't lowest on yield.

Finally, I conditionally formated the Results column so that the winners stand out.

That spreadsheet may be found at http://focusratings.com/data/LOYS1.xls

Now, it seems that, due to the way that the data was compiled, I am left with only 23 months worth of data.

Still, that should be enough for us.

The core data is shown on the Raw Data tab (at the bottom of the spreadsheet.)

The Analysis

Unfiltered - Before starting on the analysis, I took a look at the raw data to see what profit could be yielded by blindly backing the top three rated horse that is lowest on yield in non-handicap races on the first page of the ratings.

I'll use this as a benchmark.

So, I did the analysis. I split it up by country but there was really no need to do that.

The results are as follows and are shown on the Basic Analysis tab of the spreadsheet (at the bottom of the spreadsheet.)

Country	Races	Wins	Strike Rate	ROI
IRE	450	136	30.22%	71.29%
UK	2087	766	36.70%	91.68%
Total	2537	902	35.55%	88.06%

Now, this surprised me; I thought that this was a blindly profitable strategy - it certainly looked that way.

But, taking a positive from a negative, it certainly works better for UK races than for Irish races.

So, rather than waste any time, I cut out the Irish races and then moved on to look at the number of runners in a race.

So, the first rule is... Only use UK races.

The UK only data is shown in the UK tab (at the bottom of the spreadsheet.)

Runners - I looked at was the number of runners.

The Runner analysis wasn't as conclusive as I expected.

Having said that, it does look like races of 14 runners and under are where the sweet spot lies.

Also, 2 horse races shouldn't be used (I think that they are a joke anyway.)

So, the second rule is... Only use races with 3 to 14 runners.

The UK only 3 to 14 runner races are shown on the UK 3 to 14 tab.

Race Type

Now, the Race Type analysis showed that certain race types are more profitable than others.

But, I don't want to apply any filters at this point as I feel there should be another filter (or maybe more than one) applied first.

However, it does seem that Flat race and NH Flat races are a bit of a washout.

R4 Status

The question has to be asked...

Does R4 Status matter?

So, I ran a query to see if I could find an answer.

The logic is that we would expect the top rated horse to win more often than the 2nd rated which should win more than the 3rd rated as, that is what happens on a day to day basis.

The results are shown on the R4 status tab.

The results are inconclusive.

R4

By formatting the results column on the UK 3 to 14 tab it made it easy to see that selections with a low R4 value didn't win so often and so, I decided to test this. I had to create a new column to show rounded values of R4.

Now, the strike rate does improve in line with the R4 value but, as we know, strike rate is just vanity and it's profit that counts.

Having said that, if we are looking for a multiple/exotic bet, then a good strike rate does help.

Yield

By sorting the UK 3 to 14 tab by yield, I noticed that the strike rate varies inversely by yield.

So, I created another column and did a query on that.

However, once again, the results were inconclusive.

Conclusion

At this point I've decided to take a break and go for a walk in the forest to see if inspiration strikes (as it often does during a walk in the forest.)

One thing that is niggling me is that, from the 1971 races over the 23 month period, 1461 selections came in the first three races so...

Perhaps I'm looking at this in the wrong way; perhaps I should be thinking of another place betting strategy.

Now I know that the 3rd place doesn't count in races of less than 8 horses but a starting point of a strike rate of 74.12% seems interesting.

So, I am publishing this inconclusive and unfinished post so that you can download the spreadsheets yourself and see if you can find another angle.

If you have any ideas or suggestions, please feel free to let me know; feedback is always welcome.

Once I get back from my walk I'll take another look at a backing strategy and I'll generate a new spreadsheet to show Betfair Place Prices so that I can take a gander at a place betting strategy.

I suppose that I ought to mention that, just by filtering by the two existing rules and by profitable race types we get 676 races over a 23 month period that look profitable.

So, there might be a win backing strategy there but it'll give us only 6 selections a week.

As always...

My kindest regards