Thursday, 1 July 2010

Some analysis

Good evening,

I've spent the last hour or two totting up all the prices backed at, and what they returned at SP, over the last three months. This is mainly to see whether anything may come to light as for the poor results in June.

Here are the findings:


83 selections

Backed at 567.5/1
SP at 501.8/1

So on average:

Backed at 6.837/1
SP at 6.046/1


Early price 13.083% higher than SP on average with each bet


104 selections

Backed at 782.5/1
SP at 720.35

So on average:

Backed at 7.524/1
SP at 6.926/1


Early price 8.627%% higher than SP on average with each bet


83 selections

Backed at 568./1
SP at 499.7/1

So on average:
Backed at 6.844/1
SP at 6.02/1


Early price 13.688% higher than SP on average with each bet

So basically, June seemed to do find in terms of finding "value" in relation to the two months prior.

Now bearing in mind June returned more than a 50pt loss, but May and April combined to return around 90pts profit (60 + 30 respectively).. the above figures suggest that the value was still being carved out on a consistent basis, and simply the horses didn't keep their side of the bargain (!). There were a very large number of seconds, with quite a few trading very short in running. There were also quite a few very poor performances too of course.

Anyway all things considered, June was a very poor month results wise, but in terms of the overall strategy of beating the starting price by getting the early value, it held up perfectly well.

Now just for a few of those winners!


  1. Have you thought of segmenting your bets in other ways to analyse whether you get a better/worse ROI depending on the type of race. I would think it would be quite interesting to split by NH/Flat, Afternoon/Evening, and grade of race - say class 3 and above versus Class 4 and below.

  2. Thanks anon, I agree, it would be very interesting to pull apart results more closely. This is something we'll be looking at in the future, although there can be a danger of reducing results down to relatively small sample sizes and therefore display potentially rogue results which may send us down a path which may not help in the future.
    All the best,