Yes i thought that was a neat way to use win/place percentages.

Cheers Gunny
 

Gunny, as a starting base that's not too shabby at all. I just ran that on selected meetings Jan 1 - May 31 2011, looking at the top 2 only:

Races: 4288
Wins: 1593 (37.2%)
Bets: 8575
Return: 7795.30
Loss: 9.1%

2 horses per race, every race & less than 10% loss on NSW TAB.

As a matter of interest, how does the top 2 or 3 perform on win% only.

Win % Top 2, same period:

Races: 4288
Wins: 1244 (29%)
Bets: 8572
Return: 6858.20
Loss: -20%

Thanks Raven

I used this system successfully for about 10 years with a 5% pot most years but in recent years I can't get the value I used to get although the winners keep coming at the same rate and I agree with your 9% lot result. I backed 2 horses a race but chose the longest priced ones from the top three at 3 mins before the race with a minimum $3.50 price, and did I get some great results.

You must have very good resources to produce these stats so quickly and I appreciated reading them. Actually, as I was trying to keep things simple, I just doubled the win% but in fact I used a factor of 2.175 as I found the average place% was 2.175 times the average win% for about 3000 horses.

I am currently trying variations with this idea in attempt to get better value.

I'm interested in this thread
 

If you checked my win/place% ratings method you will find in BR at least some satisfactory results for the past two weeks. I can improve on them by using the following formula to covert win% to an equivalent place%. ( My earlier approach was a simplified version of this.)

EqivalentPlace% = (Win%-22.8)/12.3*15.2+49.5

This is easy to do in a spreadsheet column, then just find MAX(EquivalentPlace%, Place%) and this is your rating for the horse. I find the top three need to be considered.

I doubt that you will find a free set of ratings that will give results any better than this will. This is a good starting point and some of you with more resources and ability than me might come up with a great filter to produce a pot.

I do not have extensive resources and my calculations for the means and standard deviations of win% and place% used in this formula are based on calculations from a few thousand horses done over several years. Anyone want to provide better figures?

Very interesting Gunny, so what your saying is for example a horse with a 50% win S/R would end up having a figure of 83% to compare against its place S/R.

That's the value I would get. It's the standard statistical way for comparing different sets of results by converting one set of values to the same mean and standard deviation as the other set. It could be done in reverse and place% converted to an equivalent win%. I just think this is a good basic rating method.

Mind you I have tried this is practice but did not give it a fair trial. Previously I used place% only and in my trial I was not doing too well and then too rub insult ino my wounds in an a MR race the third place% horse won at $27 and I missed it. I then reverted to my place% approach. This was fortunate because in the very next MR race third place% won and paid $73 and I collected and did not give the standardised approach any more consideration. But of course place% by itself has its problems, more so these days as there are not too may good priced winners.

G'day I'm new. I posted the following question in the wrong thread. I think I'm at the right thread now. The Win%/Place% idea sounds very appealing. Are there followup stats kept since June 2011?

gunny, I have had a look at your ratings approach recently and I have been impressed with the results. It seems so simple yet it's very effective :)