Calling Compound Magic

CM uses the metric system , i think he did post a table befoire stating his opinion of the % chance of horses with different kg ratings beating each other. Can't recall what thread that was on

I have a similar system , i use lb like the Official Ratings/Timeform etc.

0.5 , 1.0 etc is the difference in rating of the 2 horses

0.5 (neck ) 1.92 v 2.087
1.0 (1/2 length) 1.84 v 2.19
1.5 ( 3/4 length) 1.80 v 2.25
2.0 ( 1 length) 1.74 v 2.351
2.5
3.0
3.5
4.0 (2 lengths) 1.50 v 3.0
4.5
5.0
5.5
6.0 ( 3 engths) 1.34 v 3.94
6.5
7.0
7.5
8.0 ( 4 lengths) 1.23 v 5.348
8.5
9.0
9.5
10.0 ( 5 lengths) 1.16 v 7.25

Those prices have not been trialled over an extensive database though! I have only recently used these to price up horse races , so how the ratings convert into prices may need to be adjusted .

Of course i have prices for all the ratings differences between a diffeence of over 40lb

I don't take results literally either . So i don't consider Sea The Stars 4.0 lb superior to Youmzain because he beat him 2 lengths .I would consider both how far Youmzain was behind at the point the race was effectively over, and how much STS had in hand .So i would rate him 12lb superior so in a match it would be ...............

STS 1.12
Youmzain 9.33

To calculate the odds of a race with multiple runners you add the Odds to 1 so a horse rated 130 vs two other rated 128 would be 0.74 + 0.74 = 1.48/1 , or 2.48

The other two would be 1.0 + 1.351 = 2.351/1 or 3.35

Thanks guys, very interesting.

My ratings are pretty good at highlighting value horses but I have no way of equating them to price directly via this kind of method. For example I may have the following in a race:

58.8
53.7
50.6
48.6
48.2
47.0
46.1
31.4
23.2

I compare ratings to prices here and the value tends to be obvious, but I would like to be able to more accurately measure the 'value' difference in price terms of the say the 8.2 gap between 1st and 3rd on this list. I think the only way would be to measure the relativity over a large series of events, and Compound you have confirmed my thinking that you must have validated your ratings by doing so.

Thanks again - I have much to consider.

Start from scratch with each race Tobermory, so 58 is my top rated for the race

like though DFC you would end up with the same prices on any 9 runner race

The fav would be 20% if he was marginally better than the 2nd , but he would still be 20% if he was different class

Also DFC tobermory's point is right with the ranking idea.

I am sure the key lies in the precise value of a 1 point rating difference. A top rated that is 3 points better than second rated has less chance of beating it than a top rated which is 4+ points clear. I thought Compound's calculation was what I was looking for, but alas no as I have no link to lbs or lengths via my ratings.

As a starting point Mudflaps you need to decide what your ratings mean in relation to distance

Would a horse rated 1pt superior be 1 length superior? or half a length?

Apologies for error in my haste to post. Odds calculations should read as follows.


[pre]136 2.91/1.00 = odds of 2.91
134 2.91/.80 = odds of 3.64
133 2.91/.67 = odds of 4.34
129 2.91/.33 = odds of 8.82
125 2.91/.11 = odds of 26.50

Total 2.91[/pre]

Thanks Top 2 I think I need to test that out

I found these formulas in an old spreadsheet of mine for converting distance beaten in lengths to a poundage figure. They may be of interest to someone reading this thread.

Flat

(lengths beaten * 16) / (race distance in furlongs)

e.g. Beaten 2.5 lengths over 7 furlongs

= (2.5 * 16) / 7 = 5.71 pounds

Jumps

(lengths beaten * 19) / (race distance in furlongs)

e.g. Beaten 2.5 lengths over 20 furlongs

= (2.5 * 19) / 20 = 2.38 pounds

The formulas of course do not take the going in to account but thinking about it I suppose you could divide the pounds beaten figure by a going allowance figure such as that used by speed figure compilers to give a more useful number. Just a thought.

Yes thanks again Top2, very helpful stuff.