PDA

View Full Version : Question about Quirin


Actor
07-16-2016, 06:13 PM
I'm reading Winning At the Races by William L. Quirin. At the bottom of page 2 he talks about ..

NH: Number of horses.
NW: Number of winners.
WPCT: Winning percentage of the horses under study."


Then in the middle of page 3 he writes ..

"... the important statistic %W will be used frequently, especially when studying handicapping characteristics common to several horses in a race. When just one horse in a field can have a certain characteristic, such as being the favorite, %W will be the same as WPCT."

What's giving me trouble is figuring out how these are calculated. When only one horse can have the characteristic it's easy ..

WPCT = NW / NH
%W = WPCT.

In these cases NH is the same as the number of races. But how are the two calculated when more than one horse in the field can have the characteristic, such as horses with 8 speed points?

I'm thinking that..

WPCT = NW / number or races
%W = NW / NH

or maybe it's the reverse?

I know that this should be terribly simple but I'm finding that it's not. Thanks for any help.

Overlay
07-16-2016, 07:13 PM
For a characteristic where it's possible to have more than one horse in a race possessing the characteristic, WPCT would be the number of winners in the sample possessing the characteristic divided by the total number of horses in the sample possessing the characteristic.

% winners would be the number of winners of the sample races that had the characteristic, divided by the total number of races in the sample.

For example (just pulling hypothetical numbers out of the air), if you have a fifty-race sample where eighty horses out of all the runners in those races had eight speed points, and twenty of the races were won by a horse with eight speed points, WPCT would be 25% (20/80). % winners would be 40% (20/50).

Overlay
07-16-2016, 07:30 PM
Just as an added note:

Calculating the impact value for the factor (as Quirin calculated it) would then require also knowing the total number of horses in the sample races, with the impact value being the percentage of the winners of the sample races that possessed the factor divided by the percentage of all horses in the sample races that possessed the factor.

(Some sources modify the impact value to also reflect the odds of the horses in the sample, to determine if horses possessing a particular factor ran better than expected given their odds. A discussion of that approach can be found at the following link:

http://www.netcapper.com/TrackTractsArchive/TT010223.htm.)

Actor
07-20-2016, 09:09 PM
A little bit overdue, but thanks for the help! :ThmbUp: