Class ratings for Stakes races

[Clocker]

Originally Posted by Tom

In his article, "The Basics of Handicapping the Turf," Rich Halvey suggested that to be a Gr1 turf race, at least a third of the horses in it have won or placed in a Gr1 race.

Do you have a link to that article? I looked for it and couldn't find it. Want to see what else he may have to say. Thanks.

[reckless]

Halvey has his own web site/blog.

Here is the link:

www.halveyonhorseracing.com/.

Thanks Tom for the heads-up.

[overthehill]

I dont think it makes a lot of sense to me. for one thing there are lots of horse who are graded stakes placed who never actually win a graded stakes. secondly there is a question of recency. a horse who won a graded stakes a year ago might not be graded stakes quality this year.

[raybo]

Originally Posted by overthehill

I dont think it makes a lot of sense to me. for one thing there are lots of horse who are graded stakes placed who never actually win a graded stakes. secondly there is a question of recency. a horse who won a graded stakes a year ago might not be graded stakes quality this year.

Nice Tom, quick and simple, and probably better than a straight "earning" based class rating! Personally I add an overall performance factor to earnings for my class ratings.

IMO, "form" trumps everything else, pretty much, so whether you analyze form first, or last, it should be used to adjust any other ratings you use for ranking or grading the horses. It does not necessarily need to be included in the original calculations of any other factor, but rather to separate closely rated horses, again IMO.

I tend to agree that horses who haven't won a particular class of race, but have been very competitive in that class, should not be significantly penalized for not winning a race in that class. Maybe the addition of a "+" to the final class rating?

[Tom]

I am flexible.
If I see a horse finished 3-nk in a Grade 3 and that is the horse that makes the one third, I may call it a 3 instead of a 4 or 5...this is all just kind of a note taking method, so if I call a race a 3 instead of a 4, so be it. What matters is I don't call it a 1 or a 2.

With HTR, I have every PP for every race for the last year and a half at my fingertips, so it is no big deal to go looking at old races, and there is a filter I can use to just see winning pace lines, so this is very quick and easy for me to do.

[raybo]

Originally Posted by Tom

I am flexible.
If I see a horse finished 3-nk in a Grade 3 and that is the horse that makes the one third, I may call it a 3 instead of a 4 or 5...this is all just kind of a note taking method, so if I call a race a 3 instead of a 4, so be it. What matters is I don't call it a 1 or a 2.

With HTR, I have every PP for every race for the last year and a half at my fingertips, so it is no big deal to go looking at old races, and there is a filter I can use to just see winning pace lines, so this is very quick and easy for me to do.

Don't know for sure, but if HTR has an export for Excel, couldn't you automate that class rating system? Use some "IF - THEN - ELSE" statements to make the decisions for you?

[classhandicapper]

Originally Posted by overthehill

I dont think it makes a lot of sense to me. for one thing there are lots of horse who are graded stakes placed who never actually win a graded stakes. secondly there is a question of recency. a horse who won a graded stakes a year ago might not be graded stakes quality this year.

These are some of the issues I ran into when trying to devise an "automated approach".

How far back can I go in time?

How many races back?

What if the horse's form has deteriorated recently?

How do you handle surface and distance changes? Suppose a horse is Grade 2 winner on dirt sprinting but today is a turf route?

I think setting a time limit and only going back a certain number of races works best if you want to automate the process. But since we are only talking about stakes races, it's not a lot of races. A manual subjective process that analyzes each horse's PPs works better and is not too demanding.

I don't assign points exactly like Tom, though that's a terrific approach.

I use a rating for each horse based on the PAR for the class of race in his PPs I am using. Then I create a weighted average of the horses in the field today. I look at the top 5 finishers in today's field and any other horse that ran competitively within the race to calculate the rating.

The system I use would work fine for every class of horse everywhere....EXCEPT that you need PARS for every class and that's borderline impossible to accumulate. So I don't bother actually creating ratings for most races. I just think in these terms.

[cj]

The more people deviate from the simple system Tom proposed, the less likely it will have any value in my opinion.

[raybo]

Originally Posted by cj

The more people deviate from the simple system Tom proposed, the less likely it will have any value in my opinion.

I agree, for what it is designed to do, calculate the class rating of fields, I think his simpler system is fine, and much better than the man-made class designations the raw data gives us. If you're attempting to create a "class performance" rating then more is needed, IMO.

[Tom]

Ray, there is an export, but getting to the point describe is beyond my pay grade. I guess it could be done, but probably not by me!

[raybo]

Originally Posted by Tom

Ray, there is an export, but getting to the point describe is beyond my pay grade. I guess it could be done, but probably not by me!

Oh come on Tom, I know better than that, you just haven't tried yet - LOL. If you can write down all the steps on paper then you can put it in Excel, with a little thought/work. I know, "work" was a shocking word for Maynard G. Krebs!

Maynard G. Krebs "Work!"

[Tom]

I don't know the steps to write down!

[raybo]

Originally Posted by Tom

I don't know the steps to write down!

Oops! Not good!! Consistency man, consistency.

[thaskalos]

Originally Posted by classhandicapper

These are some of the issues I ran into when trying to devise an "automated approach".

How far back can I go in time?

How many races back?

What if the horse's form has deteriorated recently?

How do you handle surface and distance changes? Suppose a horse is Grade 2 winner on dirt sprinting but today is a turf route?

I think setting a time limit and only going back a certain number of races works best if you want to automate the process. But since we are only talking about stakes races, it's not a lot of races. A manual subjective process that analyzes each horse's PPs works better and is not too demanding.

I don't assign points exactly like Tom, though that's a terrific approach.

I use a rating for each horse based on the PAR for the class of race in his PPs I am using. Then I create a weighted average of the horses in the field today. I look at the top 5 finishers in today's field and any other horse that ran competitively within the race to calculate the rating.

The system I use would work fine for every class of horse everywhere....EXCEPT that you need PARS for every class and that's borderline impossible to accumulate. So I don't bother actually creating ratings for most races. I just think in these terms.

IMO...the class factor loses "value" when it is closely associated with speed variables. I believe that the different handicapping factors should be contemplated SEPARATELY...not lumped together.

[classhandicapper]

Originally Posted by thaskalos

IMO...the class factor loses "value" when it is closely associated with speed variables. I believe that the different handicapping factors should be contemplated SEPARATELY...not lumped together.

I agree.

I am not suggesting that you use speed figures in the rating. I am suggesting you use PARs (where available) to help determine the class pecking order and gaps between classes. Over the long term, higher classifications tend to be run faster.

For example, the simplest pecking order that everyone understands is Grade 1 is best, then Grade 2, and then Grade 3.

But the gap between Grade 1 and Grade 2 tends to be larger than the gap between Grade 2 and Grade 3. So whatever points you assign should probably reflect that.

The gap between class levels in turf stakes tends to be narrower than on dirt. It's easier to jump out of the ALW ranks and move up in stakes on turf.

The gap between males and females tends to be narrower in Grade 1 sprints than routes.

etc...

I learned these things by making PARs and by observing the results when horses made class moves. Some single steps are easier than others.

If you wanted to extend this system down to all classes, you'd have to understand where statebred fits with open, claiming fits with ALW, Starter ALW conditions fit the others, 3yos fit with older throughout the year etc...

It's becomes a monstrous task for more than one circuit. But for stakes, it's pretty easy to put together a systematic approach that reflects field quality and current form WAY better than the official class designations without (as you suggest) using any speed figures. I do use the speed figure PAR scale though.