TRC Global Horse Rankings

[classhandicapper]

I'm not advocating for these ratings or rankings, but I thought this was a very good discussion of some of the issues handicappers face when trying to create a performance, class, or ranking system that's probably equally applicable to a speed figure approach.

https://www.thoroughbredracing.com/a...st-racehorses/

[MJC922]

"Notice we do not use information from outside these races, which would be an advantage in accuracy but would depart from our ethos as a ranking system, rather than a prediction system."

A strange comment. What then are they optimizing on if not the predictive, year end awards? lol

[classhandicapper]

Originally Posted by MJC922

"Notice we do not use information from outside these races, which would be an advantage in accuracy but would depart from our ethos as a ranking system, rather than a prediction system."

A strange comment. What then are they optimizing on if not the predictive, year end awards? lol

I found that a little strange too because I'm not sure what kind of information they are referring to, but it could be something like recent workouts, days between starts, surface and distance preferences, equipment changes, whether the horses is likely to improve or decline and things like that.

Those things would help you predict how the horse is going to run in the future, but they aren't part of the past performances they are evaluating.

[MJC922]

Originally Posted by classhandicapper

I found that a little strange too because I'm not sure what kind of information they are referring to, but it could be something like recent workouts, days between starts, surface and distance preferences, equipment changes, whether the horses is likely to improve or decline and things like that.

Those things would help you predict how the horse is going to run in the future, but they aren't part of the past performances they are evaluating.

I read into it as being any non-graded stakes races on the horse's record but I may have misread it. To me every race is relevant to the classification process and if someone is shunning races on a horse's record that have tested to improve accuracy there's something else being left unsaid here, they can't handle all of the data processing or someone out there already has something which has tested better etc. Anyway I probably shouldn't be too harsh with folks who do try to make an effort. I know a lot of man hours are involved.

[classhandicapper]

The other thing I found interesting was that they basically called all ratings subjective estimates. That's something I've been harping on for years.

They also said that using multiple weighted ratings gives you a more accurate picture of the horse than if you select a single race, like last race, best of last 5, best of of last 2, recent race most similar to today's conditions etc..

Unfortunately, the overall public knows that and builds it into the odds fairly accurately, but many handicappers focus all their attention on a single race they think is most representative.

The biggest problem is trying to automate a multi race approach.

The math and complexities are over my head.

I may subjectively want to toss a certain race because of a bad trip, bias, wrong surface, wrong distance, or want to change the typical weighting for let's say 2yos that may develop rapidly vs. 5yos that won't, but trying to code for all those condition and that subjective analysis is kind of a nightmare and outside what I want to put into any project.

So I start my thinking from a fixed weighting for each race based on "the data I have tested" and subjectively tweak my thinking based on the individual horses and situations the best I can.

When I look back on my youth, I wish I had gotten more involved in statistics beyond a handful of classes.