Trainer Past Performances

[Gakiss2]

The idea came up in the thread on gaging a horse's 'fitness' based on the works and races recorded in the PP's.

And is also a tangential thought on the Trainer Superstars thread.

Is it worthwhile to think about coming up with a 'score' and/or a 'style' for trainers in much the same way we have speed and pace figures as well as Quirin Points and Running Styles for horses?

Perhaps there could be Byers like 'figure' based on a trainer's past success with getting and keeping horses in shape. I am sure it would be based somewhat on their Win percent and ROI but go a bit deeper. Also it appears that besides pure ability / resources, trainers take on different 'styles' or approaches. If those could be categorized then it might help us get more out of the 'moves' that are being made such as distance switches, layoff and work patterns and etc.

My image is different look to the Trainer section of the PPs. Instead of: Baffert wins 78% of the time with and ROI of -$0.18. It would say: Baffert Trainer Score 112, Style W 4 (Workout intensive, 4 points).

I don't know how Trainer's would feel about being rated in the same way their horses are but it seems fair game to me.

Also, specifically to the 'superstar' trainer question: Shouldn't we come to grips with the fact that (I'll pick on Baffert) Baffert is actually the executive of a large successful training organization and he has several top notch trainers working for him in that organization. Why can't we track on the actual person in charge of training the horse instead of calling them all Baffert horses?

Anyway, maybe a couple of juicy thoughts to kick off some discussion.

[Tom]

I have two "PP" reports for trainers - one by horse,date, so I can see the moves he makes and the other by date, so I can see what he is doing right now. Very helpful.

[TheOracle]

Originally Posted by Gakiss2

The idea came up in the thread on gaging a horse's 'fitness' based on the works and races recorded in the PP's.

And is also a tangential thought on the Trainer Superstars thread.

Is it worthwhile to think about coming up with a 'score' and/or a 'style' for trainers in much the same way we have speed and pace figures as well as Quirin Points and Running Styles for horses?

Perhaps there could be Byers like 'figure' based on a trainer's past success with getting and keeping horses in shape. I am sure it would be based somewhat on their Win percent and ROI but go a bit deeper. Also it appears that besides pure ability / resources, trainers take on different 'styles' or approaches. If those could be categorized then it might help us get more out of the 'moves' that are being made such as distance switches, layoff and work patterns and etc.

My image is different look to the Trainer section of the PPs. Instead of: Baffert wins 78% of the time with and ROI of -$0.18. It would say: Baffert Trainer Score 112, Style W 4 (Workout intensive, 4 points).

I don't know how Trainer's would feel about being rated in the same way their horses are but it seems fair game to me.

Also, specifically to the 'superstar' trainer question: Shouldn't we come to grips with the fact that (I'll pick on Baffert) Baffert is actually the executive of a large successful training organization and he has several top notch trainers working for him in that organization. Why can't we track on the actual person in charge of training the horse instead of calling them all Baffert horses?

Anyway, maybe a couple of juicy thoughts to kick off some discussion.

Hey Gakiss

I rank Trainers based on points and return on investment but the ROI has to be greater than 0

Sometimes I come up with Trainers with nice payouts across the board even though they may not have won their particular race

I’ve been toying with a Ranking system
http://www.paceadvantage.com/forum/s...44#post2452844

Looking at it this way led me to some great finds
http://www.paceadvantage.com/forum/s...25#post2452125

I missed Gargan at Aqueduct on the Turf twice but I’m waiting for him to strike again hopefully he will hit the board 2 or 3 more times before the season is finished!!!

[Gakiss2]

What was the method to assign the points to the trainers?

ROI makes a lot of sense but how about ROI when you are convinced the trainer has Win intentions. And then of course, know how to determine when the trainer has Win intentions.

I think I recall equibase having publically available trainer rankings but I think it is straight ROI or worse, win percentage.

Thank you for your help.

[jay68802]

Only problem I see is that not all the changes a trainer makes are reported. More like only 25% of the changes you see. Not all works are reported either.

[castaway01]

I get what you're saying, but I'd argue that win percentage and ROI are the most important ways to tell if a trainer is effective at a "move". You could have a trainer who improves horses but doesn't have the stock to win, or you could have a trainer that wins at 35% on the drop but the average horse pays $4.40. While it would be nice to give credit to the (seemingly) skilled trainer who doesn't win, it doesn't really do you any good financially to do so, at least not in broad strokes.

Now granted, those 8% moves sometimes turn into 25% moves when you combine them with two OTHER moves, but I'm not sure how you would tell that from a rating. It's more something you have to dig (or program) to find.

[thaskalos]

The most important thing to me is whether or not the trainer is able to improve upon a horse's performance after newly acquiring the animal...or if the trainer can keep a horse's form intact while placing the horse in different race situations than the horse has previously been exposed to. And I judge such "improvement" and 'form' not by things like trainer win% or trainer-tactic ROI...but by using speed/pace figures as a performance determinant. A horse's performance can improve substantially even if a horse LOSES the race...while a winning performance can sometimes be deemed to be a DECLINING effort. I want to know if a trainer can improve the horse FIGURE-WISE...and then it's up to ME to decide if such a figure-improvement will lead the horse to the winners circle.

[Gakiss2]

I can't disagree but how to translate that ability into a score we can then take to that Trainer's next race.


The horse can't be judged independently of the trainer any more than the trainer can be judged (scored) independent of the horse.

Its all situational but can we use scores (figures) of horses and scores (figures?) of trainers to support the decision making process.

I really liked the thoughts on whether or not a trainer can take a horse to a new situation and support his winning effort as that should produce odds. I know I am hesitant when I see a long term sprinter suddenly taking on some classy Routers. In fact I have lost some bets that way, tossing the sprinter out then watching him wire the field.

Thank You for your input and advice.

[eqitec]

I use the trainer performance stats supplied by BRIS in the $1.drf data files. They supply as many as 6 stats, some of which ridiculously irrelevant and/or misleading.

For the stats which are useful, I've built analytics showing when trainer's performance for any reported attribute is N% higher than his/her overall winning %s (for at least X#starts.) Todd Pletcher's winning % with 1st-time starters might be 28%, but if that's not much different than his overall winning %, then it's no big deal.

Alternatively, if another trainer has an overall winning performance stat for 1st time starters = 28% AND their overall winning % is 10%, then that is a big deal. The variance between the 28% and the 10% are highlighted on my analytics (shown in the image) and scoring (not shown) is proportional to the size of the variance. The scoring for these attributes are then factored in to the other analytics for that horse.

[Dave Schwartz]

You mean like these?



Better look...

[cj]

Formulator has customizable trainer past performances that show every running line for up to five years under all conditions and can be filtered to show the ones you want to see.

[Gakiss2]

Originally Posted by Dave Schwartz

You mean like these?



Better look...

Dave

Yes, that is certainly in the spirit of what I am thinking of. If I understood it looks like it is a look at various forms of ROI based on the situation. I think ROI is an important part of 'Trainer Performance', I'm not sure its the whole picture. Imagine if we bet on horses based on ROI, even if it is a very detailed picture of ROI under different circumstances, and not anything else; like speed, pace, running style and etc. So not knocking what you have there, just thinking it could be considered more deeply. I am particularly interested in 'Style' or methodology and thinking it might be applied in an analogous way as we use 'Running Style' for horses. So just like that approach, we would have to wait for a data producer to start including that in their offering.

Thank You very much for your insights, I have used some of your products with some success.

[Gakiss2]

Originally Posted by cj

Formulator has customizable trainer past performances that show every running line for up to five years under all conditions and can be filtered to show the ones you want to see.

I did not know that. Thank you, I'll check it out.

[Dave Schwartz]

Originally Posted by Gakiss2

Dave

Yes, that is certainly in the spirit of what I am thinking of. If I understood it looks like it is a look at various forms of ROI based on the situation. I think ROI is an important part of 'Trainer Performance', I'm not sure its the whole picture. Imagine if we bet on horses based on ROI, even if it is a very detailed picture of ROI under different circumstances, and not anything else; like speed, pace, running style and etc. So not knocking what you have there, just thinking it could be considered more deeply. I am particularly interested in 'Style' or methodology and thinking it might be applied in an analogous way as we use 'Running Style' for horses. So just like that approach, we would have to wait for a data producer to start including that in their offering.

Thank You very much for your insights, I have used some of your products with some success.

I especially like that last part. ;-)

Thank you.

The key to using trainer stats is to see them from the 10,000-foot level.

What we do is to create a subset of the "angles" that apply to each horse, and then average all the data for each category of the subset.



What I show above (lower-left) are the "systems." At the top are the trainers in this race, and the bottom-right are the applicable categories (or angles) for this horse.

Thus, if you look in the top section, at the highlighted trainer, you see that the categories below together average 25 starts and 3 wins.

Of course, there are a lot more columns in the table.

Here's the key point...

Each of the columns is available as a HANDICAPPING FACTOR!

Thus, looking at the "systems" - each column (such as win IV or place IV) could be populated directly into a weighted handicapping calculation!

This has a lot pf potential for pushing winners to the top, which is, of course, the logical goal.

[Gakiss2]

Originally Posted by Dave Schwartz

I especially like that last part. ;-)

Thank you.

The key to using trainer stats is to see them from the 10,000-foot level.

What we do is to create a subset of the "angles" that apply to each horse, and then average all the data for each category of the subset.



What I show above (lower-left) are the "systems." At the top are the trainers in this race, and the bottom-right are the applicable categories (or angles) for this horse.

Thus, if you look in the top section, at the highlighted trainer, you see that the categories below together average 25 starts and 3 wins.

Of course, there are a lot more columns in the table.

Here's the key point...

Each of the columns is available as a HANDICAPPING FACTOR!

Thus, looking at the "systems" - each column (such as win IV or place IV) could be populated directly into a weighted handicapping calculation!

This has a lot pf potential for pushing winners to the top, which is, of course, the logical goal.

Thank you for clarifying further what the data means. I will certainly keep this approach in mind as I go forward. I feel your example does well at exploring percentages when the angle 'conditions' present themselves (e.g. distance switch, layoff, etc.).
Teasing out angles, trainer intentions, trainer motivation is an interesting and complicated topic.