TimeformUS Figs Key to Belmont Longshot Win

[Tom]

It is extremely interesting.
I'll be digesting it and playing with it for a long time!

[classhandicapper]

Originally Posted by thaskalos

Very enlightening! This is EXACTLY the type of pace analysis that I find most interesting...because of its "practical" value.

I can provide more detailed analysis with percentages. I have all the data stored. But I'm not sure how useful that would be to other people. The projected running styles, projected pace pressure, analysis of the race flow based on the charts etc.. all use my own metrics and standards.

So what the data suggests is true for me may not be true for people using different information or techniques.

It would be like me doing a study on TG figures and how to use them. If would be interesting, but if you were using Beyer figures without ground and weight adjustments it might screw you up.
Both can be good, but they have to be used differently.

What I was trying to get at was that you can demonstrably prove with data that you can make valid pace projections often enough to be useful. You can also demonstrate that more is going on than just fractions. If you track your results, you can start estimating how often you will be right.

[classhandicapper]

Originally Posted by cj

So what I posted was not? Good to know.

What you are posting is excellent.

I suspect he was being sarcastic in that he didn't find my post very useful because I didn't provide the data, just the general findings and conclusions.

[bobphilo]

Originally Posted by cj

What should I do next? Suggestions?

You can do with patterns is exactly what you've been doing with the single pace figure. To see what effect each pattern has on a horses performance and to see if requires a horses' final figure be upgraded or down graded. You've done the research which on how to adjust a horses final figure based on pace figure (fast or slow) All you need to do now is the same thing based on the different pace patterns.
What you want to do is to compare impact values of the different pace patterns. Beyond that you want to do beyond that is see to what extent the degree of unevenness of the pattern determines to what degree to you will adjust the final figure, just as you now use the degree of deviation of the single pace figure to determine the to what extent to adjust the final figure. You can then develop an algorithm for adjusting the final figure based on the pattern, just as you now have an algorithm for the effect of a single pace figure on the final figure. In order to measure the performance of the horses in question and how they are affected by pace patterns you can use win%, ITM%, impact value and most importantly, adjustment to final figure.

It's basically what you've been doing in measuring the effect of a pace figure to the final figure only it's a more complicated due to the greater number of pace patterns than there are variations to a single pace figure.

Hope that helps.

[bobphilo]

Originally Posted by classhandicapper

My data suggests:

1. Among races I projected to have a fast or slow pace, on average they DO tend to have fast or slow paces according to pace figures. But as we all know some don't go according to plan.

2. Among races that pace figures suggest were fast to slow early, the winner tends to come from further back or closer up as you'd expect. However, it gets tricky to measure impacts correctly. Large fields tend to have faster paces and vice versa. So you have to control for field size to know how much of where the average winner came from was related to the pace and how much was related to the field size.

3. Among races I projected to have a fast/slow pace that pace figures suggest DID go according to plan, sometimes the race flow does not follow the pace figures and projections. How the track was playing or whether there was a dominant speed seems to impact that.

4. Among races I projected to have a fast/slow pace that pace figures suggest did NOT go according to plan, the race flow sometimes follows the projection instead of the pace figures. That suggests that sometimes things other than the fractions impact the flow.

Getting at some of those probabilities using your own techniques and figures seems to help me with value judgments.

If I'm projecting a fast pace and given how much pressure there is I'm expecting to be right 60% of the time, and 50% of the time I get the projection right the race falls apart, that means 30% of the time horse A, B, and C are in deep trouble. At least I have a ballpark basis for my thinking instead of just "it looks like a fast pace".

What you are saying is true, but remember, what cj is studying is the effect of pace patterns on the final figure as opposed to the effect of the single pace figure that is currently being used to adjust the final figure. Much of what you are saying may be useful but is not relevant to the specific purpose of the study. This question is what I raised in the beginning of this thread and what cj is attempting to answer in his study.
Not once in your post did I see mention of the term "pace pattern', whose effect is what this thread is about. Whatever can be learned in using the pace pattern to understand and predict a race result by calculate the final figure as accurately as possible is of great practical value in handicapping.

[cj]

Originally Posted by bobphilo

You can do with patterns is exactly what you've been doing with the single pace figure. To see what effect each pattern has on a horses performance and to see if requires a horses' final figure be upgraded or down graded. You've done the research which on how to adjust a horses final figure based on pace figure (fast or slow) All you need to do now is the same thing based on the different pace patterns.
What you want to do is to compare impact values of the different pace patterns. Beyond that you want to do beyond that is see to what extent the degree of unevenness of the pattern determines to what degree to you will adjust the final figure, just as you now use the degree of deviation of the single pace figure to determine the to what extent to adjust the final figure. You can then develop an algorithm for adjusting the final figure based on the pattern, just as you now have an algorithm for the effect of a single pace figure on the final figure. In order to measure the performance of the horses in question and how they are affected by pace patterns you can use win%, ITM%, impact value and most importantly, adjustment to final figure.

It's basically what you've been doing in measuring the effect of a pace figure to the final figure only it's a more complicated due to the greater number of pace patterns than there are variations to a single pace figure.

Hope that helps.

Definitely working on that part of it. My question was more for public consumption, what should I do next? Randall made some broad suggestions, but I'm looking for more specific. Dirt route? Turf sprints? Break down 6, 6.5, 7? Probably won't get to much until Monday. Monday through Wednesday, even Thursday sometimes, have much less figure making workload.

[cj]

Originally Posted by classhandicapper

What you are posting is excellent.

I suspect he was being sarcastic in that he didn't find my post very useful because I didn't provide the data, just the general findings and conclusions.

I did assume that, just busting his chops a little.

[Tom]

Dirt route? Turf sprints? Break down 6, 6.5, 7?

Yes!

[thaskalos]

In all sincerity, this is a great topic...but this debating format is too restrictive, IMO, and it doesn't allow the full development of the different opinions that are sure to arise. Can we perhaps set up a roundtable discussion on Skype, or something?

[Maximillion]

Originally Posted by bobphilo

What you are saying is true, but remember, what cj is studying is the effect of pace patterns on the final figure as opposed to the effect of the single pace figure that is currently being used to adjust the final figure. Much of what you are saying may be useful but is not relevant to the specific purpose of the study. This question is what I raised in the beginning of this thread and what cj is attempting to answer in his study.
Not once in your post did I see mention of the term "pace pattern', whose effect is what this thread is about. Whatever can be learned in using the pace pattern to understand and predict a race result by calculate the final figure as accurately as possible is of great practical value in handicapping.

I think that something like this "pace pattern" could be a reason to maybe look at another recent race?...if it was a bunch of horses trying a route for the first time I would probably be leaning in that direction.
Hard to say,since it was a "sprint" and usually these are all about attacking early, etc etc and since these horses have no real history and could improve greatly from one start to the next it only adds to the confusion, at least for me.

[classhandicapper]

Originally Posted by bobphilo

What you are saying is true, but remember, what cj is studying is the effect of pace patterns on the final figure as opposed to the effect of the single pace figure that is currently being used to adjust the final figure. Much of what you are saying may be useful but is not relevant to the specific purpose of the study. This question is what I raised in the beginning of this thread and what cj is attempting to answer in his study.
Not once in your post did I see mention of the term "pace pattern', whose effect is what this thread is about. Whatever can be learned in using the pace pattern to understand and predict a race result by calculate the final figure as accurately as possible is of great practical value in handicapping.

Here's reality as I see it.

1. We have a theory on what should happen given a certain pace pattern.

2. We have data that suggests "on average" certain things do happen given that pattern

3. We have a specific race where the pace pattern suggests something specific should have happened.

4. The race didn't develop anywhere near what our theory, the averages, or the pace pattern suggests should have happened.

So what happened?

We can pretend that what should have happened did happen and misunderstand the race or we can try to figure out why what we thought would happen didn't happen.

What I am adding is that I have data that suggests the pace or pace pattern can become less relevant depending on the makeup of the field, how the race developed positionally, and how the track was playing.

Pace handicappers sort of understand that biases matter, but the makeup of the field part of it less so.

If there were 6 speed horses in the race and it developed in pattern A that's way different than if there were 0 speed horses in the race and it developed in the exact same pace pattern and fractions.

[classhandicapper]

Originally Posted by cj

Definitely working on that part of it. My question was more for public consumption, what should I do next? Randall made some broad suggestions, but I'm looking for more specific. Dirt route? Turf sprints? Break down 6, 6.5, 7? Probably won't get to much until Monday. Monday through Wednesday, even Thursday sometimes, have much less figure making workload.

You definitely have to break up dirt and turf because every study I have done indicates that small variations of pace from the average in turf races have way less of an impact on the result than similar variations on dirt.

[classhandicapper]

Originally Posted by classhandicapper

You definitely have to break up dirt and turf because every study I have done indicates that small variations of pace from the average in turf races have way less of an impact on the result than similar variations on dirt.

If you want to check something new, I have an idea. This is a study I am eventually going to do, but I haven't had the time yet.

I want to check "off the turf" races.

What makes those races interesting is that even though an MTO or two may sneak in, they contain some turf horses also. Turf horses are generally trained differently than dirt horses. They are trained to go slow early and finish well.

If you look at pace variations in those races, there might be some interesting results.

For example if the pace is slow on turf, closers tend to do OK anyway.

What happens if the pace is slow on dirt but it's a bunch of turf horses?

Do they still close better than in the typical dirt race because they are trained to do that or does the race develop the same as any other dirt race?

In between?

You might have to weed out the ones on sloppy tracks. You'll be looking a small subset where the race came off the turf but the dirt was fast/good, but it might be worth it to understand pace and style impacts.

[bobphilo]

Originally Posted by cj

Definitely working on that part of it. My question was more for public consumption, what should I do next? Randall made some broad suggestions, but I'm looking for more specific. Dirt route? Turf sprints? Break down 6, 6.5, 7? Probably won't get to much until Monday. Monday through Wednesday, even Thursday sometimes, have much less figure making workload.

By all means, you can do the same analysis for different types of races such as dirt, further broken down for dirt sprint vs, route, or grass, sprint vs. route, and all combinations thereof. I think that at least one of the things you were were asking for in was how you should go about it (methodology), - the all important step in deciding how to best answer the research question.
It's the same as what you did where you related variations in a horses speed figures in a race to variations in pace and creating an algorithm determining just how much a given variation in pace affects the final speed figure. That would be a regression analysis. You just have to do the same thing for pace patterns and break it down into as many types of races (dirt, grass, synthetic, sprint, route) as you want to as long as you don't reduce the sample size too much in being very specific. An issue in research is always sample size compared to specificity. I will deal with that issue in more detail in my upcoming reply to Classhandicapper's post.

As a statistician, I will be glad to answer any questions you have in methodology and study design, though I think you've got that down based on your previous work. Similarly, if you are seeking relevant observations based on my personal experience, I will be glad to oblige, as I'm sure other posters would.

[cj]

Originally Posted by classhandicapper

You definitely have to break up dirt and turf because every study I have done indicates that small variations of pace from the average in turf races have way less of an impact on the result than similar variations on dirt.

Of course, I would never lump them together. Just to be clear I didn't in the first one either. It is 6, 6.5 and 7f dirt races. I think I said that but don't want to scroll through the whole thread.