Improving or Decline

[wiretowire68]

I have a question, maybe someone will answer me, maybe not. I am not strong enough to come up with an equation that used randomness when it comes to predicting a horse will improve on a given race or decline. My experience has taught me in reading past performance that as certain angles have been suggested i.e. Mike Pizzolla using CE, CL in terms of Advance Form Pattern recognition in terms of class drops in a form cycle.

I also agree as Dave Schwartz explains that there are two types of speeds as a horse moves to hitting the board or winning. I have always seen that with certain conditions of a form cycle that you obviously see negative lengths lossed between 2nd and 3rd Calls, sometimes back to back, sometimes 1 race improv..sometimes the horses manages to stay with the pace and lengths behind i.e. one paced etc. My question is in my design as I go forward, I am try to use forecasting model in terms of predicting that a horse may improve after two-three times losing ground.. Then there are the horses that show positive improvement within their latest starts.....My question is there a function that can be used within the framework of averaging out lost or gained lengths and lost and gained speed within three starts that can show a prediction of a horse mathematically improving or declining ... Help or suggestions, that are easily understandable is great admired and appreciated as my learning curve goes .. as usual.. spreadsheeting takes a long time to do.. trying to shorten it, instead seems like it takes longer. LOL>

[denniswilliams]

Here's what I'm struggling with:


I have 3 apples: red delicious; macintosh; granny smith


I need an algorithm that would capture the taste nuances between the three: sweet, to 'middle' to tart. I'm sure there must be some machine learning algorithm that will very easily and precisely classify these apples. Some formula surely exists.



Oh, and I omitted that I have NEVER TASTED any of these apples. Have no idea how they rank in terms of quality of taste.



Shouldn't be a problem, right?

[MJC922]

https://youtu.be/nlpsAwzo_wk

[Dave Schwartz]

I also agree as Dave Schwartz explains that there are two types of speeds as a horse moves to hitting the board or winning.

I'm not actually sure what you have attributed to me here. Do not recall a reference to two speeds.

I am not strong enough to come up with an equation that used randomness when it comes to predicting a horse will improve on a given race or decline.

While I am a fan of randomness in some situations, I do not see it as applicable to form.

In NewPace (https://store.pacemakestherace.com/newpace-early-vs-late-video-seminar-download/), I taught an approach that used what is called 90% CI.

So, named by the author of this book - meaning 90% confidence interval.
How to Measure Anything: Finding the Value of Intangibles in Business (https://www.amazon.com/How-Measure-Anything-Intangibles-Business-ebook/dp/B00INUYS2U/ref=sr_1_1?crid=2CKXZ4C7HDKIK&keywords=how+to+measure+anything&qid=1706060741&sprefix=how+to+measure%2Caps%2C178&sr=8-1)


That approach uses the concept that if you had a horse with 5 speed ratings, you would take the best and the worst, and have a 90+% statistical likelihood that what the horse runs today would be between those two numbers.
As the number of races increase, you'd toss a race, taking the 2nd best and the 2nd worst.


This works surprisingly well, especially considering that the entire premise begins on the wrong foot.
That is, the races going back thru history, ARE NOT equal. There is a strong bias towards more recent races.



There is also the issue of horses being poorly placed (or simply not trying) in some races.

It demanded research to sort it out.

Still, despite all the warts, it works reasonably well.

But, just to be clear, this is not FORM analysis.


Dave

[wiretowire68]

I do understand Dave, it is more about as you suggest tiered speed. My experience in watching races has shown me that e.g. if a horse in his latest form cycle has shown negative lengths (lost ground between 2nd-3rd Call) i.e. not trying, bad start, shuffled etc. not fit.. I have picked many horses horses that have been showing negative form....not picking up ground and not even hitting the board.. the question was mathematical and if it is not form analysis, what would it be???????? If a handicapper looks at the past performances..Brisnet, Equibase, DRF, Post Time Solutions, Betmix..presuming that all running lines appear the same, only differences in their idea of speed figures..Beyer, Bris, Equibase, SR..your fixing of speed numbers based on Repent (Which is a simple example) or Jim Cramer's idea of running style, positioning.

You are saying that there is no function i.e. like forecasting, linear regression analysis that can give a horse predictable type number based on form cycle on whether he can improve or decline. It becomes obvious that when the whales see the 3/5 shot on the board and his form indicates that he should be 3/5 because of all the readable intangibles the horse shows. However, he runs a clunker or gets beat but the horse with negative form goes off at a good odd wins or hits the board. I have seen horses run negative form 3-4 races in a row, maybe a small class drop, maybe the same class and it appears that he should run bad again, but my theory believes that there are many horses who either decide to wake up and run a big number or other horses who have won in a cycle show positive gain in their calls... and naturally some of them decline and get beat and the public gets suckered because his form shows positive gains in his running lines.

So my point was is there a calculus function or a derivative function that can be used against the field in terms of independent variable.

[Dave Schwartz]

(1)You are saying that there is no function i.e. like forecasting, linear regression analysis that can give a horse predictable type number based on form cycle on whether he can improve or decline. (2)It becomes obvious that when the whales see the 3/5 shot on the board and his form indicates that he should be 3/5 because of all the readable intangibles the horse shows.

1. I am not saying that.
One can create a formula that regresses just about anything. Whether or not it works, or whether it represents form, is another matter.

The big question is: "Is it causation or correlation only?"


2. It is NOT FORM that allows them to make their decisions.
It is a combination of iterative regression of known horse handicapping factors and CALIBRATION of how the public will bet such horses, considering the GATE ODDS.



IOW, they're not predicting FORM. They're predicting hit rate and it is VERY accurate - especially the CALIBRATION part.

Because the horses typically run well, it MANIFESTS as form.




.

[wiretowire68]

Yes...need a crystal ball. I had a question for you regarding your speed fig analysis on Repent as I know this was an easy example to use in a teaching/seminar format. As a handicapper or a teacher when you come to doing those figures up against the par time, let us say in pen and paper format. Do you still do the figures at other distances if there are horses in the field that have not run that distance or have varying distances in their past performances. Secondly, do you adjust for the distance..?

[Dave Schwartz]

Yes...need a crystal ball. I had a question for you regarding your speed fig analysis on Repent as I know this was an easy example to use in a teaching/seminar format. As a handicapper or a teacher when you come to doing those figures up against the par time, let us say in pen and paper format. Do you still do the figures at other distances if there are horses in the field that have not run that distance or have varying distances in their past performances. Secondly, do you adjust for the distance..?

There are some adjustments to be made, but, frankly, paper & pencil can be a lot of work. Don't overdo it.

I mean, sprints-to-routes (and vice-versa) can be tricky. I'd probably not bother for SPEED ratings.

Pace ratings are a different story.


Did you watch the video Using Pars, Speed & Pace Ratings (https://pacemakestherace.com/using-pars-speed-pace-ratings-video/)?

Might help.

[wiretowire68]

The question still remains... let us say, we take a race whereby we are using the last 3-4 races within the cycle or within the last two cycles as in the current cycle a horse has 1 start or only two starts. My question is I will take the 2nd call to the stretch, or the stretch to wire and will put it into a cell on whether they have lost ground or picked up ground.. ie. lengths gained or lengths lost I.e. +-. Within my function in terms of a possible regression or "Improvement or prediction of improvement without a whole whack of data, Is there a statistical or function, ie. like =(data, range of data, range of data) is there a greater than you can use, or less than...

For instance: #1 horse 2nd call-third call : Positionally 6th -4.00 lengths
#2 horse 2nd call-third call: Positionally. 3rd.- loses no ground say hd. than a head.

And you have array of these types in the field. Let us add up or use a probability as a percent who is likely to improve or decline in terms of running. Easy math shows who has improved and who has declined in their bad running line. There is no function that can take these against the others and show a possible + improvement vs. - improvement?


Me.

[Jake]

Forget beaten lengths and running positions for all of this. it's a mistake.
Think about what you can do with either pace figures or Sartin energy
calculations. Scale them and connect the dots. Only advice I can offer here,
for what's it worth.

[Dave Schwartz]

Forget beaten lengths and running positions for all of this. it's a mistake.
Think about what you can do with either pace figures or Sartin energy
calculations. Scale them and connect the dots. Only advice I can offer here,
for what's it worth.

^^^^^^
THIS.

What Jake said.

I'd add that an energy expenditure number would be better than SPEED or PACE because both are highly dependent upon the early pace of the race.

[wiretowire68]

copy of my work on spreadsheet designing through Brisnet Past Performances. My improve/decline showing yesterday's 8th race at Aqu with rankings. Like Bris in terms of Prime Power it shows the # 5 horse Midtwown Rose as the number 1 rank after strength of race, speed and in categories including my design of post position and Jockey Trainer angles with improvement/declining in terms Positive gains in lengths or negative and other factors which is simple. Brisnet did not show the # 2 Darn That Song in the top 3 and I had him as top 3 horse ready to improve based on pps.etc. which he had the lead late and finished 2nd at 14-1.

I cashed the exactor in this race and the triactor as you notice in my rankings, the # 8 and the #3. My improve or decline numbers, the way I designed show it as the horses lowest number needed to show the best chance at improving in this race.
Have a look, tell me what you think?

[wiretowire68]

or think Bris uses Quirin points to base a horses running style based on their data base of PP's. Funny, Midtown Rose shows as E horse but wins the race closing as what Mr. Cramer on RS Positioning calls a PS or a Presser who Sustains his bid and takes over at the 1/8 pole. So.. maybe do to a sloppy track.

I maintain that trainers do try to teach their horses to obviously be versatile. However, many end up with different running styles throughout their form cycles.

P.S. thanks to Dave Schwartz for pointing me to HDW and Jim Cramer's positioning ideas.. I did understand this but have kind of added it to my understanding of how to judge form cycles to understand running style. Of course depending on Pace and if sprinting or routing.

[Dave Schwartz]

That's a nice looking piece of work.

Glad I could help.

[wiretowire68]

it takes forever, even with Macros because Post Position and Trainer % fluctuate