Optimal intra track-surface adjustments

[DeltaLover]

I am currently revisiting the implementation of my speed figures, focusing on a topic that I find very interesting and sufficiently complex both from the data collection process scope of view and the optimal algorithm to use as well.

The problem lies in the creation an optimal intra track-surface adjustments, which I believe is very poorly addressed in the existing bibliography. Currently I have a few ideas of how to attack the problem although I would like to hear from others about it.

What is your solution?

[raybo]

"Intra" - "within", meaning within the same track. I think the more difficult adjustment is "inter" ("between" or "among"), meaning from track to track.

"Intra" surface adjustments include air temperature, air density, surface density and friction, wind force and direction, path bias changes, etc..

"Inter" surface adjustments include all the above, plus, track configuration differences, turn radii differences, turn slope differences, length of turns differences, length of straights differences, starting gate locations differences, etc..

Good luck with your study/potential solutions.

[DeltaLover]

"Intra" - "within", meaning within the same track. I think the more difficult adjustment is "inter" ("between" or "among"), meaning from track to track.

"Intra" surface adjustments include air temperature, air density, surface density and friction, wind force and direction, path bias changes, etc..

"Inter" surface adjustments include all the above, plus, track configuration differences, turn radii differences, turn slope differences, length of turns differences, length of straights differences, starting gate locations differences, etc..

Good luck with your study/potential solutions.

You are right about your definitions. The way I see the problem is to derive adjustments for any kind of track - surface pair though. For example comparing BEL - D to BEL - T or AQU - d to TP - A both fall into the same category.

As far as track configuration differences, turn radii differences, turn slope differences, length of turns differences etc, of course they affect the final times and the raw figures but my approach is purely numeric and do not take any of these factors into account based in the assumption that they are already embedded in them.

From a pure algorithmic scope of view. the solution to the problem ranges from the naive approach that is described in Beyer's books to a some very sophisticated methodologies involving graphs, simplex method etc..

Starting this thread, I was curious about how others are attacking this challenging and pragmatic problem but as it turns out there is no interest in this topic as the PA community seems to be more attracted to threads like What is your favorite type of wager? (http://www.paceadvantage.com/forum/showthread.php?t=127428) or even Creed-End of the Rocky movies? (http://www.paceadvantage.com/forum/showthread.php?t=127445)

[raybo]

So, by only considering one track, but dirt versus turf, these adjustments will only come into play when horses at that track change surfaces. How often does that happen? Or are you thinking that, in the case of horses who have run both surfaces, at the same track, you are hoping to equalize them in order to not have to ignore past performances due to them being on a different surface than today's race?

[Tom]

No need to get insulting about it.
Maybe most of us here do not see a problem with the ways in use now - TFUS, T-Graph/Rags, Beyer, HTR, J-Capper.

If you thing there is something better, you have proven standards to measure the results, which are all that matter.

[cj]

Don't get your panties in a wad. The thread has been up a day, and it was a very busy day of racing. It would have drawn some interest I'm sure. Now, who knows?

[DeltaLover]

So, by only considering one track, but dirt versus turf, these adjustments will only come into play when horses at that track change surfaces. How often does that happen? Or are you thinking that, in the case of horses who have run both surfaces, at the same track, you are hoping to equalize them in order to not have to ignore past performances due to them being on a different surface than today's race?

Exactly these are some of the questions that we need to answer in this exercise.

I general, we should start with a hypothesis that I try to support with empirical data. Note that we have several problems to solve here:


The creation of the raw figures, something that probably involves the development of a track variant, class pars, distance to distance adjustments etc
The hypothesis of what might be a good way to compare figures across distances and surface. For example we might translate every winning time to its corresponding figure and then assume that shippers who finish first or second in two different tracks within a specific time span are good candidates for adjustment estimation
The algorithms to use to optimize the adjustment times. For example, in my database I have over 1,800 possible pairs of track - distance and some of them are connected directly by one or more suitable horses while some of them are connected by a certain path (meaning Track X might not be directly connected to Track Y but still there is a track Z that connects them)


As I said before this is a challenging problem to solve efficient and parametrically and this is why I started this thread..

[DeltaLover]

Don't get your panties in a wad. The thread has been up a day, and it was a very busy day of racing. It would have drawn some interest I'm sure. Now, who knows?

I know CJ, it certainly was a busy day for most of us!

[DeltaLover]

No need to get insulting about it.
Maybe most of us here do not see a problem with the ways in use now - TFUS, T-Graph/Rags, Beyer, HTR, J-Capper.

If you thing there is something better, you have proven standards to measure the results, which are all that matter.

I am not sure that there is something better, I am currently trying to see if it can be discovered

[DeltaLover]

To make the conversation more tangible, you can see a list of all the pairs and their adjustments based on my database and my proprietary raw speed figure and method to select matching horses in the following spreadsheet:

http://www.themindofagambler.com/cross-track-adjustments.xlsx

The actual methodology of the raw speed figure creation is not important here as the selection of matching horses is neither. The objective is to optimize the adjustments of each pair in an optimal way to minimize errors. A negative adjustment means that the second track - surface pair needs to be subtracted by this number in order to equate the first. Note that we can have a very large number of connecting paths from track to track and how to handle this is a good part of the challenge..

Note that what i call raw figures are already comparable for the same track and surface, so their rawness is only in comparison to other track - surfaces

[PaceAdvantage]

And to be fair, handicapper's corner isn't the most popular section of the site. There are only 12 people viewing this section right now, while there are 75 viewing General Handicapping Discussion. Want me to move to to General Handicapping discussion?

[DeltaLover]

And to be fair, handicapper's corner isn't the most popular section of the site. There are only 12 people viewing this section right now, while there are 75 viewing General Handicapping Discussion. Want me to move to to General Handicapping discussion?

Sure, I think it makes perfect sense!

[PaceAdvantage]

No problem...thread moved...

[Cratos]

I am currently revisiting the implementation of my speed figures, focusing on a topic that I find very interesting and sufficiently complex both from the data collection process scope of view and the optimal algorithm to use as well.

The problem lies in the creation an optimal intra track-surface adjustments, which I believe is very poorly addressed in the existing bibliography. Currently I have a few ideas of how to attack the problem although I would like to hear from others about it.

What is your solution?
The reply in post #2 from the poster, Raybo offers the correct solution if you have the correct empirical data.

However if your hypothesis is based on anecdotal data as the majority if not all of the speed figure methodologies are, the Raybo’s post might be of little help to you.

Therefore you asked a question about “track-surface adjustments” (intra/inner) and my suggestion would be to start with the coefficient of kinetic friction.

For example, if you have calculated the coefficient of kinetic friction for the dirt surface and the coefficient of kinetic friction for the turf surface at a particular racetrack or two different racetracks the next step would to determine the normal force for the horse(s) in question and multiply that normal force by the respective coefficients of kinetic friction.

The difference when divided by the horses’ total mass will give you acceleration in m/seconds.

Convert that into feet/second and divide by the speed of the race of each horse measured in feet/second and you will have the speed loss (so-called track variant) determined by the kinetic friction of both the dirt surface and the turf surface.

Some useful equations:

D (distance) = (1/2) at squared

Velocity =at

Where t = time and a = acceleration

[DeltaLover]

The reply in post #2 from the poster, Raybo offers the correct solution if you have the correct empirical data.

However if your hypothesis is based on anecdotal data as the majority if not all of the speed figure methodologies are, the Raybo’s post might be of little help to you.

Therefore you asked a question about “track-surface adjustments” (intra/inner) and my suggestion would be to start with the coefficient of kinetic friction.

For example, if you have calculated the coefficient of kinetic friction for the dirt surface and the coefficient of kinetic friction for the turf surface at a particular racetrack or two different racetracks the next step would to determine the normal force for the horse(s) in question and multiply that normal force by the respective coefficients of kinetic friction.

The difference when divided by the horses’ total mass will give you acceleration in m/seconds.

Convert that into feet/second and divide by the speed of the race of each horse measured in feet/second and you will have the speed loss (so-called track variant) determined by the kinetic friction of both the dirt surface and the turf surface.

Some useful equations:

D (distance) = (1/2) at squared

Velocity =at

Where t = time and a = acceleration

My approach is completely different and not based on physics but on statistics and algorithms rather, assuming that the chaotic details of surface friction, air resistance, humidity etc are already embedded in the final times . The problem as I defined in my previous posts lies in the selection of the proper data structures and graph algorithms..

[Cratos]

My approach is completely different and not based on physics but on statistics and algorithms rather, assuming that the chaotic details of surface friction, air resistance, humidity etc are already embedded in the final times . The problem as I defined in my previous posts lies in the selection of the proper data structures and graph algorithms..
It will interesting to see the results of your approach and since it is statistical based I hope you will post the confidence level of your conclusion.

[steveb]

You are right about your definitions. The way I see the problem is to derive adjustments for any kind of track - surface pair though. For example comparing BEL - D to BEL - T or AQU - d to TP - A both fall into the same category.

As far as track configuration differences, turn radii differences, turn slope differences, length of turns differences etc, of course they affect the final times and the raw figures but my approach is purely numeric and do not take any of these factors into account based in the assumption that they are already embedded in them.

From a pure algorithmic scope of view. the solution to the problem ranges from the naive approach that is described in Beyer's books to a some very sophisticated methodologies involving graphs, simplex method etc..

Starting this thread, I was curious about how others are attacking this challenging and pragmatic problem but as it turns out there is no interest in this topic as the PA community seems to be more attracted to threads like What is your favorite type of wager? (http://www.paceadvantage.com/forum/showthread.php?t=127428) or even Creed-End of the Rocky movies? (http://www.paceadvantage.com/forum/showthread.php?t=127445)

it's actually quite simple to figure the difference between any surface, track, distance.
you just need to start from the same baseline figures, and do your calculations that way.

for instance if you started used 70 seconds for 1200m on dirt for 'dirthere' and the same 70 seconds for 1200 on turf for 'turfhere' and you had plenty of data, then the difference between the speeds of those tracks would show up in the remainder left after your calculations are done.

of course that is a 'simple as' explanation and there is a more involved, but it works.
i guess one just needs to now how to make it work for themselves without being given the detail.

it works everyhwere i have tried it, and my information is that most tracks in usa have basically the same basic shape, so it would probably be more accurate than other places where one needs to account for topography differences.

[proximity]

I am currently revisiting the implementation of my speed figures, focusing on a topic that I find very interesting and sufficiently complex both from the data collection process scope of view and the optimal algorithm to use as well.

The problem lies in the creation an optimal intra track-surface adjustments, which I believe is very poorly addressed in the existing bibliography. Currently I have a few ideas of how to attack the problem although I would like to hear from others about it.

What is your solution?

although idk if he even made figures for turf, in general for a big computerized figure making project like this i would start by checking out a lot of the old sjk posts here on pa.

[maddog42]

it's actually quite simple to figure the difference between any surface, track, distance.
you just need to start from the same baseline figures, and do your calculations that way.

for instance if you started used 70 seconds for 1200m on dirt for 'dirthere' and the same 70 seconds for 1200 on turf for 'turfhere' and you had plenty of data, then the difference between the speeds of those tracks would show up in the remainder left after your calculations are done.

of course that is a 'simple as' explanation and there is a more involved, but it works.
i guess one just needs to now how to make it work for themselves without being given the detail.

it works everyhwere i have tried it, and my information is that most tracks in usa have basically the same basic shape, so it would probably be more accurate than other places where one needs to account for topography differences.

I am doing something similar. There is one monkey wrench that gets in the way: the improving 3 year old. Beyer talks about how many points a 3 year old will improve, especially earlier in the year. I have a fudge factor to account for this. Also the differences between tracks seems to change a little bit as the meets progress.

[cj]

I am doing something similar. There is one monkey wrench that gets in the way: the improving 3 year old. Beyer talks about how many points a 3 year old will improve, especially earlier in the year. I have a fudge factor to account for this. Also the differences between tracks seems to change a little bit as the meets progress.

This is the biggest challenge I face with figure making. Projecting figures for lightly race two and three year old horses is tough. The easy answer is to just base track variants on older, more established horses. But in today's game, that is just not possible most of the time.

It isn't rare to see a card with more than half being maiden races. The best tracks are running more and more turf races. Tracks often change during the card so you can't just assume the track is the same when the older horses and younger horses run. And, most tracks are running less races with smaller fields which gives a smaller sample size.

When it comes to track to track adjustments (as well as distance to distance) though, it should be fairly easy to stick to the older horses and ignore the young ones if you have a big enough data set.

[DeltaLover]

This is the biggest challenge I face with figure making. Projecting figures for lightly race two and three year old horses is tough. The easy answer is to just base track variants on older, more established horses. But in today's game, that is just not possible most of the time.

It isn't rare to see a card with more than half being maiden races. The best tracks are running more and more turf races. Tracks often change during the card so you can't just assume the track is the same when the older horses and younger horses run. And, most tracks are running less races with smaller fields which gives a smaller sample size.

When it comes to track to track adjustments (as well as distance to distance) though, it should be fairly easy to stick to the older horses and ignore the young ones if you have a big enough data set.

This is a handicapping challenge that represents a different problem than what I am trying to attack here, where I assume that I already have a set of per track figures that I am trying to connect with an adjustment. Obviously, the handicapping part is very important but once we have solve the adjustments problem it becomes easier to apply different handicapping approaches trying to select the optimal.

[cj]

This is a handicapping challenge that represents a different problem than what I am trying to attack here, where I assume that I already have a set of per track figures that I am trying to connect with an adjustment. Obviously, the handicapping part is very important but once we have solve the adjustments problem it becomes easier to apply different handicapping approaches trying to select the optimal.

Yes, that is what I tried to address in the last paragraph. But handicapping and figure making are linked whether we like it or not. You can't be a good figure make if you aren't a good handicapper in my opinion.

Further, I think handicapping is needed for your endeavor as well which is a HUGE part of figure making. You can't figure out the speed difference between tracks and distances if you aren't able to properly handicap the quality of the horses running on each track.

[DeltaLover]

Yes, that is what I tried to address in the last paragraph. But handicapping and figure making are linked whether we like it or not. You can't be a good figure make if you aren't a good handicapper in my opinion.

Further, I think handicapping is needed for your endeavor as well which is a HUGE part of figure making. You can't figure out the speed difference between tracks and distances if you aren't able to properly handicap the quality of the horses running on each track.

:ThmbUp:

Exactly.

The objective of this thread is about the creation of a generic adjustment mechanism which will serve as part of the optimize of the handicapping method.

[Magister Ludi]

And to be fair, handicapper's corner isn't the most popular section of the site.

Mr. PA, I don't believe that you got the memo. "Handicapper's Corner" is now "Capper Al's Corner". Do you suppose that there's a correlation between the content and the popularity, or lack thereof?

[cj]

Mr. PA, I don't believe that you got the memo. "Handicapper's Corner" is now "Capper Al's Corner". Do you suppose that there's a correlation between the content and the popularity, or lack thereof?

To be fair it was never that popular and probably would be extinct without Al's postings.

[thaskalos]

The idea of "equating" the different surfaces and distances relies on the assumption that the horses are being entered to run with the same INTENTION, across all surfaces and distances. And this isn't the impression that I've gotten during my long tenure in the game. The game may appear at first glance to be a mathematical endeavor...but the horses are actually handled by humans who harbor different motives and expectations as the distances and the surfaces change. The turf runner is usually not "well meant" when he is entered on the dirt, nor is the router entered in a sprint with the expectation of winning the race. Trying to come up with an "adjustment" to smooth out the differences between the performances across the different surfaces and distances becomes a futile pursuit when the horses who are changing distance or surface are doing so with motives other than winning in mind.

A pet project of mine had been to try to come up with a suitable adjustment to smooth out the differences in the performances of the horses who were moving up and down the class ladder. It occured to me that such an adjustment would be highly beneficial to me in my pursuit to properly evaluate a horse's "form". But my efforts towards that end were stymied by the realization that the horses who were rising in class were not all doing it with the same intention in mind. Some class risers gave a good account of themselves at the new class level...while others underperformed miserably on the class rise...only to return to their prior good form on the subsequent class drop. And, while some class droppers ran better on the drop, others were "damaged goods"...and were entered not to try and win the race...but to become someone ELSE'S "damaged goods". A numerical adjustment to smooth-out the vast differences of performance in these cases not only appeared difficult to me...it seemed downright unrealistic. And mighty dangerous to the wallet.

As much as I enjoy making adjustments in this game...I must draw the line when I face "apple-and-orange" comparisons. Some aspects of this game must be analyzed separately...IMO.

[classhandicapper]

A pet project of mine had been to try to come up with a suitable adjustment to smooth out the differences in the performances of the horses who were moving up and down the class ladder. It occured to me that such an adjustment would be highly beneficial to me in my pursuit to properly evaluate a horse's "form". But my efforts towards that end were stymied by the realization that the horses who were rising in class were not all doing it with the same intention in mind. Some class risers gave a good account of themselves at the new class level...while others underperformed miserably on the class rise...only to return to their prior good form on the subsequent class drop. And, while some class droppers ran better on the drop, others were "damaged goods"...and were entered not to try and win the race...but to become someone ELSE'S "damaged goods". A numerical adjustment to smooth-out the vast differences of performance in these cases not only appeared difficult to me...it seemed downright unrealistic. And mighty dangerous to the wallet.

That has been a decades long nightmare for me.

Here's another complication.

There is a difference between a lightly raced well bred horse from a top barn moving up in class off a win and the run of the mill horse moving up. The lightly raced one often has more in the tank than it has shown to date and will run FASTER when pressed against better (but only up until a point) while the cheaper one with the exact same figures and running lines will go backwards.

Finally, I have statistical evidence that if you use speed figures alone or class figures alone you may not pick as many top choice winners as if you use both, but the value can deteriorate when you combine them. You have to be really careful about combining 2 factors even when they both outperform the track take independently. The combination sometimes creates WORSE value.

Of course making more accurate odds lines helps, but you can really screw yourself up by adding too many ingredients to the stew.

[DeltaLover]

That has been a decades long nightmare for me.

Here's another complication.

There is a difference between a lightly raced well bred horse from a top barn moving up in class off a win and the run of the mill horse moving up. The lightly raced one often has more in the tank than it has shown to date and will run FASTER when pressed against better (but only up until a point) while the cheaper one with the exact same figures and running lines will go backwards.

Finally, I have statistical evidence that if you use speed figures alone or class figures alone you may not pick as many top choice winners as if you use both, but the value can deteriorate when you combine them. You have to be really careful about combining 2 factors even when they both outperform the track take independently. The combination sometimes creates WORSE value.

Of course making more accurate odds lines helps, but you can really screw yourself up by adding too many ingredients to the stew.


Again, as I have mentioned before, your absolutely (as Thask's as well) correct points are not (directly) connected to the topic of the thread...

[thaskalos]

Again, as I have mentioned before, your absolutely (as Thask's as well) correct points are not (directly) connected to the topic of the thread...
I must have misunderstood. I thought you were trying to make adjustments so you could compare turf races to dirt...and sprint races to routes.

[ReplayRandall]

Finally, I have statistical evidence that if you use speed figures alone or class figures alone you may not pick as many top choice winners as if you use both, but the value can deteriorate when you combine them. You have to be really careful about combining 2 factors even when they both outperform the track take independently. The combination sometimes creates WORSE value.

I have statistical evidence that the INVERSE is also true. Combining workouts and late pace, as a theoretical example, let's just say can show a +EV, but as stand-alone factors are a -EV.....Filtering, combining and layering countless factors takes hard work, immense time and energy resources to come up with profitable algorithms/angles. To use an analogy, it's like trying to predict the movements of an amoeba that's constantly in a state of flux. Beating your competitors to the punch is how you become profitable, knowing when a select method has become non-profitable and moving on, is the hardest part of all, IMO.....

[DeltaLover]

I must have misunderstood. I thought you were trying to make adjustments so you could compare turf races to dirt...and sprint races to routes.

Yes, you misunderstood it Thask.

The purpose of this exercise is to adjust based on track - surface assuming that we already have a valid method to define raw figures ( comparing any race ran in a specific track)

[thaskalos]

I have statistical evidence that the INVERSE is also true. Combining workouts and late pace, as a theoretical example, let's just say can show a +EV, but as stand-alone factors are a -EV.....Filtering, combining and layering countless factors takes hard work, immense time and energy resources to come up with profitable algorithms/angles. To use an analogy, it's like trying to predict the movements of an amoeba that's constantly in a state of flux. Beating your competitors to the punch is how you become profitable, knowing when a select method has become non-profitable and moving on, is the hardest part of all, IMO.....
Horseplayers will never agree on what it "takes to become profitable". We all have our own opinions...and that's how it should be. The only "truth" that I've ever been able to discover in this game is that we are all different...and the same playing style isn't suitable for all of us. My playing style would probably drive you crazy...and yours would probably do the same to me.

[ReplayRandall]

My playing style would probably drive you crazy...and yours would probably do the same to me.

You drive me crazy NOW, I can't imagine what it would be like if the two of us handicapped/bet together... :cool:

[thaskalos]

You drive me crazy NOW, I can't imagine what it would be like if the two of us handicapped/bet together... :cool:

It might get physical... :)

[ReplayRandall]

It might get physical... :)

As long as we win, what's a few bumps and bruises.... ;)

[Cratos]

This is the biggest challenge I face with figure making. Projecting figures for lightly race two and three year old horses is tough. The easy answer is to just base track variants on older, more established horses. But in today's game, that is just not possible most of the time.

It isn't rare to see a card with more than half being maiden races. The best tracks are running more and more turf races. Tracks often change during the card so you can't just assume the track is the same when the older horses and younger horses run. And, most tracks are running less races with smaller fields which gives a smaller sample size.

When it comes to track to track adjustments (as well as distance to distance) though, it should be fairly easy to stick to the older horses and ignore the young ones if you have a big enough data set.
In reading your posts you have referred to using projections in your speed figure assessments and I am not understanding how you are using projections.

As I understand projections from academic training and applied use, the idea behind making future projections is based on the fact that many real-world random processes are considered to be stationary, meaning that the statistical characteristics (such as mean, variance) of these processes do not change over time. This will enables you to use the theoretical distribution based on historical data to predict future events.

However there are also certain real-world processes that are non-stationary, and using past data generated by those random processes to make future projections can be incorrect.

[Cratos]

Mr. PA, I don't believe that you got the memo. "Handicapper's Corner" is now "Capper Al's Corner". Do you suppose that there's a correlation between the content and the popularity, or lack thereof?
Hi Magister Ludi I am glad to see that you are still posting here; are you still in contact with “TrifectaMike”?

[Cratos]

That has been a decades long nightmare for me.

Here's another complication.

There is a difference between a lightly raced well bred horse from a top barn moving up in class off a win and the run of the mill horse moving up. The lightly raced one often has more in the tank than it has shown to date and will run FASTER when pressed against better (but only up until a point) while the cheaper one with the exact same figures and running lines will go backwards.

Finally, I have statistical evidence that if you use speed figures alone or class figures alone you may not pick as many top choice winners as if you use both, but the value can deteriorate when you combine them. You have to be really careful about combining 2 factors even when they both outperform the track take independently. The combination sometimes creates WORSE value.

Of course making more accurate odds lines helps, but you can really screw yourself up by adding too many ingredients to the stew.
Are you suggesting that there is interdependence between speed figures and class figures or is this just serendipity?

[classhandicapper]

Again, as I have mentioned before, your absolutely (as Thask's as well) correct points are not (directly) connected to the topic of the thread...

If I told you what I think about this subject it would be of little help.

I think there are such wide differences between surfaces that figures that try to equalize them often have very little to do with how fast the horses are actually running. They may help us equate performances, but they aren't reality.

I even think that some dirt surfaces are different enough from each other (or from day to day) that the relationships you are looking for break down.

[raybo]

The idea of "equating" the different surfaces and distances relies on the assumption that the horses are being entered to run with the same INTENTION, across all surfaces and distances. And this isn't the impression that I've gotten during my long tenure in the game. The game may appear at first glance to be a mathematical endeavor...but the horses are actually handled by humans who harbor different motives and expectations as the distances and the surfaces change. The turf runner is usually not "well meant" when he is entered on the dirt, nor is the router entered in a sprint with the expectation of winning the race. Trying to come up with an "adjustment" to smooth out the differences between the performances across the different surfaces and distances becomes a futile pursuit when the horses who are changing distance or surface are doing so with motives other than winning in mind.

A pet project of mine had been to try to come up with a suitable adjustment to smooth out the differences in the performances of the horses who were moving up and down the class ladder. It occured to me that such an adjustment would be highly beneficial to me in my pursuit to properly evaluate a horse's "form". But my efforts towards that end were stymied by the realization that the horses who were rising in class were not all doing it with the same intention in mind. Some class risers gave a good account of themselves at the new class level...while others underperformed miserably on the class rise...only to return to their prior good form on the subsequent class drop. And, while some class droppers ran better on the drop, others were "damaged goods"...and were entered not to try and win the race...but to become someone ELSE'S "damaged goods". A numerical adjustment to smooth-out the vast differences of performance in these cases not only appeared difficult to me...it seemed downright unrealistic. And mighty dangerous to the wallet.

As much as I enjoy making adjustments in this game...I must draw the line when I face "apple-and-orange" comparisons. Some aspects of this game must be analyzed separately...IMO.

That's the main reason that I filter pacelines by sprint and route, dirt and turf. If today is a dirt sprint then that's what I want to use to compare the field, and vice versa, if today is a turf race I don't want to use any dirt races in the analysis, unless forced to do so. If today's race was supposed to be turf, but has been taken off the turf, then I simply click a button named "Take Off Turf" which re-filters the PPs for dirt pacelines. I can of course toggle back and forth in cases where some horses only turf PPs, and some only have dirt PPs, while some have both.

[classhandicapper]

Are you suggesting that there is interdependence between speed figures and class figures or is this just serendipity?

I think they measure some of the same things in different ways and some different things, but the tote board reflects a kind of consensus view.

[steveb]

I am doing something similar. There is one monkey wrench that gets in the way: the improving 3 year old. Beyer talks about how many points a 3 year old will improve, especially earlier in the year. I have a fudge factor to account for this. Also the differences between tracks seems to change a little bit as the meets progress.

i don't know how they program race types for 3yo's in your country, but in australia there are a variety of them

you can mostly solve the improvement month to month by regression analysis.
use the open age races of the various kinds that you have the most instances of for comparison, and see how the 3yo speeds measure up against them.

over here you would find that races for maiden 3yo's have no improvement during the course of a racing season and in fact regress late in their 3yo season.

the unrestricted races for 3yo's tend to improve about 5.5 to 6kg during the season, with nearly all of it happening over the first 4 months particularly for the sprinters.

the other types all fit somewhere between those 2 extremes.

and of course it is not lineal, so you need a transformation when running the regression.

[Cratos]

I think they measure some of the same things in different ways and some different things, but the tote board reflects a kind of consensus view.
In my opinion neither measure speed; both attempt to measure “class” using timefigures.

[cj]

Hi Magister Ludi I am glad to see that you are still posting here; are you still in contact with “TrifectaMike”?


PM this kind of stuff please.

[Cratos]

That's the main reason that I filter pacelines by sprint and route, dirt and turf. If today is a dirt sprint then that's what I want to use to compare the field, and vice versa, if today is a turf race I don't want to use any dirt races in the analysis, unless forced to do so. If today's race was supposed to be turf, but has been taken off the turf, then I simply click a button named "Take Off Turf" which re-filters the PPs for dirt pacelines. I can of course toggle back and forth in cases where some horses only turf PPs, and some only have dirt PPs, while some have both.
That is a very good method to adhere too in statistical analysis because it minimizes extraneous noise and subjectivity.

[steveb]

Yes, that is what I tried to address in the last paragraph. But handicapping and figure making are linked whether we like it or not. You can't be a good figure make if you aren't a good handicapper in my opinion.

Further, I think handicapping is needed for your endeavor as well which is a HUGE part of figure making. You can't figure out the speed difference between tracks and distances if you aren't able to properly handicap the quality of the horses running on each track.

my question would be....how could you know the quality of horses at any particular track, if you don't already know the speed differences between those tracks?

if i had a group of tracks i wanted to learn about, then the only thing i would need is data with the fields and format i required.

then you would easily figure the difference between the inherent speed of the tracks/distances/surfaces and the quality of the horses of those tracks.....all found at the one time, because they all depend on other.
imo you can't find one and then find the other, you need to solve them all at once.

[DeltaLover]

If I told you what I think about this subject it would be of little help.


Does this mean that you do not use any kind of speed figures (as far as I know, all the figures are using some kind of track to track adjustments)?

[DeltaLover]

then you would easily figure the difference between the inherent speed of the tracks/distances/surfaces and the quality of the horses of those tracks.....all found at the one time, because they all depend on other.
imo you can't find one and then find the other, you need to solve them all at once.

The problem is not as easy as you think since we not only have many different distances, track geometries and surfaces and we need a specific methodology to select the common horses to use but because we also have a huge different number of paths that are connecting any pair of tracks (the latter is the topic of this thread)

[steveb]

The problem is not as easy as you think since we not only have many different distances, track geometries and surfaces and we need a specific methodology to select the common horses to use but because we also have a huge different number of paths that are connecting any pair of tracks (the latter is the topic of this thread)

actually it IS that simple, as i have already been doing it for 30 years or more.
you just have not thought of it, and as far as i know nobody has.


i don't know why you would think australia does not have the same issues....or england/europe in general, or south africa or new zealand or japan, etc etc.

i will admit that i don't know anything much about usa, but their courses as a whole, are much more similar in layout, than tracks in most other countries, so it would probably be easier in fact.

the biggest problem with my way is that the data needs to be clean, although the odd mistake will be drowned out if there is sufficient data.

[classhandicapper]

Does this mean that you do not use any kind of speed figures (as far as I know, all the figures are using some kind of track to track adjustments)?


This is my current thinking in a long winded nutshell.

Final times are the result of the quality of the horses, the race development (pace, competitive nature of race, positions of the horses etc..), the speed of the racetrack, how tiring the racetrack is, and how the jockeys adjust their aggressiveness to the characteristics of that track.

On top of that you have issues related to run ups, rails, wind, track speed change etc..

Even if you could assume perfect figure charts and incredible figure making skills (which you cannot) you'd still get fluctuations of times of a few fifths in either direction for the same horses based on those race development issues.

IMO, between the complexity and differing methodologies, speed figures are best used as a ballpark estimate of what happened.

Now for Class.

The class structure at any track will be relatively efficient because winners tend to move up in class searching for bigger purses and badly beaten horses move down until they all reach their appropriate level.

What makes "classing" horses difficult is learning that pecking order, identifying strong and weak fields within a class, dealing with very lightly raced horses that haven't sorted themselves out yet, dealing with shippers, and just plain knowing what kind of performance equals what (is a 3rd at class A better than a 2nd at class B etc…)

Both approaches will pick a similar number of winners if you are skilled, but both have strengths and weaknesses.

My goal has long been to combine them and use each of their strengths while simultaneously avoiding the weaknesses. That’s a lot easier said than done because it leads to mental conflict. When they disagree, you aren’t always sure which is right or what the fair odds should be. When they agree, the prices are unattractive.

Now, instead of trying to use figures to help me fill the holes in my classing approach, I am searching for better methods of filling those holes using different classing techniques. I'm heavily into "data", "research" and "stats" mode.

I don’t think there is a right or wrong way. It’s about value. I’m trying to create ways of measuring things that other people will not have. There’s no reason you can’t do that with “times” too. It’s just that the competition seems to be a lot deeper in that area and it’s a more popular approach among handicappers.

[cj]

my question would be....how could you know the quality of horses at any particular track, if you don't already know the speed differences between those tracks?

if i had a group of tracks i wanted to learn about, then the only thing i would need is data with the fields and format i required.

then you would easily figure the difference between the inherent speed of the tracks/distances/surfaces and the quality of the horses of those tracks.....all found at the one time, because they all depend on other.
imo you can't find one and then find the other, you need to solve them all at once.

It is a chicken and egg type thing, which is why the old time figure makers started with pars and eventually moved to projections. But were the track to track comparison, for example all 10k claimers set to equal, really accurate? Almost assuredly not. I think it takes time. You start with a few assumptions and do the best you can, but then you build in checks to see where you are wrong and where you are right and perfect it over time.

[Capper Al]

I am currently revisiting the implementation of my speed figures, focusing on a topic that I find very interesting and sufficiently complex both from the data collection process scope of view and the optimal algorithm to use as well.

The problem lies in the creation an optimal intra track-surface adjustments, which I believe is very poorly addressed in the existing bibliography. Currently I have a few ideas of how to attack the problem although I would like to hear from others about it.

What is your solution?

Hoping to prove the old model of parellel time plus variant to be obsolete soon with the testing of AMS or AMS EuroSpd figures.

[raybo]

Hoping to prove the old model of parellel time plus variant to be obsolete soon with the testing of AMS or AMS EuroSpd figures.

As far as speed figure creation goes, I see nothing wrong with parallel time charts plus variant. The key is in the adjustments made before the time charts are used. If one is trying to use something to represent times, across distances, surfaces, and tracks, etc., then after adjustments to the raw times and variants, the real work is already done, IMO. The goal is to arrive at a figure that is equal, by distance, surface, and track, etc.. An 80 at a specific distance, surface, track, on a specific day and time period on the card, should be equal to an 80 at any other distance, surface, track, day, and time period on any other card. So, the 80 figure in a time chart is not the problem, it's how you adjust everything before you ever correlate the adjusted time to the figure.

Why would time charts be obsolete? The chart figure is just the representation of all previous adjustments to raw time, isn't it? What is obsolete is the raw timing methodology and lack of other technology required to make the other adjustments to time.

[classhandicapper]

If I was going to try to create time based charts from scratch I would start by studying the most consistent and versatile horses I could find based on PPs alone. That includes shippers for the circuit to circuit comparisons.

A major problem with shippers though is that there are different drug rules from circuit to circuit. So if a bunch of of horses ship from track A to B and run poorly (or vice versa) you can't always be sure if you need a track adjustment or it's a drug/treatment issue.

[Capper Al]

Variants are the obstacle. This all started because weren't available speed figs for Euros. I have done well with BRIS pace and speed figs for North American horses. I'm now seeing a possibility to do better than BRIS.

[cj]

I'm now seeing a possibility to do better than BRIS.

You should set your goal much higher than that.

[Capper Al]

You should set your goal much higher than that.


If you have any comparison between BRIS, Beyer, Jcapper, and TFUS please post them.

[cj]

If you have any comparison between BRIS, Beyer, Jcapper, and TFUS please post them.

All I said was you can do better than BRIS. I didn't mention Beyer or Jcapper. What does the above I quoted have to do with my post?

[Capper Al]

If you have any comparison between BRIS, Beyer, Jcapper, and TFUS please post them.

I find BRIS speed a pretty good standard. It might be the best speed rating out there from what I have read.

[cj]

I find BRIS speed a pretty good standard. It might be the best speed rating out there from what I have read.

OK, fair enough. I found the speed ratings ok, the pace ratings terrible, but that was years ago. Maybe they have gotten better.

[Capper Al]

CJ,

Anyone in the business, like you are with TFUS, should know how their speed ratings stack up with their competition. It's your business to do so. Having said that, I don't really expect any vendor to start releasing their figures because business interest would dictate not to. So I'm sorry for bringing this up. You have been more than fair with me. But my study goes on, and I'll continue to post my findings.

Between BRIS and myself, I think BRIS will win this round. I have an update coming out shortly that I hang my hopes on. Please note- that my speed figs, as they currently are, are good enough to use in absence of traditional speed figs such as BRIS or Beyer for races with Euro horses. That was their original purpose.

[cj]

CJ,

Anyone in the business, like you are with TFUS, should know how their speed ratings stack up with their competition. It's your business to do so. Having said that, I don't really expect any vendor to start releasing their figures because business interest would dictate not to. So I'm sorry for bringing this up. You have been more than fair with me. But my study goes on, and I'll continue to post my findings.

Between BRIS and myself, I think BRIS will win this round. I have an update coming out shortly that I hang my hopes on. Please note- that my speed figs, as they currently are, are good enough to use in absence of traditional speed figs such as BRIS or Beyer for races with Euro horses. That was their original purpose.

If that information was available I would know, but I'm not about to buy DRF and BRIS subscriptions, let alone Thorograph, to find out. But I have a lot of experience with both and have had unlimited subscriptions to both as well. That is how I got started with making my own figures which is what I was getting at for you.

Good luck!

[thaskalos]

OK, fair enough. I found the speed ratings ok, the pace ratings terrible, but that was years ago. Maybe they have gotten better.
They haven't gotten any better...IMO. The pace ratings are still laughable.

[Capper Al]

If that information was available I would know, but I'm not about to buy DRF and BRIS subscriptions, let alone Thorograph, to find out. But I have a lot of experience with both and have had unlimited subscriptions to both as well. That is how I got started with making my own figures which is what I was getting at for you.

Good luck!

Thanks. I know that you have been around the block a few times with these figs.

[Tom]

Anyone in the business, like you are with TFUS, should know how their speed ratings stack up with their competition. It's your business to do so. Having said that, I don't really expect any vendor to start releasing their figures because business interest would dictate not to. So I'm sorry for bringing this up. You have been more than fair with me. But my study goes on, and I'll continue to post my findings.

Will your study include TF, Beyer, EB comparisons. If not, how will you know if your results are optimal?

[Capper Al]

Will your study include TF, Beyer, EB comparisons. If not, how will you know if your results are optimal?

I won't know. All I'll have is a guess, when the time comes, on how my numbers might compare to BRIS and Beyer.

[DeltaLover]

So, nobody wants to talk about the specifics of cross track adjustments? (meaning data/algorithms instead of opinions/beliefs)..

[Tom]

I am happy with the current way it is done. Certainly not as you trivialize it to be.

Looks like the ball's in your court...........

[Capper Al]

So, nobody wants to talk about the specifics of cross track adjustments? (meaning data/algorithms instead of opinions/beliefs)..

Delta,

What are you looking for, parallel chart + track variant + daily variant? (And some might add additional distance travelled or wind resistance.)

[DeltaLover]

Delta,

What are you looking for, parallel chart + track variant + daily variant? (And some might add additional distance travelled or wind resistance.)

In your definition, what I am looking for is the creation of the parallel chart (assuming that it contains the adjustments from track to track)

[no breathalyzer]

I find Equibase speed ratings work just fine

[DeltaLover]

I find Equibase speed ratings work just fine

Good for you but your comment is out of topic

[thaskalos]

Good for you but your comment is out of topic
If you weren't so strict about keeping all the replies in your thread "on topic", maybe you'd get more replies...and you wouldn't get so aggravated because your thread is being "ignored".

[classhandicapper]

So, nobody wants to talk about the specifics of cross track adjustments? (meaning data/algorithms instead of opinions/beliefs)..

Honestly, I would start with someone else's figures that I trust has done plenty of high quality research already. If you start from scratch and just use raw times there's going to be such wild variance and small samples you won't know what's going on for years. There is no point in rebuilding the mousetrap. Just try to make a better one.

Let's assume you had access to Beyer figures in a database. You could slowly back into his "current" parallel time charts for each track (they have changed a lot since publication). Then once you had them, you could look for ways to improve on them. You could look at the distance to distance relationships, distance to distance relationships at the extremes of class, track to track relationships, surface to surface relationships etc... by checking the results and tweaking. Then you could start making your own variants etc.

It doesn't have to be Beyer figures. It could be your personal preference.

[classhandicapper]

Hopefully this link works. It's a section from Beyer's book that deals with track to track.

https://books.google.com/books?id=3jXCQlsfmn4C&pg=PA129&lpg=PA129&dq=claiming+10,000++par&source=bl&ots=3bP8q3e_7J&sig=RCky27jBeU9g9Lc8VR9JvXVLwHw&hl=en&sa=X&ei=kNfKVICYNKaMsQTHsIGICA&ved=0CDwQ6AEwBQ#v=onepage&q&f=false

[DeltaLover]

If you weren't so strict about keeping all the replies in your thread "on topic", maybe you'd get more replies...and you wouldn't get so aggravated because your thread is being "ignored".

Replies that are based on "I think" ,"I believe" etc are of no value for the topic of this thread Thask and they only pollute the namespace with irrelevant noise. What I am asking here is very concrete: data, methodologies and nullable hypothesis that can be tested and proven correct based on objective criteria

[DeltaLover]

Hopefully this link works. It's a section from Beyer's book that deals with track to track.

https://books.google.com/books?id=3jXCQlsfmn4C&pg=PA129&lpg=PA129&dq=claiming+10,000++par&source=bl&ots=3bP8q3e_7J&sig=RCky27jBeU9g9Lc8VR9JvXVLwHw&hl=en&sa=X&ei=kNfKVICYNKaMsQTHsIGICA&ved=0CDwQ6AEwBQ#v=onepage&q&f=false

Of course I am aware about this but I have already clarified that I find it very naive and dated. Today we can do much better..

[classhandicapper]

Of course I am aware about this but I have already clarified that I find it very naive and dated. Today we can do much better..

I'm not sure what you can do to improve on this method other than making the figures themselves more accurate to start (in ways I suggested in the other post) because looking at figure comparisons from track to track among shippers, win%, and ROI (which represent reality) is better than any theory I can think of.

[thaskalos]

Of course I am aware about this but I have already clarified that I find it very naive and dated. Today we can do much better..
And if we can do much better today...we should publicize our work for all to see? What is the benefit of divulging important information such as this to people with whom we are in direct competition? You say that you are not looking for "I think", or "I believe"...you want "very concrete methodologies". You honestly expect people to share their "very concrete methodologies" with you?

"What is your solution?"...you ask in your opening post here. And you seriously expect someone to supply you with a direct and "very concrete" answer to this question? This is a COMPETITION, DeltaLover...where we are trying to win each other's MONEY. We may share opinions and beliefs...but our "very concrete" findings are retained for our own private use. Making a buck out there is hard enough already...IMO.

[cj]

In your definition, what I am looking for is the creation of the parallel chart (assuming that it contains the adjustments from track to track)

I use different parallel charts for different configurations. I one turn route is different than a two turn route, a mile track different than a 6 furlong track, etc. But why can't the rest just be handled by the track variant?

[ReplayRandall]

And if we can do much better today...we should publicize our work for all to see? What is the benefit of divulging important information such as this to people with whom we are in direct competition? You say that you are not looking for "I think", or "I believe"...you want "very concrete methodologies". You honestly expect people to share their "very concrete methodologies" with you?

"What is your solution?"...you ask in your opening post here. And you seriously expect someone to supply you with a direct and "very concrete" answer to this question? This is a COMPETITION, DeltaLover...where we are trying to win each other's MONEY. We may share opinions and beliefs...but our "very concrete" findings are retained for our own private use. Making a buck out there is hard enough already...IMO.

Delta, please don't be offended, but what Thask has told you here is reality. Conspicuous by their absence to this thread, are 2 PA members who probably have what your looking for. However, they both won't contribute for free, for obvious reasons. You can PM them and inquire,( For a FEE they might help, who knows?),----> Dave Schwartz and JCapper.....

[DeltaLover]

I use different parallel charts for different configurations. I one turn route is different than a two turn route, a mile track different than a 6 furlong track, etc. But why can't the rest just be handled by the track variant?

My understanding is that track variants adjusts times for the same track - surface while here we are talking about cross track adjustments

[cj]

My understanding is that track variants adjusts times for the same track - surface while here we are talking about cross track adjustments

I'm saying if you get the variant right, it won't matter.

[classhandicapper]

My understanding is that track variants adjusts times for the same track - surface while here we are talking about cross track adjustments

It's 6 to one and 1/2 dozen to another.

It doesn't matter if the adjustment is done at the track level or track variant level as long as the final number that results is equating performances correctly.

That's why I am saying that what Beyer is doing is fine.

Maybe 6F times are slower at KEE than they are at SA and Beyer has 6F in 1:10 set as equal in his charts for both tracks, but the track variants will adjust for that. He can cross check the accuracy by monitoring horses that ship back and forth to ensure the final figures are being consistent.

What you really have to do is get the distance relationships at the same track correct because those can vary from track to track.

[cj]

It's 6 to one and 1/2 dozen to another.

It doesn't matter if the adjustment is done at the track level or track variant level as long as the final number that results is equating performances correctly.

That's why I am saying that what Beyer is doing is fine.

Maybe 6F times are slower at KEE than they are at SA and Beyer has 6F in 1:10 set as equal in his charts for both tracks, but the track variants will adjust for that. He can cross check the accuracy by monitoring horses that ship back and forth to ensure the final figures are being consistent.

What you really have to do is get the distance relationships at the same track correct because those can vary from track to track.

I said the same thing in like 4% of the words :)

[ReplayRandall]

I use different parallel charts for different configurations. I one turn route is different than a two turn route, a mile track different than a 6 furlong track, etc. But why can't the rest just be handled by the track variant?

I find this to be true for lower class/purse horses, but for higher class/purse, I find the track-to-track pars cause the variants to skew dramatically. Don't you have the same problem?

[cj]

I find this to be true for lower class/purse horses, but for higher class/purse, I find the track-to-track pars cause the variants to skew dramatically. Don't you have the same problem?

Sorry, I don't really understand what you are trying to say. I don't use track to track pars. I make the figures for tracks and then make sure as horses move around the country they are lined up.

[Tom]

Of course I am aware about this but I have already clarified that I find it very naive and dated. Today we can do much better..

Well, then stop asking and start talking about your ideas.
YOU started the thread.
Did you ever intend to take part it in or not?

Replies that are based on "I think" ,"I believe" etc are of no value for the topic


And yet that is all you have offered so far - you think there is a better way.

I will offer the premise - No, there is no better way than currently is used - no one can improve upon it. Prove me wrong.

[thaskalos]

Sorry, I don't really understand what you are trying to say. I don't use track to track pars. I make the figures for tracks and then make sure as horses move around the country they are lined up.
Do you find that the majority of the horses are able to reproduce their best figures across different tracks...or are these high figures "track specific'?

[ReplayRandall]

Sorry, I don't really understand what you are trying to say. I don't use track to track pars. I make the figures for tracks and then make sure as horses move around the country they are lined up.

Lined up by what criteria? What adjustments do you make to line them up?

To add: What % of subjectivity (your opinion) goes into your figures?

[cj]

Lined up by what criteria? What adjustments do you make to line them up?

To add: What % of subjectivity (your opinion) goes into your figures?

If horses that run at Track A and often ship to Track B (a different circuit) and run about 5 points slower at Track B, I'll reevaluate the figures for both. It is like a spider web, you have to piece it all together. But once you've been doing this a while it doesn't take a lot of work. In the beginning, it takes a ton of work.

The second part, I have no idea how to answer that. I try to use as little subjectivity as possible. But some races and some days leave you no choice or you'll make a lot of figures that look silly in the near future.

[cj]

Do you find that the majority of the horses are able to reproduce their best figures across different tracks...or are these high figures "track specific'?


Yes, most are able to, but not all of course. A lot depends on horse and trainer. And as we all know, not all jurisdictions have similar drug policies. Some horses need certain drugs more than others and probably have no shot when shipped to a place not as liberal with drug use. But really, there is no way for me to begin to try build that into speed figures.

[Cratos]

I find this to be true for lower class/purse horses, but for higher class/purse, I find the track-to-track pars cause the variants to skew dramatically. Don't you have the same problem?
Figure making for horseracing handicapping is primarily if not exclusively an anecdotal based postulation and to get what DL is requesting the answer lies in an empirical analysis as put forth in post #2 in this thread by the poster, “Raybo.”

Additionally for the poster who asked if pace change by a small amount, will final time change?

The answer is yes because pace correctly stated is acceleration and time over distance is a function of acceleration: Newton’s laws of motion proved that nearly 200 years ago.

[DeltaLover]

If horses that run at Track A and often ship to Track B (a different circuit) and run about 5 points slower at Track B, I'll reevaluate the figures for both. It is like a spider web, you have to piece it all together. But once you've been doing this a while it doesn't take a lot of work. In the beginning, it takes a ton of work.

The second part, I have no idea how to answer that. I try to use as little subjectivity as possible. But some races and some days leave you no choice or you'll make a lot of figures that look silly in the near future.

Good post..

Exactly this "spider web" is where I am referring in this thread! Optimizing the the weights of the "web" in such a way to minimize discrepancies is the way to go and this is exactly what my solution to the problem consists of.

[DeltaLover]

Figure making for horseracing handicapping is primarily if not exclusively an anecdotal based postulation and to get what DL is requesting the answer lies in an empirical analysis as put forth in post #2 in this thread by the poster, “Raybo.”


No, i do not agree... It certainly is not empirical analysis (especaially anything that relies on mechanics and physics in general) but optimization algorithms based on discrete math that solve the problem optimally

[steveb]

Well, then stop asking and start talking about your ideas.
YOU started the thread.
Did you ever intend to take part it in or not?



And yet that is all you have offered so far - you think there is a better way.

I will offer the premise - No, there is no better way than currently is used - no one can improve upon it. Prove me wrong.


word for word what i was thinking.
but there IS a better way.
it's called iteration by trial and error, with each iteration lessening the error until it is no more.

he ignored mine, and i am probably the only guy that can tell him how to do it accurately, very quickly(a minute or so after code written for however many tracks desired) and completely non-subjectively, provided one has the (accurate)data with required fields.
and the coding knowledge.

maybe mine is trivial though and i am just too naive!!:lol:

[DeltaLover]

word for word what i was thinking.
but there IS a better way.
it's called iteration by trial and error, with each iteration lessening the error until it is no more.

he ignored mine, and i am probably the only guy that can tell him how to do it accurately, very quickly(a minute or so after code written for however many tracks desired) and completely non-subjectively, provided one has the (accurate)data with fields.
and the coding knowledge.

maybe mine is trivial though and i am just too naive!!:lol:

Obviously there will be some sort of an iteration. The problem is the path complexity is N! which translates to a huge number of paths that is impossible to handle sequentially.. The question is how you represent the paths and how what algorithms you use.. Can you answer providing an example?

[Cratos]

No, i do not agree... It certainly is not empirical analysis (especaially anything that relies on mechanics and physics in general) but optimization algorithms based on discrete math that solve the problem optimally
Your response is circumventive, but I am not going to debate the fallacy of your assumption because it would detract away from the answer that you are attempting to garner.

However what you want to measure is not theoretical; it is real and tangible with science and math being the correct tools to quantify your solution.

The major weakness in thoroughbred racing handicapping today is its inability to embrace and take advantage of new technologies, but in 2015 the handicapper is using a flawed speedfigure methodology which is supported by the innocuous beaten length metric.

[cj]

Your response is circumventive, but I am not going to debate the fallacy of your assumption because it would detract away from the answer that you are attempting to garner.

However what you want to measure is not theoretical; it is real and tangible with science and math being the correct tools to quantify your solution.

The major weakness in thoroughbred racing handicapping today is its inability to embrace and take advantage of new technologies, but in 2015 the handicapper is using a flawed speedfigure methodology which is supported by the innocuous beaten length metric.

There is no flawed beaten lengths metric at the finish. Pace calls, absolutely.

[DeltaLover]

Your response is circumventive, but I am not going to debate the fallacy of your assumption because it would detract away from the answer that you are attempting to garner.

However what you want to measure is not theoretical; it is real and tangible with science and math being the correct tools to quantify your solution.

The major weakness in thoroughbred racing handicapping today is its inability to embrace and take advantage of new technologies, but in 2015 the handicapper is using a flawed speedfigure methodology which is supported by the innocuous beaten length metric.

I disagree with your approach but this is not the point here.. Please let's focus on the topic of the discussion... If you have any data , procedures or algorithms that we can test and possibly confirm or reject please go ahead.

[steveb]

Obviously there will be some sort of an iteration. The problem is the path complexity is N! which translates to a huge number of paths that is impossible to handle sequentially.. The question is how you represent the paths and how what algorithms you use.. Can you answer providing an example?

no, it's your topic as you keep reminding us.
YOU need to lead the way.
if i can see something of value,that appeals to me, then......
if i see fit i may comment, or critique.

but i am in the box seat as i already know, and the only other people that know(via me), like the late entropy syndicate and another bigger one still, are not going to tell you.
but i am free to do as i wish, as i refused to sign NDA's as they were my ideas and not theirs.

oh, and nothing is impossible if you are determined and keep persisting.
too many people get tied up in complexities, and they talk the talk, but they can't walk the walk.

[cj]

Your response is circumventive, but I am not going to debate the fallacy of your assumption because it would detract away from the answer that you are attempting to garner.

However what you want to measure is not theoretical; it is real and tangible with science and math being the correct tools to quantify your solution.

The major weakness in thoroughbred racing handicapping today is its inability to embrace and take advantage of new technologies, but in 2015 the handicapper is using a flawed speedfigure methodology which is supported by the innocuous beaten length metric.

It does no good to tell everyone how they are wrong if you aren't going to tell them what is right. No more of these posts if you are only going to say the same thing you've said many times without any substance. I'll simply delete them.

[Capper Al]

Delta,

All I can say is that you must use your imagination and test the outcomes.

[Cratos]

It does no good to tell everyone how they are wrong if you aren't going to tell them what is right. No more of these posts if you are only going to say the same thing you've said many times without any substance. I'll simply delete them.
I didn't say anyone was wrong, I stated a different point of view and intellectually that is what debate is about.

You appear to be cynical when speedfigures are discussed. My handicapping knowledge coupled with my quantitative intellect in math and science is strong enough to understand when the content is wrong.

Delete whatever you like, that is your own insecurity; I am not intimidated.

[cj]

I didn't say anyone was wrong, I stated a different point of view and intellectually that is what debate is about.

You appear to be cynical when speedfigures are discussed. My handicapping knowledge coupled with my quantitative intellect in math and science is strong enough to understand when the content is wrong.

Delete whatever you like, that is your own insecurity; I am not intimidated.

Nobody is trying to intimidate you. Cynical? Please. I love discussing speed figures. But you constantly come in and say the things over and over again and derail threads. Thus, the delete warning. That is the end of our discussion in this thread.

[raybo]

My understanding is that track variants adjusts times for the same track - surface while here we are talking about cross track adjustments

The longer this thread gets, the more confused I get. In your first post in this thread, you specify "intra" track adjustments, meaning at the same track. But, later you mention that you are assuming that there already exists a reliable method for adjustments, within the same track ("intra"). And, in the quoted post, you say the thread is about "cross" track adjustments ("inter" track, meaning from track to track, not within the same track).

Do you already have a reliable "intra" track adjustment method? If so, then you should have a speed figure conversion chart for each track that could possibly be involved in any movement by horses from one track to another.

If you already know how to adjust times at the same track for equalization of surfaces, distances, etc., then the best way for adjusting speed figures from track to track, IMO, would be to have a fully adjusted time to speed figure chart, for each and every track.

For example, a 6f dirt fast surface, with track and time of year normal temp, air, surface, wind, etc., that resulted in an adjusted time of 1:12 at track A, equates to an 80 speed figure, but at track B, under track and time of year normal factors, that same fully adjusted 6f time of 1:12 equates to a 78 speed figure, because you're using a different time to speed figure chart at track B than the one used at track A. Multiple tracks equals multiple charts, each chart specific to a particular track. So the difference between track A and track B, at 6f on the dirt under normal racing conditions would be 2 speed points.

[DeltaLover]

The longer this thread gets, the more confused I get. In your first post in this thread, you specify "intra" track adjustments, meaning at the same track. But, later you mention that you are assuming that there already exists a reliable method for adjustments, within the same track ("intra"). And, in the quoted post, you say the thread is about "cross" track adjustments ("inter" track, meaning from track to track, not within the same track).

Do you already have a reliable "intra" track adjustment method? If so, then you should have a speed figure conversion chart for each track that could possibly be involved in any movement by horses from one track to another.

If you already know how to adjust times at the same track for equalization of surfaces, distances, etc., then the best way for adjusting speed figures from track to track, IMO, would be to have a fully adjusted time to speed figure chart, for each and every track.

For example, a 6f dirt fast surface, with track and time of year normal temp, air, surface, wind, etc., that resulted in an adjusted time of 1:12 at track A, equates to an 80 speed figure, but at track B, under track and time of year normal factors, that same fully adjusted 6f time of 1:12 equates to a 78 speed figure, because you're using a different time to speed figure chart at track B than the one used at track A. Multiple tracks equals multiple charts, each chart specific to a particular track. So the difference between track A and track B, at 6f on the dirt under normal racing conditions would be 2 speed points.

How you treat a case where there is no direct connection of two specific tracks? For example you might not have enough data for Santa Rosa to Belmont, in this case how you are making the adjustments? More than this how you treat longer paths connecting the same tracks (meaning that you have enough data from AQU -> BEL but also you have AQU -> PRX and PRX -> BEL that contradict the direct connection)

[steveb]

For example, a 6f dirt fast surface, with track and time of year normal temp, air, surface, wind, etc., that resulted in an adjusted time of 1:12 at track A, equates to an 80 speed figure, but at track B, under track and time of year normal factors, that same fully adjusted 6f time of 1:12 equates to a 78 speed figure, because you're using a different time to speed figure chart at track B than the one used at track A. Multiple tracks equals multiple charts, each chart specific to a particular track. So the difference between track A and track B, at 6f on the dirt under normal racing conditions would be 2 speed points.

you sir are on the money.
but the proof of the pudding is in how it's done, and how well it works.
you would have to equalise for class too, or you will get nowhere no matter how well you do the other equalising factors.
and that will be the primary problem for most i would imagine.
in fact i would do the whole lot at once.....figure the times, classes, tracks distances, surfaces, whatever.
simply because they all depend on each other.
if you do one first and then get the others, your 'first' has to be correct or it follows that everything derived from the 'first' is wrong too.
but if you are doing trial and error with all those factors at once and learning a little more after each iteration, then at the end you will have the answers to everything, rather than just one, and they all figured each other out.

but i would do it so that an 80 is worth exactly the same where ever it happens to be, because it then becomes completely unambiguous.
that does not mean that a horse that runs an 80 here, should necessarily run an 80 there.
it's easier than saying a 72 seconds is worth 80 points here and 78 there.

[raybo]

How you treat a case where there is no direct connection of two specific tracks? For example you might not have enough data for Santa Rosa to Belmont, in this case how you are making the adjustments? More than this how you treat longer paths connecting the same tracks (meaning that you have enough data from AQU -> BEL but also you have AQU -> PRX and PRX -> BEL that contradict the direct connection)

That is exactly my point. If you already have a reliable method of adjusting times at each track, then the track to track adjustments are right in front of you. I suspect that you do NOT have a reliable intra-track time adjustment method, so it is probably impossible for you to adjust from track to track. if you had a reliable intra-track time adjustment method, that method would produce different speed figures, for the same adjusted times, for tracks that are significantly different. if your intra-track conversions do not produce differing speed figures for tracks that are significantly different, then you are most likely not performing the intra-track adjustments accurately.

[steveb]

How you treat a case where there is no direct connection of two specific tracks? For example you might not have enough data for Santa Rosa to Belmont, in this case how you are making the adjustments? More than this how you treat longer paths connecting the same tracks (meaning that you have enough data from AQU -> BEL but also you have AQU -> PRX and PRX -> BEL that contradict the direct connection)

you don't need ANY data that is common for any 2(or more) tracks.
just because horse 'a' has been there AND there, does not in itself prove anything.
you do need lots of data for each individual track though.

[raybo]

you sir are on the money.
but the proof of the pudding is in how it's done, and how well it works.
you would have to equalise for class too, or you will get nowhere no matter how well you do the other equalising factors.
and that will be the primary problem for most i would imagine.
in fact i would do the whole lot at once.....figure the times, classes, tracks distances, surfaces, whatever.
simply because they all depend on each other.
if you do one first and then get the others, your 'first' has to be correct or it follows that everything derived from the 'first' is wrong too.
but if you are doing trial and error with all those factors at once and learning a little more after each iteration, then at the end you will have the answers to everything, rather than just one, and they all figured each other out.

but i would do it so that an 80 is worth exactly the same where ever it happens to be, because it then becomes completely unambiguous.
that does not mean that a horse that runs an 80 here, should necessarily run an 80 there.
it's easier than saying a 72 seconds is worth 80 points here and 78 there.

I totally agree. I did not mention "class" simply because "class" is an ambiguous term among players. Define "class" unambiguously, from track to track, and you have the key to this thread. Of course, anyone who has "class", from track to track and race to race, all figured out, is NOT going to tell anyone how to do it (not even me, and I'll usually tell people exactly how I do almost everything I do, if they really want to know). There are some things that, if you are a wagerer/investor, you just can't afford to reveal to the public.

[DeltaLover]

you don't need ANY data that is common for any 2(or more) tracks.
just because horse 'a' has been there AND there, does not in itself prove anything.
you do need lots of data for each individual track though.

Sure, I did not mean to use a single horse but many of them matching a certain set of conditions.. In this case how do you proceed?

[raybo]

Sure, I did not mean to use a single horse but many of them matching a certain set of conditions.. In this case how do you proceed?

I believe that CJ answered that for you (yes, this is an opinion and not fact or "empirical evidence" but then almost everything in this game is opinion, even your stuff). His "intra" track adjustments are reliable, so his track to track comparisons are reliable, until they aren't, at which point he makes individual adjustments, after the fact. If you are looking for perfection, through numerical methods, in individual performances, without making adjustments later on down the line, I'm afraid you are fighting a losing battle. Variances in individual performances are normal, but eventually, over time, some of those formerly assumed variances (or lack thereof) are going to be, simply wrong, and require manual adjustments based on smoothing them out so that they make "sense" to the body of work of the individuals involved.

[steveb]

I totally agree. I did not mention "class" simply because "class" is an ambiguous term among players. Define "class" unambiguously, from track to track, and you have the key to this thread. Of course, anyone who has "class", from track to track and race to race, all figured out, is NOT going to tell anyone how to do it (not even me, and I'll usually tell people exactly how I do almost everything I do, if they really want to know). There are some things that, if you are a wagerer/investor, you just can't afford to reveal to the public.

i actually don't have any problems telling what i know, because the investing aspect of it no longer interests me, but the whole puzzle that is handicapping does.

how much money does one person need?
i don't need any more than what i already have, in fact i don't need bugger all of what i already have, as everything i like doing in life is basically free.
i would rather be doing my volunteer conservation work, as i discovered it fulfils me far more than making money on horse racing ever did.
so there are people that have received off me for nothing what entropy and others have paid me many hundreds of K's for.
i just choose who i help though.
i would be surprised if a genuinely smart person could not read my past posts on here, and not figure how to do it.
i have given plenty of clues over a fairly long period of time

you are correct though, what i had for myself for many years, is not worth now what it once was.
never mind, my day is done and dusted, i have no ambitions to have that life again.

[raybo]

i actually don't have any problems telling what i know, because the investing aspect of it no longer interests me, but the whole puzzle that is handicapping does.

how much money does one person need?
i don't need any more than what i already have, in fact i don't need bugger all of what i already have, as everything i like doing in life is basically free.
i would rather be doing my volunteer conservation work, as i discovered it fulfils me far more than making money on horse racing ever did.
so there are people that have received off me for nothing what entropy and others have paid me many hundreds of K's for.
i just choose who i help though.
i would be surprised if a genuinely smart person could not read my past posts on here, and not figure how to do it.
i have given plenty of clues over a fairly long period of time

you are correct though, what i had for myself for many years, is not worth now what it once was.
never mind, my day is done and dusted, i have no ambitions to have that life again.

I understand, unfortunately, most of us did not get paid by others for our "work", rather we used it to make money, and continue to this day. I'm not rich, nor do I have any desire to be, but I do need income to support my retirement, and racing is a very good source of support income. I give, and have given, away lots of stuff to almost anyone (some reserved for obviously serious and open-minded players), but as the old saying goes, "I taught him everything he knows, but not everything I know". :lol:

[thaskalos]

I understand, unfortunately, most of us did not get paid by others for our "work", rather we used it to make money, and continue to this day. I'm not rich, nor do I have any desire to be, but I do need income to support my retirement, and racing is a very good source of support income. I give, and have given, away lots of stuff to almost anyone (some reserved for obviously serious and open-minded players), but as the old saying goes, "I taught him everything he knows, but not everything I know". :lol:

I wouldn't mind becoming rich...but I seem to have a difficult time getting these horses to comply with my wishes often enough...

[steveb]

I understand, unfortunately, most of us did not get paid by others for our "work", rather we used it to make money, and continue to this day. I'm not rich, nor do I have any desire to be, but I do need income to support my retirement, and racing is a very good source of support income. I give, and have given, away lots of stuff to almost anyone (some reserved for obviously serious and open-minded players), but as the old saying goes, "I taught him everything he knows, but not everything I know". :lol:


for most of my adult life i did it on my own.
it was more or less foisted on me as i was once a printer, but knew my life working for others was soon to be over for reasons i don't want to get into.

it is only in this millennium that people had paid me for my knowledge, and that is no more.


i have always struggled to understand why people strive to gamble for a living, when it is infinitely harder and more stressful than any other occupation i can think of.
hobby yes, living no.
especially in these times.
i honestly doubt i could do it again if i had to, because the syndicates have made it so much harder for your average joe to survive.
i am 63 now and i don't know ANY others that do it for a living that are not members of a syndicate.
once i knew about 10 or more that could make it alone but they have all been put out of business by the betting syndicates(or have joined them).

i would have happily remained a tradesman if i could have.
then i would probably have just been the same mug punter as most others are.
it's amazing what necessity does.

[raybo]

for most of my adult life i did it on my own.
it was more or less foisted on me as i was once a printer, but knew my life working for others was soon to be over for reasons i don't want to get into.

it is only in this millennium that people had paid me for my knowledge, and that is no more.


i have always struggled to understand why people strive to gamble for a living, when it is infinitely harder and more stressful than any other occupation i can think of.
hobby yes, living no.
especially in these times.
i honestly doubt i could do it again if i had to, because the syndicates have made it so much harder for your average joe to survive.
i am 63 now and i don't know ANY others that do it for a living that are not members of a syndicate.
once i knew about 10 or more that could make it alone but they have all been put out of business by the betting syndicates(or have joined them).

i would have happily remained a tradesman if i could have.
then i would probably have just been the same mug punter as most others are.
it's amazing what necessity does.

Oh, I don't ever want to do this "for a living", did that. Although I was successful during that short period of time, it was not pleasant, and turned out to be downright harmful to my mental/emotional health. But, as an activity and additional support function it's quite pleasant and rewarding.

My early jobs were in maintenance of military airborne computerized navigation and weapons control systems, which included basic computer operation training (very basic!), and then drafting and design of custom products (millwork). The computer stuff caused me to ask questions about how data from aircraft sensors was used to project altitude, angle of attack, speed, and direction requirements, and how that and other data were used to project data relationships for optimum weapons delivery. The drafting and design stuff required that I be able to "see" the finished product first, in my mind, and then work backwards to find the individual steps and materials/tools required to produce that finished product.

So, when you combine those two disciplines, it's not a giant leap to apply them to horse racing, and especially to creating handicapping tools. Although I didn't "love" the electronics/computer portion, I did find a way to use it in support of what I did love (using historical and projected data as the basis for designing and making things work in regard to horse race analysis and wagering), by combining the two into a "love" for creating handicapping tools and methods. But, I had to learn horse racing first, and that took many years of study. I'm still doing this stuff, because I really like the design and incremental creation process, and I make some money wagering as a result. Best of both worlds.

[DeltaLover]

Oh, I don't ever want to do this "for a living", did that. Although I was successful during that short period of time, it was not pleasant, and turned out to be downright harmful to my mental/emotional health. But, as an activity and additional support function it's quite pleasant and rewarding.

My early jobs were in maintenance of military airborne computerized navigation and weapons control systems, which included basic computer operation training (very basic!), and then drafting and design of custom products (millwork). The computer stuff caused me to ask questions about how data from aircraft sensors was used to project altitude, angle of attack, speed, and direction requirements, and how that and other data were used to project data relationships for optimum weapons delivery. The drafting and design stuff required that I be able to "see" the finished product first, in my mind, and then work backwards to find the individual steps and materials/tools required to produce that finished product.

So, when you combine those two disciplines, it's not a giant leap to apply them to horse racing, and especially to creating handicapping tools. Although I didn't "love" the electronics/computer portion, I did find a way to use it in support of what I did love (using historical and projected data as the basis for designing and making things work in regard to horse race analysis and wagering), by combining the two into a "love" for creating handicapping tools and methods. But, I had to learn horse racing first, and that took many years of study. I'm still doing this stuff, because I really like the design and incremental creation process, and I make some money wagering as a result. Best of both worlds.

WOW!! :bang: :bang: :bang:


May I please ask what these stories have to do with this thread??

With absolutely no intention to offend you or steveb, who really cares about what you are writing here?

Why not try to stick to the topic or simply start a new one if you really think your stories are interesting to some of the participants of the forum....

[raybo]

WOW!! :bang: :bang: :bang:


May I please ask what these stories have to do with this thread??

With absolutely no intention to offend you or steveb, who really cares about what you are writing here?

Why not try to stick to the topic or simply start a new one if you really think your stories are interesting to some of the participants of the forum....

You may think that our "stories" have no value in this thread, and I'm sorry you didn't get anything from them, but sometimes, the answer is right in front of you, all you have to do is recognize it and determine the steps needed to create it. You already have the answer, you just need to figure out how to get from point A to the finished product. Nobody can, or will, do it for you, that probably wouldn't satisfy you anyway, it would just be a mirror of someone else's work, knowledge, and experience, satisfaction comes from within, applying your own research, knowledge, and experience (and a big dose of common sense).

[cj]

WOW!! :bang: :bang: :bang:


May I please ask what these stories have to do with this thread??

With absolutely no intention to offend you or steveb, who really cares about what you are writing here?

Why not try to stick to the topic or simply start a new one if you really think your stories are interesting to some of the participants of the forum....

Delta, threads often go in odd directions. Nothing wrong with trying to get them back on track but the best way to do that is just post something relevant, not trying to control the flow. Been doing this a long time, it doesn't work.

[steveb]

WOW!! :bang: :bang: :bang:


May I please ask what these stories have to do with this thread??

With absolutely no intention to offend you or steveb, who really cares about what you are writing here?

Why not try to stick to the topic or simply start a new one if you really think your stories are interesting to some of the participants of the forum....

thank you, you have guaranteed that you never get help from me.
and no i am not offended in the slightest.

you wanted others to give you information, but without offering anything up yourself.
you have buckleys and none with an attitude like that.
you only want help on YOUR terms, so good luck with that.

[thaskalos]

I too am a professional gambler, with a pretty interesting story to tell...but it is slightly off topic...and I don't want to aggravate my friend DeltaLover any more than he already is. :)

[Capper Al]

Delta,

You can't expect to come up with better speed figs by applying the same standard formula of parallel time + TV + DV. Doing the same thing and expecting different results is the definition of insanity. Besides, us little guys, don't have the resources and the data to compete against the big commercial vendors when applying the standard formula. You must be creative and think outside the box and test the outcomes. Use the force Delta.

[DeltaLover]

Delta,

You can't expect to come up with better speed figs by applying the same standard formula of parallel time + TV + DV. Doing the same thing and expecting different results is the definition of insanity. Besides, us little guys, don't have the resources and the data to compete against the big commercial vendors when applying the standard formula. You must be creative and think outside the box and test the outcomes. Use the force Delta.

I am not using the parallel time + TV + DV formula. Instead I first create "raw figures" in a per track - surface basis, converting the final times of all the races (all distances - surfaces) to a speed figure, then based on these figures I estimate the track variants that can be applied to the "raw figure" to produce a per track figure and finally I use my cross track algorithms to calculate the optimal track to track adjustments assuming one specific track to be the starting point.

What I am trying to discuss here is the last part of this process (not the TV or the distance adjustments). I have solved the problem with several different ways, the most elegant of them been based on a graph representation of the possible connections, where the vertices are the tracks while the edges contain the figure diffs (note that I am not mentioning the methodology to estimate the diffs as they can change in many different ways as we test our handicapping).

The solution is to manipulate the figure diffs (weight of each edge) in an optimal way, trying to minimize the discrepancies when the graph is walked in such as way to visit all the edges. The problem is not trivial, as a brutal force BFS or DFS becomes a very expensive operation that practically never ends and this is where I came to a variation of Tarjan's algorithm that gives the solution in a matter of few seconds when applied to a very complex universe of thousands of edges...

[Tom]

Why not try to stick to the topic or simply start a new one if you really think your stories are interesting to some of the participants of the forum....

You start a thread and then contribute nothing to it. Try moving it along yourself, that will give everyone something to build on. Frankly the thread has been very boring. A few less insults to us all in a passive-aggressive manor might get people to participate.

Do you have a way to somehow equate SA and Pen? If so, put it out there and let us ponder it.

[Capper Al]

I am not using the parallel time + TV + DV formula. Instead I first create "raw figures" in a per track - surface basis, converting the final times of all the races (all distances - surfaces) to a speed figure, then based on these figures I estimate the track variants that can be applied to the "raw figure" to produce a per track figure and finally I use my cross track algorithms to calculate the optimal track to track adjustments assuming one specific track to be the starting point.

What I am trying to discuss here is the last part of this process (not the TV or the distance adjustments). I have solved the problem with several different ways, the most elegant of them been based on a graph representation of the possible connections, where the vertices are the tracks while the edges contain the figure diffs (note that I am not mentioning the methodology to estimate the diffs as they can change in many different ways as we test our handicapping).

The solution is to manipulate the figure diffs (weight of each edge) in an optimal way, trying to minimize the discrepancies when the graph is walked in such as way to visit all the edges. The problem is not trivial, as a brutal force BFS or DFS becomes a very expensive operation that practically never ends and this is where I came to a variation of Tarjan's algorithm that gives the solution in a matter of few seconds when applied to a very complex universe of thousands of edges...

The commercial vendor have deeper pockets and hence more resources than we do. You are not going to beat them at their game. You need to rethink the proposition and come up with an effective but different way of arriving at a solution. MY speed figs aren't Parallel Time + TV + DV, yet I can hang in there with the big boys. Matter of fact, my methods aren't burdened with TV and DV where the big boys have the advantage. If you are going to be the David against Goliath, you'll need your own slingshot. BTW, I have been posting my speed fig research here in PA in the thread 'What's up with the Capper'?

Good Luck!

[classhandicapper]

How you treat a case where there is no direct connection of two specific tracks? For example you might not have enough data for Santa Rosa to Belmont, in this case how you are making the adjustments? More than this how you treat longer paths connecting the same tracks (meaning that you have enough data from AQU -> BEL but also you have AQU -> PRX and PRX -> BEL that contradict the direct connection)

In the cases where you don't have a direct relationship, you can try to use indirect relationships. Maybe you don't have BEL to Santa Rosa, but you might have BEL to SA/DMR and SA/DMR to Santa Rosa. So you can tie them that way.

In the case of conflict like your AQU, BEL, and PRX case you should be positive about AQU and BEL because the sample size would be gigantic.

So the issue is what's going on with PRX. That could be a sample size thing or it could be a trainer intent thing and not a figure issue. I would probably dig through the details manually.

If you insist on doing it in a purely mathematical way instead of looking at the data and analyzing it using your handicapping knowledge and common sense, some sort of regression analysis where the goal is maximizing the results might work, but that's outside my range. As more data comes in you could keep running the analysis and tweaking.

[steveb]

Parallel Time


So just exactly what is 'Parallel Time' may I ask?
An example please?

[cj]

So just exactly what is 'Parallel Time' may I ask?
An example please?

I've always heard that term used to express a chart that attempts to match up times at different distances to numerical ratings.

For example, this chart would be considered a parallel time chart:

http://www.angelfire.com/la2/LouisianaRacing/1turnspeedratings.html

[Capper Al]

CJ has you covered on the parallel time chart.

[steveb]

I've always heard that term used to express a chart that attempts to match up times at different distances to numerical ratings.

For example, this chart would be considered a parallel time chart:

http://www.angelfire.com/la2/LouisianaRacing/1turnspeedratings.html


thank you.
not what i was expecting to see, but the relationships between times & distances look right(except for rounding)
i have never believed it matters one iota what the actual times are, the only thing that matters is the relationships between them and in that regard they look fine

EDIT: just looked at the beaten margin chart, and that is the opposite.
it is nonsense.
it should be obvious that you can't treat also rans the same as you treat the winners.

[cj]

EDIT: just looked at the beaten margin chart, and that is the opposite.
it is nonsense.
it should be obvious that you can't treat also rans the same as you treat the winners.

I somewhat agree, but curious how do you treat also rans? I have some ideas that I will share but would like to hear your point of view.

[steveb]

I somewhat agree, but curious how do you treat also rans? I have some ideas that I will share but would like to hear your point of view.


assuming you know the conversion used by the lynx(or whatever) operator then.....
i am assuming 1 length = .16 second

so 6 lengths would be .96 of a second.

according to that chart 57.8 = 99 points for 5 furlongs
96.4 = 99 for 8 furlongs

so .......57.8 +.96 = 58.76
96.4 + .96 = 97.36

((57.8/58.76) ^ 1.0-0.9) * 1000 -100 + 99 = 82.7
((96.4/97.36) ^ 1.6-0.9) * 1000 -100 + 99 = 83.3

of course it will NEVER be perfect no matter how you do it, but this way it keeps things in proportion better..... for instance if the 8f race was run in 99.4 then .....
it would end up 83.7

the -100 + 99 are only there because the 57.8 and the 96.4 were equal to 99 on that chart instead of the 100 that all my base(pars, standards, or whatever you call them) times are.

so basically 6 lengths beaten at any distance would be about the same, just slightly in favour of the longer distances.
the slower the time, then the more the beaten margin will mean

i don't have the problem of converting margins to times, as i already know the times all the individual horse have run.

anyway i would have the respective rating for beaten 6 lengths over 5f and 8f as 82.7 and 83.3, which is bugger all in the scheme of things, whereas that chart would give you 81 an 90 respectively.
of course it is blatantly obvious that the longer the distance then the better off the beaten horse are.

[classhandicapper]

I think all the beaten length charts and values assigned to a 1/5th of a second are suspect.

For example, let's say we have 2 horses that are versatile enough to run identical races at 6F and 12F.

Horse "A" beats horse "B" by 2 lengths at 6F. Let say they earn figures of 100 and 95 for their performances. According to theory he should beat him by quite a bit more at 12F because they are running twice as far and 5 points is worth more lengths at 12F.

But all the evidence I have collected says this is not true. Horse A will probably run his race the same way, make his move at the same time, will win again, it will be by more than 2 lengths, but it will be by less than than expected.

Horses generally use each other as prompters and only do their best running for part of the race (especially on turf). So even though average margins do expand as distances stretch out, it's not by as much as theory suggests. I'm not sure how you would incorporate that into figures, but the theory seems amiss to me.
(an exception might be a front runner).

[raybo]

I think all the beaten length charts and values assigned to a 1/5th of a second are suspect.

For example, let's say we have 2 horses that are versatile enough to run identical races at 6F and 12F.

Horse "A" beats horse "B" by 2 lengths at 6F. Let say they earn figures of 100 and 95 for their performances. According to theory he should beat him by quite a bit more at 12F because they are running twice as far and 5 points is worth more lengths at 12F.

But all the evidence I have collected says this is not true. Horse A will probably run his race the same way, make his move at the same time, will win again, it will be by more than 2 lengths, but it will be by less than than expected.

Horses generally use each other as prompters and only do their best running for part of the race (especially on turf). So even though average margins do expand as distances stretch out, it's not by as much as theory suggests. I'm not sure how you would incorporate that into figures, but the theory seems amiss to me.
(an exception might be a front runner).

I agree that any chart, or player, that uses the same time per beaten length, arbitrarily, simply doesn't care what the real time difference is. They just want something easy and static to work with. The value of a beaten length, in reference to time, varies with the speed the horses are traveling. The faster the horses are running the less time it takes to run a length, and the slower they run the more time it takes to run a length. So, any chart, or anyone who uses a static time per beaten length is starting the adjustment process with built-in errors. Some of them, obviously, will be huge errors.

[Tom]

Good point, but what time do you use?
For 6 furlongs, do you use the final times to calculate the value of a length?
Of the final fraction? or final eighth?

Horse A goes to the front and runs hard all 6 furlongs, losing ground in the stretch and just holds on to win a photo.

horse B lags back 5 lengths and rockets down the stretch to just miss.
Essenstially the same final times, but B ran a much faster last fraction and eighth. Wold you say he covered the last length the in the same time as the tiring A?

[raybo]

Good point, but what time do you use?
For 6 furlongs, do you use the final times to calculate the value of a length?
Of the final fraction? or final eighth?

Horse A goes to the front and runs hard all 6 furlongs, losing ground in the stretch and just holds on to win a photo.

horse B lags back 5 lengths and rockets down the stretch to just miss.
Essenstially the same final times, but B ran a much faster last fraction and eighth. Wold you say he covered the last length the in the same time as the tiring A?

Obviously, this current discussion is in reference to final beaten lengths, not fractional beaten lengths. So, the obvious answer is the shortest distance, at the end of the race, that has a viable raw time associated with it.

If we were talking about a fractional beaten length then it would, obviously, be that specific fraction. But here, we are talking about the final beaten lengths, so it will obviously be the last timed fraction of the race.

This points to another fact, traditional speed figures or final times, cannot be used, in isolation, to measure the value of a performance. Everything that happened before the finish line must be incorporated in any performance evaluation.

[RXB]

I think all the beaten length charts and values assigned to a 1/5th of a second are suspect.

For example, let's say we have 2 horses that are versatile enough to run identical races at 6F and 12F.

Horse "A" beats horse "B" by 2 lengths at 6F. Let say they earn figures of 100 and 95 for their performances. According to theory he should beat him by quite a bit more at 12F because they are running twice as far and 5 points is worth more lengths at 12F.

But all the evidence I have collected says this is not true. Horse A will probably run his race the same way, make his move at the same time, will win again, it will be by more than 2 lengths, but it will be by less than than expected.

Horses generally use each other as prompters and only do their best running for part of the race (especially on turf). So even though average margins do expand as distances stretch out, it's not by as much as theory suggests. I'm not sure how you would incorporate that into figures, but the theory seems amiss to me.
(an exception might be a front runner).

Agreed. On turf and synthetic, a person could actually use the same beaten lengths value for the frequently raced distances in North America and not be very far off. On dirt, the margins do expand with distance but the increase in average winning margin isn't nearly as much as the increase in distance.

Average winning margin for each distance on each surface is a better base for a beaten lengths value chart. Any chart that is making the value of a beaten length proportional to the race distance is underrating defeated sprinters compared to defeated routers.

[Tom]

I'm talking final, not fractional.
So you vote for the final 8th?

But now, do you calculate a different value for each horse?
This might be what Steve and CJ were talking about with the also rans.

Interesting topic.

[classhandicapper]

Any chart that is making the value of a beaten length proportional to the race distance is underrating defeated sprinters compared to defeated routers.

It's a mathematical dilemma of sorts.

Even if you have the exactly correct final time for the beaten horse in my example, you will still underrate the defeated 6F horse relative to at 12F because of the way the value of 1/5th of a second varies from distance to distance.

Your approach should correct that, but it moves away from the actual final time.

When I think about it my head explodes. :lol:

It's stuff like this makes me think just looking at field quality and the horses' trips and finishes within those fields is the better approach. The times matter, but there are so many technical and other difficulties, I just want to know if it was fast, slow, or average for the class.

[raybo]

I'm talking final, not fractional.
So you vote for the final 8th?

But now, do you calculate a different value for each horse?
This might be what Steve and CJ were talking about with the also rans.

Interesting topic.

Unless you have GPS data (and have access to each individual horse's time and distance traveled in the last fraction), or hand time every horse and somehow accounted for changes in direction to obtain individual distances traveled), then the simple method would be to use the winner's time minus the previous call leader's time and the distance from that previous call to the finish line. But of course, that is not optimal, for obvious reasons. The problem is that "also rans" could have actually run faster in the final fraction than the winner, which would mean that their beaten lengths conversion to time, would be worth less time than that "simple method" would result in. And, I think that is exactly what Steve and CJ were referring to. That's also why "performance figures" are much better than "speed figures", and also why it would be much better if we had actual times for each call, for each horse running in the race. Everything we do in handicapping, is an "estimation", but the closer our estimations are to reality, the better we will be at projecting future outcomes. "Every little bit helps"

If your records say that good enough is good enough, then good enough is good enough (until you figure out a way to make good enough, better ;) )

[Capper Al]

Agreed. On turf and synthetic, a person could actually use the same beaten lengths value for the frequently raced distances in North America and not be very far off. On dirt, the margins do expand with distance but the increase in average winning margin isn't nearly as much as the increase in distance.

Average winning margin for each distance on each surface is a better base for a beaten lengths value chart. Any chart that is making the value of a beaten length proportional to the race distance is underrating defeated sprinters compared to defeated routers.

My research disagrees. Turf and synthetic and dirt are all different.

[raybo]

My research disagrees. Turf and synthetic and dirt are all different.

LOL - my research disagrees. All races are different, regardless of the surface or distance.

[steveb]

I'm talking final, not fractional.
So you vote for the final 8th?

But now, do you calculate a different value for each horse?
This might be what Steve and CJ were talking about with the also rans.

Interesting topic.


it would automatically be a different value, depending on winner time & time of also ran.
but the differences in the scheme of things would be bugger all, so it's not that important.

what IS important is that those charts for beaten margins that cj pointed me to, are bad in the extreme.

i should also say the post i put up explaining my way is not quite correct, in that i kind of cheated.
instead of using 5f & 8f i used 1000m & 1600m because i could no be bothered doing the conversion. and that will explain the ^1 & ^1.6
if you used 1 for both of those calcs they would be basically the same as that chart, and favouring the beaten runners far too much as the distances increase.

[RXB]

My research disagrees. Turf and synthetic and dirt are all different.

I know that; perhaps I wasn't quite clear.

The average winning margin in turf sprints is not very different from middle-distance turf races.

The average winning margin in synthetic sprints is not very different from middle-distance synthetic races.

The average winning margin in dirt sprints is noticeably different from middle-distance dirt races but that average margin increases by a factor less than the increase in race distance.

[steveb]

I know that; perhaps I wasn't quite clear.

The average winning margin in turf sprints is not very different from middle-distance turf races.

The average winning margin in synthetic sprints is not very different from middle-distance synthetic races.

The average winning margin in dirt sprints is noticeably different from middle-distance dirt races but that average margin increases by a factor less than the increase in race distance.

even intuitively that would be correct.
because dirt tires more than turf.
if you measured wet turf with dry turf it would be true too.
as the distances increase the margins would increase too, but would be more dependent on pace.

the reality is that both me and those charts are wrong, because there can be no right that covers it all.
the truth for any one race would likely fall between me and that chart, but i am 100% which is closer to the truth.

[steveb]

I somewhat agree, but curious how do you treat also rans? I have some ideas that I will share but would like to hear your point of view.

your turn

[RXB]

the reality is that both me and those charts are wrong, because there can be no right that covers it all.
the truth for any one race would likely fall between me and that chart, but i am 100% which is closer to the truth.

Yes, Beyer made those charts in the 1970's and he along with many other people have better data now. I know he has made some adjustments, especially regarding turf and synthetic. And of course there are so many variations on race speeds/dynamics that nothing will provide perfect accuracy. Especially on dirt where early running positions are so linked to finish position tendencies.

[steveb]

Yes, Beyer made those charts in the 1970's and he along with many other people have better data now. I know he has made some adjustments, especially regarding turf and synthetic. And of course there are so many variations on race speeds/dynamics that nothing will provide perfect accuracy. Especially on dirt where early running positions are so linked to finish position tendencies.

i have no idea if beyer has changed or not,but it has nothing to do with data.

all he was doing in that chart was keeping the proportions the same as the time charts, and it would be correct if you WERE NOT using it on the beaten runners.
in other words he was just keeping the percentage differences constant.

anyway, there's this bloke called dl, and i have this vivid image in my mind of smoke coming from his ears, coz we are not addressing the thread topic.
so maybe i best drop off, although in a way i AM addressing his topic, i just doubt he would see it that way.

[RXB]

i have no idea if beyer has changed or not,but it has nothing to do with data.

all he was doing in that chart was keeping the proportions the same as the time charts, and it would be correct if you WERE NOT using it on the beaten runners.
in other words he was just keeping the percentage differences constant.

anyway, there's this bloke called dl, and i have this vivid image in my mind of smoke coming from his ears, coz we are not addressing the thread topic.
so maybe i best drop off, although in a way i AM addressing his topic, i just doubt he would see it that way.

Whether in this thread or others, I hope you keep posting as you are very learned and I appreciate your sharing of knowledge.

[classhandicapper]

Let's remove beaten lengths from the conversation and just talk about actual running times (under the assumption that we have the actual times). That will both simplify the conversation and highlight the issue I am raising.

Horse A beats horse B by 2/5ths of a second at 6F.

The race again next time at 12F

Will A beat B by 4/5ths or more as the speed figure charts suggest?

I say NO.

[thaskalos]

Let's remove beaten lengths from the conversation and just talk about actual running times (under the assumption that we have the actual times). That will both simplify the conversation and highlight the issue I am raising.

Horse A beats horse B by 2/5ths of a second at 6F.

The race again next time at 12F

Will A beat B by 4/5ths or more as the speed figure charts suggest?

I say NO.

What are you suggesting, CH?

What should be the difference between the point value of a length at six furlongs, and a length at a mile and a quarter?

[RXB]

Let's remove beaten lengths from the conversation and just talk about actual running times (under the assumption that we have the actual times). That will both simplify the conversation and highlight the issue I am raising.

Horse A beats horse B by 2/5ths of a second at 6F.

The race again next time at 12F

Will A beat B by 4/5ths or more as the speed figure charts suggest?

I say NO.

I see the problem as being the adjustment for beaten lengths, not the figure chart for the winning horse.

[Capper Al]

LOL - my research disagrees. All races are different, regardless of the surface or distance.

Lol, I'm talking averages over the long run.

[Cratos]

I agree that any chart, or player, that uses the same time per beaten length, arbitrarily, simply doesn't care what the real time difference is. They just want something easy and static to work with. The value of a beaten length, in reference to time, varies with the speed the horses are traveling. The faster the horses are running the less time it takes to run a length, and the slower they run the more time it takes to run a length. So, any chart, or anyone who uses a static time per beaten length is starting the adjustment process with built-in errors. Some of them, obviously, will be huge errors.
The problem here is that there is an attempt by some handicappers to solve a nonlinear downward sloping curve with linearity.

[raybo]

The problem here is that there is an attempt by some handicappers to solve a nonlinear downward sloping curve with linearity.

I agree. Linearity is fine if that linearity depends on the speed the horse is traveling when the beaten lengths are determined, otherwise linearity doesn't apply to beaten length to time conversions.

[johnhannibalsmith]

Whether in this thread or others, I hope you keep posting as you are very learned and I appreciate your sharing of knowledge.

Agreed, at least I get the sense that you (steveb) are saying something.

[cj]

assuming you know the conversion used by the lynx(or whatever) operator then.....
i am assuming 1 length = .16 second

so 6 lengths would be .96 of a second.

according to that chart 57.8 = 99 points for 5 furlongs
96.4 = 99 for 8 furlongs

so .......57.8 +.96 = 58.76
96.4 + .96 = 97.36

((57.8/58.76) ^ 1.0-0.9) * 1000 -100 + 99 = 82.7
((96.4/97.36) ^ 1.6-0.9) * 1000 -100 + 99 = 83.3

of course it will NEVER be perfect no matter how you do it, but this way it keeps things in proportion better..... for instance if the 8f race was run in 99.4 then .....
it would end up 83.7

the -100 + 99 are only there because the 57.8 and the 96.4 were equal to 99 on that chart instead of the 100 that all my base(pars, standards, or whatever you call them) times are.

so basically 6 lengths beaten at any distance would be about the same, just slightly in favour of the longer distances.
the slower the time, then the more the beaten margin will mean

i don't have the problem of converting margins to times, as i already know the times all the individual horse have run.

anyway i would have the respective rating for beaten 6 lengths over 5f and 8f as 82.7 and 83.3, which is bugger all in the scheme of things, whereas that chart would give you 81 an 90 respectively.
of course it is blatantly obvious that the longer the distance then the better off the beaten horse are.


Thanks, I'll get back to this in a few days. Got whacked in the eye by a tree branch so limited computer time for a few days, but I do appreciate the response.

[cj]

I think all the beaten length charts and values assigned to a 1/5th of a second are suspect.

For example, let's say we have 2 horses that are versatile enough to run identical races at 6F and 12F.

Horse "A" beats horse "B" by 2 lengths at 6F. Let say they earn figures of 100 and 95 for their performances. According to theory he should beat him by quite a bit more at 12F because they are running twice as far and 5 points is worth more lengths at 12F.

But all the evidence I have collected says this is not true. Horse A will probably run his race the same way, make his move at the same time, will win again, it will be by more than 2 lengths, but it will be by less than than expected.

Horses generally use each other as prompters and only do their best running for part of the race (especially on turf). So even though average margins do expand as distances stretch out, it's not by as much as theory suggests. I'm not sure how you would incorporate that into figures, but the theory seems amiss to me.
(an exception might be a front runner).

This is exactly what I found when I wrote this article for HANA:

http://horseplayersassociation.org/jan15issue.pdf

Somebody I'm going to tackle exploring using values like this, though more distance specific than sprint/route. The way beaten lengths are treated by most now would be perfect if horses were all out for the full distance of the race, but of course that isn't possible. Races are tactical and the "race" isn't really for the full distance.

[cj]

I'm talking final, not fractional.
So you vote for the final 8th?

But now, do you calculate a different value for each horse?
This might be what Steve and CJ were talking about with the also rans.

Interesting topic.

If you want to be most accurate, that is the way to do it in my opinion.

[cj]

Here is an example for dirt sprints on a fast track, the average winning margin:

Surf Dist AvgOfBLF
1-Dirt 500 2.5622
1-Dirt 550 2.6263
1-Dirt 600 2.6126
1-Dirt 650 2.4304
1-Dirt 700 2.6031
1-Dirt 750 2.6388


So it is actually really consistent. If time has less value as distances increase, why don't we see it in the finish margins? It almost looks like winner's times and times of also rans should not be treated in the same manner. This is what I suggest in the article I posted.

[cj]

If you want to be most accurate, that is the way to do it in my opinion.

I was mistaken when I posted this, I was thinking about the fractional calls, not the finish.

[cj]

Here is more interesting data. Like has been stated, "parallel speed charts" that all speed figure makers I am aware of given more value to time at shorter distances. For example, with Beyer a 6f race gives a second a value of about 14 points. At 7f, a second is worth about 12 points, and at a mile a second is equal to about 10. The prevailing wisdom has always been that as the races get longer, it makes sense that horses would lose by more to superior horses.

But, I already posted this doesn't really happen in sprints. This is what the most common routes look like:

Surf Dist Races AvgOfBLF
1-Dirt 800 20726 2.9102
1-Dirt 832 10699 3.0140
1-Dirt 850 9290 2.7701
1-Dirt 900 1286 2.6479

The margins at the finish are actually getting smaller as the race gets longer. Doesn't this contradict the whole framework of parallel time charts?

(sorry Delta, you never know where a thread will go)

[classhandicapper]

What are you suggesting, CH?

What should be the difference between the point value of a length at six furlongs, and a length at a mile and a quarter?

I haven't been able to resolve this issue in my mind in mathematical terms.

If you measure horses' performances purely in terms of time/speed figures and build the time charts the way theory suggests, it will tell you one thing about the relationships between times/beaten lengths and various distances.

But if you look at the actual results across distances it will tell you something else.

The reality is probably that the charts that form the basis of all speed figures have an implied assumption in them that horses are doing close to their best all the way. But since the reality is horses kind of use each other as prompters and make their last moves at similar times in the race and then do their best, the charts are a little off.

Is is possible to build a better chart?

I have no idea. It would have to be tested. But if you did, the figures would lose their connection to the actual times a little bit. :confused:

I doubt that all made sense because it's hard for me wrap my head, let alone explain it. It's not a major issue, but it's something.

[cj]

I haven't been able to resolve this issue in my mind in mathematical terms.

If you measure horses' performances purely in terms of time/speed figures and build the time charts the way theory suggests, it will tell you one thing about the relationships between times/beaten lengths and various distances.

But if you look at the actual results across distances it will tell you something else.

The reality is probably that the charts that form the basis of all speed figures have an implied assumption in them that horses are doing close to their best all the way. But since the reality is horses kind of use each other as prompters and make their last moves at similar times in the race and then do their best, the charts are a little off.

Is is possible to build a better chart?

I have no idea. It would have to be tested. But if you did, the figures would lose their connection to the actual times a little bit. :confused:

I doubt that all made sense because it's hard for me wrap my head, let alone explain it. It's not a major issue, but it's something.

I wonder if keeping the charts for winners as is, but using different values for the amount beaten is the way to go. I'm always looking for stuff to do between December and February :)

[Magister Ludi]

Intra-track adjustments should be based upon three factors: going, pace, and ability of the competitors. These three factors can be separated and quantified by a systematic procedure which is based upon calculating the energy expenditure of the competitors. A simple example would require hours of typing, which I can’t spare.

[cj]

A simple example would require hours of typing, which I can’t spare.

Must not be all that simple then, right?

[Cratos]

I haven't been able to resolve this issue in my mind in mathematical terms.

If you measure horses' performances purely in terms of time/speed figures and build the time charts the way theory suggests, it will tell you one thing about the relationships between times/beaten lengths and various distances.

But if you look at the actual results across distances it will tell you something else.

The reality is probably that the charts that form the basis of all speed figures have an implied assumption in them that horses are doing close to their best all the way. But since the reality is horses kind of use each other as prompters and make their last moves at similar times in the race and then do their best, the charts are a little off.

Is is possible to build a better chart?

I have no idea. It would have to be tested. But if you did, the figures would lose their connection to the actual times a little bit. :confused:

I doubt that all made sense because it's hard for me wrap my head, let alone explain it. It's not a major issue, but it's something.
If you want to solve this problem in mathematical terms, the task will be daunting and rigorous to arrive at the correct solution because the answer is very complex and it involves an understanding and application of the nonlinear system which the thoroughbred horseracing curve represents.

However let’s look at it in terms of linearity (which the racing curve is not) and we will find the superposition property which states that in the linear system the net response at a given place and time caused by two or more stimuli is the sum of the responses which would have been caused by each stimulus individually. So that if input A produces response X and input B produces response Y then input (A + B) produces response (X + Y).

Therefore “the difference between the point value of a length at six furlongs, and a length at a mile and a quarter” can best be described from a definition of nonlinearity from the Internet which states that a nonlinear system of equations is a set of simultaneous equations in which the unknowns (or the unknown functions in the case of differential equations) appear as variables of a polynomial of degree higher than one or in the argument of a function which is not a polynomial of degree one.

In other words, in a nonlinear system of equations, the equation(s) to be solved cannot be written as a linear combination of the unknown variables or functions that appear in them. It does not matter if nonlinear known functions appear in the equations. In particular, a differential equation is linear if it is linear in terms of the unknown function and its derivatives, even if nonlinear in terms of the other variables appearing in it.

[Magister Ludi]

Must not be all that simple then, right?

Simple and lengthy are not mutually exclusive terms.

[Cratos]

Intra-track adjustments should be based upon three factors: going, pace, and ability of the competitors. These three factors can be separated and quantified by a systematic procedure which is based upon calculating the energy expenditure of the competitors. A simple example would require hours of typing, which I can’t spare.
If you have the mass (weight + load) and the velocity of each horse the kinetic energy is a simple and fast calculation.

What am I missing?

[cj]

Simple and lengthy are not mutually exclusive terms.

Sure, but if it requires hours of typing than the simple part will be relative to the readers.

[Magister Ludi]

If you have the mass (weight + load) and the velocity of each horse the kinetic energy is a simple and fast calculation.

What am I missing?

Aerodynamic drag and energy cost of locomotion. Decomposition of the three adjustment factors of going, pace, and ability of the competitors is the lengthy part of the process.

[RXB]

I wonder if keeping the charts for winners as is, but using different values for the amount beaten is the way to go. I'm always looking for stuff to do between December and February :)

Beaten lengths adjustments can be improved in a couple of ways.

One is that they need to reflect the realities of each distance and surface.

The other is that the figure deduction for a unit of time should be less when applied to beaten horses than on the chart for winning times. The goal should be to measure and predict performance, rather than time; they are correlated but not the same. Overall it's more difficult for horses to run to their capabilities when outclassed than when in a favourable spot.

[classhandicapper]

I wonder if keeping the charts for winners as is, but using different values for the amount beaten is the way to go. I'm always looking for stuff to do between December and February :)

Someone suggested something along those lines earlier in the thread (RXB?). I have no problem with anything that works better, but you would lose a bit of the connection to the actual times.

I've been screwing around with alternatives. They are testing better so far, but it's really hard to test beaten lengths in an objective way.

One horse loses by 10 lengths because he's inferior to that class by 10 lengths.

One horse loses by 10 lengths because the top 2 finishers were 6 lengths better than the class.

It hard to systematically deal with all the possibilities.

[Cratos]

Aerodynamic drag and energy cost of locomotion. Decomposition of the three adjustment factors of going, pace, and ability of the competitors is the lengthy part of the process.
You have somewhat expanded the task, but I am still not seeing the “hours” of work. Yes, I understand that your time is valuable to you and you might not want to spend it on this example, but it is not a lifelong endeavor because some of the factors you listed can be integrated together.

[whodoyoulike]

I wonder if keeping the charts for winners as is, but using different values for the amount beaten is the way to go. I'm always looking for stuff to do between December and February :)

A while back, you made a suggestion that tracks should install tele- timers similar to the one at the finish line at each fractional point of call which would provide more accurate pp's. This would resolve the BL problems.

What's the reason for the opposition?

Are these timers that expensive? We're probably talking about 4 add'l timers.

[whodoyoulike]

... but it's really hard to test beaten lengths in an objective way.

One horse loses by 10 lengths because he's inferior to that class by 10 lengths.

One horse loses by 10 lengths because the top 2 finishers were 6 lengths better than the class.

It hard to systematically deal with all the possibilities.

Also, keep in mind that a horse behind 10 lengths may ease up if the jockey doesn't think he will lose final positioning. I figure the winner's rating will always be correct but the also ran may not be accurate. They probably could do better in most cases unless they're sucking wind at the finish.

[steveb]

If you want to solve this problem in mathematical terms, the task will be daunting and rigorous to arrive at the correct solution because the answer is very complex and it involves an understanding and application of the nonlinear system which the thoroughbred horseracing curve represents.

However let’s look at it in terms of linearity (which the racing curve is not) and we will find the superposition property which states that in the linear system the net response at a given place and time caused by two or more stimuli is the sum of the responses which would have been caused by each stimulus individually. So that if input A produces response X and input B produces response Y then input (A + B) produces response (X + Y).

Therefore “the difference between the point value of a length at six furlongs, and a length at a mile and a quarter” can best be described from a definition of nonlinearity from the Internet which states that a nonlinear system of equations is a set of simultaneous equations in which the unknowns (or the unknown functions in the case of differential equations) appear as variables of a polynomial of degree higher than one or in the argument of a function which is not a polynomial of degree one.

In other words, in a nonlinear system of equations, the equation(s) to be solved cannot be written as a linear combination of the unknown variables or functions that appear in them. It does not matter if nonlinear known functions appear in the equations. In particular, a differential equation is linear if it is linear in terms of the unknown function and its derivatives, even if nonlinear in terms of the other variables appearing in it.

nobody is going to get a correct solution because there ISN'T one.
all you need is the times of the horses or know the conversion factor used and the distance of the race.
nothing else is needed.
despite the fact i know it is wrong(there is no right),my way works fine,the only thing i would change these days if i still wanted to invest would be the 'to the power of' number and get best fit as far as dollars returned, although it has worked just fine for many many years.
it always bemuses me when people try to make something simple look daunting.

you use OTHER factors to account for the other stuff like how fast they are going at the end, or if they are making or losing ground or eased or whatever.

[cj]

Beaten lengths adjustments can be improved in a couple of ways.

One is that they need to reflect the realities of each distance and surface.

The other is that the figure deduction for a unit of time should be less when applied to beaten horses than on the chart for winning times. The goal should be to measure and predict performance, rather than time; they are correlated but not the same. Overall it's more difficult for horses to run to their capabilities when outclassed than when in a favourable spot.

I think most are using different values for distance and surface.

As far as the second part, I've actually been doing that for long time. I could probably even decrease the amount more but people might freak out over it. :)

[RXB]

Someone suggested something along those lines earlier in the thread (RXB?). I have no problem with anything that works better, but you would lose a bit of the connection to the actual times.

I've been screwing around with alternatives. They are testing better so far, but it's really hard to test beaten lengths in an objective way.

One horse loses by 10 lengths because he's inferior to that class by 10 lengths.

One horse loses by 10 lengths because the top 2 finishers were 6 lengths better than the class.

It hard to systematically deal with all the possibilities.

Sometimes, loosening the connection with the times will improve performance ratings. There's no doubt that competition affects performance. To whatever extent a compounded rating (such as the TFUS "speed figure") can factor in the effects of competition, I'm all for it.

I'd rather have a general or incremental improvement and accept its limitations than bog down trying to deal with every specific variable in every specific race and end up getting nowhere at all. Wheel invented long before automobile. Start with simplicity; add complexity only where needed and as able.

[whodoyoulike]

Mr. PA, I don't believe that you got the memo. "Handicapper's Corner" is now "Capper Al's Corner". Do you suppose that there's a correlation between the content and the popularity, or lack thereof?

Al can correct me if I'm incorrect but I recall "Handicapper's Corner" was suggested by him. When first brought up, I thought it had a lot of potential for learning about handicapping.

[Capper Al]

Al can correct me if I'm incorrect but I recall "Handicapper's Corner" was suggested by him. When first brought up, I thought it had a lot of potential for learning about handicapping.

What's this all about? Yeah, I started the Handicapper's Corner with PA. So what?

[RXB]

I think most are using different values for distance and surface.


Different: yes. Accurate: doubtful in many cases.

[whodoyoulike]

What's this all about? Yeah, I started the Handicapper's Corner with PA. So what?

I was just pointing it out to Magister Ludi and I always thought it was a good idea.

Why are you coming back with this type of response?

Actually, I started noticing from this idea that you had some good ideas which I've mentioned to you in other threads.

[Capper Al]

I was just pointing it out to Magister Ludi and I always thought it was a good idea.

Why are you coming back with this type of response?

Actually, I started noticing from this idea that you had some good ideas which I've mentioned to you in other threads.

No Problem.

[RXB]

As far as the second part, I've actually been doing that for long time. I could probably even decrease the amount more but people might freak out over it. :)

I think I've worked it out but can you just confirm that this is accurate on your scale for 6f on dirt run in 1:11?

.20 of a second difference in winning times = 3 points
Beaten length ~ 1/7 of a second
Raw (not applied) figure value of a beaten length ~ 2.14 points
Applied figure value of a beaten length = 2 points

[cj]

I think I've worked it out but can you just confirm that this is accurate on your scale for 6f on dirt run in 1:11?

.20 of a second difference in winning times = 3 points
Beaten length ~ 1/7 of a second
Raw (not applied) figure value of a beaten length ~ 2.14 points
Applied figure value of a beaten length = 2 points

Nope, I cannot confirm or deny :)

[steveb]

I think I've worked it out but can you just confirm that this is accurate on your scale for 6f on dirt run in 1:11?

.20 of a second difference in winning times = 3 points
Beaten length ~ 1/7 of a second
Raw (not applied) figure value of a beaten length ~ 2.14 points
Applied figure value of a beaten length = 2 points

my opinion is that you don't need one for dirt and one for turf.
for the same reason that wet turf would not be the same as dry turf.
you just let the values figure themselves, you don't need 'constructs'.

to sidetrack a little although it's very valid.
let's go back to beyer numbers.
the ONLY time when beyer points for beaten runners will be correct, is for 1000 metres(or in your lingo 5f would be near enough), and the reason for that is that beyer's are only percentage figures, nothing more.
an 80 of beyers is always going to be 98% of what the 100 rater will be.
a 60 will be 96% and so on.
so the 1000m lines up perfectly with your beyers.
thus you can also say that 1 beyer point equals 1 metre at 1000 metres.

on that chart for 5f then about 56 seconds equals 131 points.
58 seconds = 96 points.

or the simple way....(56/58-.9)*1000 = 65.5 + 31(because the 56 was 131 and not 100)=96.5

but just say the track was really slow, and after your variant you found that 61 was worth 131 points and to get the value of a horse beaten 2 seconds.... (61/63-.9)= 68.2 + 31 = 99.2
thus you can see it is automatically compensating for that difference in time, and it's why i don't think you need to worry about surface, in your calculations.
and why i don't think you need even worry about how many points per length, as it is automatically accounted for.

you would notice i did not include the power sign(^) as it is only 1 for 1000 metres or near enough for 5 furlongs.

so like beyer(even if he does not know!) you should make your base distance 5 furlongs.

unlike beyer, if the race you are trying to get numbers for is 8 furlongs then you raise to the power of 8f/5f which is 1.6
5f to 10f would be ^2 and so on.

so your 8 furlongs beyer......94 seconds = 124 points
96 seconds = 103 points

my way of doing beyer......
(94/96-.9)*1000 =79 + 24 = 103
but of course for beaten margins we have already ascertained that is nonsense.
this is where your power comes into it.
i would do ((94/96)^1.6-.9)*1000 = 66.9 plus that extra 24 = 90.9
but let's assume the beyer for 94 seconds actually equalled the same 131 as in the 5f example.
then beaten 2 seconds will be 97.9 and just a little bit better than beaten 2 seconds over 5f.

[cj]

This is what I do. The value of time varies based on the time of the race. It is more accurate.

That said, it doesn't make much difference than using a fixed value, IMO, since the range of actual times we deal with at each distance in the US isn't very large.

[classhandicapper]

For the record, we shouldn't be talking about Beyer's beaten length charts in the present tense. They have been changed and to my knowledge no one has spoken publicly about the new secret sauce.

[classhandicapper]

steveb,

You are explaining how to make better "time based" numbers so they are more accurate. That's a worthy goal.

I think the entire system of equalizing times at various distances is flawed because of the way races develop and see no good solution other than ignoring the time in anything other than a very general sense.

[steveb]

This is what I do. The value of time varies based on the time of the race. It is more accurate.

That said, it doesn't make much difference than using a fixed value, IMO, since the range of actual times we deal with at each distance in the US isn't very large.

a fixed value would indeed not make much difference, IF they all raced on tracks racing the same speed, AND, the margins were small.
the deeper the variations in going, and the larger the margins then the bigger the difference in values.

but it's certainly NOT rocket science like some on here would have you believe.

and a little note to delta lover to keep on topic!.....that everything is proportionate is the VERY reason i can line all the jurisdictions up with no bother at all.

[steveb]

steveb,

You are explaining how to make better "time based" numbers so they are more accurate. That's a worthy goal.

I think the entire system of equalizing times at various distances is flawed because of the way races develop and see no good solution other than ignoring the time in anything other than a very general sense.

of course it's flawed, there is NO un-flawed way.
all i am suggesting is the best way 'I' have discovered.
but the flaws that can't be fixed this way, ARE fixed, using other things like section times and position in running rankings at various points.
after all, time is just one part of a very big puzzle.
in my models, time accounted for only about .3 of the form factors in them.
you can't expect to use time as a standalone method, it does not work like that.

and just a little side note.
maybe it's because i am no longer involved in racing that i don't mind saying the odd thing, but it strikes me that i am the only one giving you more than talk, and am giving examples of what i think.
if you or somebody can do it better, then by all means show us by example?

the rocket science and fancy talk i would leave to the theorists, and those that get lost in complexity.
there is only one thing that matters if your aim is to win money for a prolonged period.
how much is in your pockets.

[Cratos]

This is what I do. The value of time varies based on the time of the race. It is more accurate.

That said, it doesn't make much difference than using a fixed value, IMO, since the range of actual times we deal with at each distance in the US isn't very large.
"The value of time varies based on the time of the race. It is more accurate."

The value of time is constant and doesn’t vary based on the time of the race; unit value of distance per unit of constant time per race length varies for a variety of reasons, but most notably the environmental influences (air resistance, surface resistance, and wind resistance) and the horse’s energy expenditure with respect to its velocity output.

[steveb]

"The value of time varies based on the time of the race. It is more accurate."

The value of time is constant and doesn’t vary based on the time of the race; unit value of distance per unit of constant time per race length varies for a variety of reasons, but most notably the environmental influences (air resistance, surface resistance, and wind resistance) and the horse’s energy expenditure with respect to its velocity output.

crikey.
forget the nonsense, cj is correct in what he meant, and that is all that matters.
the slower the time the race is run, for whatever reason, then the values should be different to keep things in proportion.

[cj]

crikey.
forget the nonsense, cj is correct in what he meant, and that is all that matters.
the slower the time the race is run, for whatever reason, then the values should be different to keep things in proportion.

Yeah, I've been avoiding that nonsense for a while now. I'm glad you understood as most horse bettors will.

[Cratos]

crikey.
forget the nonsense, cj is correct in what he meant, and that is all that matters.
the slower the time the race is run, for whatever reason, then the values should be different to keep things in proportion.
It is not nonsense and CJ is not correct in what he stated; that is the problem with horserace handicapping, too many cockamamie ideas about what is correct.

Whether you like it or not, the science underbelly of horseracing is physics.

You can dance around with a lot of “ifs & ands”, but the bottom-line in this business is gambling to make money.

Therefore if I am wrong then mathematically and scientifically prove it to me and I will gladly accept it if it is correct.

[steveb]

It is not nonsense and CJ is not correct in what he stated; that is the problem with horserace handicapping, too many cockamamie ideas about what is correct.

Whether you like it or not, the science underbelly of horseracing is physics.

You can dance around with a lot of “ifs & ands”, but the bottom-line in this business is gambling to make money.

Therefore if I am wrong then mathematically and scientifically prove it to me and I will gladly accept it if it is correct.

the bit in bold you got right, and from reading you for a good while, i am positive YOU are NOT one those that do.
not that i would give a fig either way.
nearly all lose anyway, regardless of how smart they are(or think they are)

i am not interested in proving you wrong, that is a child's game.
you can play that on your own.

cj is perfectly correct in what he means, regardless of what you think.
give us an example in figures(not talk) of why he is wrong?
until you do that you are just a noise maker, nothing else in my humble opinion

one of my friends is a physicist(and trained actuary) in australia and was one of the leaders in the team behind the cochlear implant system that was invented here in melbourne australia.
he has helped me with lots of things in my racing, because time and the like is right up his alley.
but one thing stands out, once he is on a racetrack, off goes his head and on goes a pumpkin.

[Cratos]

the bit in bold you got right, and from reading you for a good while, i am positive YOU are NOT one those that do.
not that i would give a fig either way.
nearly all lose anyway, regardless of how smart they are(or think they are)

i am not interested in proving you wrong, that is a child's game.
you can play that on your own.

cj is perfectly correct in what he means, regardless of what you think.
give us an example in figures(not talk) of why he is wrong?
until you do that you are just a noise maker, nothing else in my humble opinion

one of my friends is a physicist(and trained actuary) in australia and was one of the leaders in the team behind the cochlear implant system that was invented here in melbourne australia.
he has helped me with lots of things in my racing, because time and the like is right up his alley.
but one thing stands out, once he is on a racetrack, off goes his head and on goes a pumpkin.
Thanks for the reply, but you are talking in circles and if I don't make any money in betting on horseracing I am having a helluva lot fun trying too.

[Tom]

What CJ said was absolutely true.
The slower the race goes, the more time it takes a horse to cover a length.
If you don't know what horse players are talking about, maybe you shouldn't jump in with your unsubstantiated theories. I would say of anyone here, CJ is the one who is walking the talk every day.

[ReplayRandall]

What CJ said was absolutely true.
The slower the race goes, the more time it takes a horse to cover a length.
If you don't know what horse players are talking about, maybe you shouldn't jump in with your unsubstantiated theories. I would say of anyone here, CJ is the one who is walking the talk every day.

That is true about CJ, but like all serious handicappers, we're all prone to slumps and being just simply infallible, no matter our pathway.....

[steveb]

Thanks for the reply, but you are talking in circles and if I don't make any money in betting on horseracing I am having a helluva lot fun trying too.


:lol: 'i' am talking in circles! :lol:

you my friend, talk in riddles, but it's fine, you can keep on pretending.
i have the same problem where cate blanchett is concerned!

[cj]

:lol: 'i' am talking in circles! :lol:

you my friend, talk in riddles, but it's fine, you can keep on pretending.
i have the same problem where cate blanchett is concerned!

You, sir, do not post nearly enough!

[Dark Target]

You, sir, do not post nearly enough!

Agree. But it's staggering that some would rather argue with posters (not just steveb) who have forgotten more than most know on a particular subject rather than try and learn from them.