Horse Racing Forum - PaceAdvantage.Com - Horse Racing Message Board

Go Back   Horse Racing Forum - PaceAdvantage.Com - Horse Racing Message Board


Horse Racing Forum - PaceAdvantage.Com - Horse Racing Message Board

Horse Racing Forum - PaceAdvantage.Com - Horse Racing Message Board (http://www.paceadvantage.com/forum/index.php)
-   General Handicapping Discussion (http://www.paceadvantage.com/forum/forumdisplay.php?f=2)
-   -   A Critique of Speed: Fallacies in Creating Accurate Speed Figures (http://www.paceadvantage.com/forum/showthread.php?t=148582)

Nitro 11-08-2018 01:36 AM

A Critique of Speed: Fallacies in Creating Accurate Speed Figures
 
There's been quite a bit of interest recently in threads titled “Speed Figure Comparisons”, “Distance Between Winner & Place Horse”, and “How Long is a Length”. All have a common theme that revolves around the attempt to quantify a horse’s performance based on time.

I was recently digging through some of my old speed handicapping notes from many years ago and I came across some things I thought I had discarded. I thought I might pass on a brief summary of why I shelved Speed handicapping. There could be others who might also be somewhat skeptical about trying to develop truly accurate Speed profiles or figures.

Before I get into it I’d like to preface this scenario by stating that we can all agree that there are many Classes of horses that will also perform at various speed level benchmarks. In other words, the classier animals will generally run faster at any given distance from gate to wire.

With that in mind I’m going use some average time fractions from a typical 6F race on a fast dirt track for a mid-level claiming race at a reasonably sized race track.

I thought an appropriate Summary Title might read:
Fallacies in Creating Accurate Speed Figures.

So, let’s start off by arranging and combining some basic related facts and conditions:
1) A 6F race is ¾ of a mile long which equates to 3,960 feet.
2) There are 3 primary points of call:
¼ mile (1320 feet), ½ mile (2640 feet), and finish - 3/4 mile (3960 feet)
3) There is no consideration taken for the track condition (variant) other than it being rated generically listed as fast without any wind.

The Race Times:

The raw time fractions used for each point of call will be generally recognized as:
¼ mile) - 22 & 3 (22.6) sec
½ mile) - 46 & 4 (46.8) sec
¾ mile) - 1:12 flat (72) sec

Factual summary:
It took the leading horse 22.6 seconds to reach the ¼ m. pole.
It took the leading horse an additional 24.2 seconds to reach the 1/2m. pole.
It took the leading horse an additional 25.2 seconds to reach the finish line.
The only portion of the race where obvious acceleration occurs is during the first ¼ mile.
The remaining two ¼ mile intervals indicate an obserable deceleration.

Controversial Premise:
So, to arrive at a time figure based on beaten lengths you might have already noticed that you simply can’t use constant numerical value like 1 length = 1/5 (0.20) sec. Why? Because any numerical constant would be based on merely utilizing an overall average velocity which is not the case as demonstrated above. Now let’s keep in the back of our mind what many seem to believe to be what a typical “length” should equal: Somewhere between 8 and 10 feet.

Establishing the “Beaten Length”
If a ¼ m = 1320 ft and it’s covered in 22.6 sec than the final avg velocity = 58.4 ft/sec
Dividing this velocity by a 1/5th of a second = 11.68 feet
If the next ¼ m is covered in 24.2 sec than the final avg velocity = 54.5 ft/sec
Dividing this velocity by a 1/5th of a second = 10.90 feet
If the final ¼ m is covered in 25.2 sec than the final avg velocity = 52.3 ft/sec
Dividing this velocity by a 1/5th of a second = 10.46 feet

By comparison, if you were to mistakenly think that the average velocity was based on just the overall final time of 72 sec (1min,12sec) for 3960 feet = 55 ft/sec which would equate to a length being equal to 11 feet.

Class Consideration

Without going any further, we have to also keep in mind that we’re just using an average Class of horse in this example. As the Class rating improves, so do the times at each point of call, which again would not only increase the average final velocity at each call, but the length equivalent as well. (And visa-versa)

This is one of the reasons why horses moving up and down the Class ladder at various times during their racing careers are affected by the new average velocities at each point of call that they must face in each race. Of course that’s both good and bad depending on how much the Class rating changes.

The Beaten Length Fantasy
The proposed clarification of a beaten length dimension is all well and good and should be considered. However, the genuine crux of the matter for attempting to establish truly accurate time figures for all entries in a race lies not only in utilizing the proper beaten length factor for a specific portion of the race, but an underlying factor that seems to be completely overlooked or maybe even ignored.

I call it the "Phantom Dynamic".

The easiest explanation for the recognition of this Phantom Dynamic is to think about what’s commonly being done to establish an accepted time figure for each entry that’s behind the horse at any point of call. Basically, a static snapshot is taken when any leading horse reaches a specific predetermined spot on the race course. Simultaneously an observed measured distance (beaten length) is established for each of the trailing entries. This stationary measurement completely disregards the running dynamic velocity of any individual entry at that moment. In point of fact, some of the entries may be accelerating while others are decelerating. This technique simply presumes (erroneously) that each entry being examined will pass that SAME point of call based on a static beaten length which is based on the velocity of the leading horse. This couldn’t be further from the truth because “acceleration” or “deceleration” is defined as a dynamic change in velocity over a specific period of time.

An example would be an entry, say 4 lengths behind a leading horse at the ½ mile pole. From the race times used above the leading horse has reached this position in 46.8 seconds and has a final average velocity of 54.4 ft/sec. If we accept that a beaten length during that portion of the race is 10.9 feet it would mean that this horse would cross the same point 0.8 seconds later or in 47.6 seconds. However, if this same pursuing horse happens to be accelerating during that portion of the race with an average velocity of 58 ft/sec. its actual time would be 47.5 seconds. This same scenario could also be applied to the final ¼ mile of any race. A 10th of second doesn’t sound like very much of a difference, but at 58 ft/sec it equates to 5.8 feet. How many races are won (or lost) by less of a margin (approx ½ a length).

Conclusion:
The actual measurement of a beaten length can vary at various fixed points in a race due to changes in the average velocities of the leading horse. However, computing the actual time of a trailing horse at any point of call should not be dependent on its beaten lengths in conjunction with the speed of the leading horse but rather on its own average velocity. The problem becomes actually measuring that velocity. Multiplying this Phantom Dynamic by the number of points of call can certainly alter the calculated Speed profile of any entry.

steveb 11-08-2018 03:38 AM

my simple observation is that you don't understand as well as you think you do.


full points though, for going to the trouble of so much detail, which in turn gives me the confidence to say the above.


just one point......there is NO completely accurate figure, there's just figures that enable one to allot probs better than most others do(or could), nothing more.

vegasone 11-08-2018 06:09 AM

Trying to put an exact number in a scenario of so many changing factors helps keep most people busy and out of trouble. Good observation and explanation though..:bang:


Think of a car race.

bobphilo 11-08-2018 07:45 AM

Quote:

Originally Posted by Nitro (Post 2393436)
There's been quite a bit of interest recently in threads titled “Speed Figure Comparisons”, “Distance Between Winner & Place Horse”, and “How Long is a Length”. All have a common theme that revolves around the attempt to quantify a horse’s performance based on time.

I was recently digging through some of my old speed handicapping notes from many years ago and I came across some things I thought I had discarded. I thought I might pass on a brief summary of why I shelved Speed handicapping. There could be others who might also be somewhat skeptical about trying to develop truly accurate Speed profiles or figures.

Before I get into it I’d like to preface this scenario by stating that we can all agree that there are many Classes of horses that will also perform at various speed level benchmarks. In other words, the classier animals will generally run faster at any given distance from gate to wire.

With that in mind I’m going use some average time fractions from a typical 6F race on a fast dirt track for a mid-level claiming race at a reasonably sized race track.

I thought an appropriate Summary Title might read:
Fallacies in Creating Accurate Speed Figures.

So, let’s start off by arranging and combining some basic related facts and conditions:
1) A 6F race is ¾ of a mile long which equates to 3,960 feet.
2) There are 3 primary points of call:
¼ mile (1320 feet), ½ mile (2640 feet), and finish - 3/4 mile (3960 feet)
3) There is no consideration taken for the track condition (variant) other than it being rated generically listed as fast without any wind.

The Race Times:

The raw time fractions used for each point of call will be generally recognized as:
¼ mile) - 22 & 3 (22.6) sec
½ mile) - 46 & 4 (46.8) sec
¾ mile) - 1:12 flat (72) sec

Factual summary:
It took the leading horse 22.6 seconds to reach the ¼ m. pole.
It took the leading horse an additional 24.2 seconds to reach the 1/2m. pole.
It took the leading horse an additional 25.2 seconds to reach the finish line.
The only portion of the race where obvious acceleration occurs is during the first ¼ mile.
The remaining two ¼ mile intervals indicate an obserable deceleration.

Controversial Premise:
So, to arrive at a time figure based on beaten lengths you might have already noticed that you simply can’t use constant numerical value like 1 length = 1/5 (0.20) sec. Why? Because any numerical constant would be based on merely utilizing an overall average velocity which is not the case as demonstrated above. Now let’s keep in the back of our mind what many seem to believe to be what a typical “length” should equal: Somewhere between 8 and 10 feet.

Establishing the “Beaten Length”
If a ¼ m = 1320 ft and it’s covered in 22.6 sec than the final avg velocity = 58.4 ft/sec
Dividing this velocity by a 1/5th of a second = 11.68 feet
If the next ¼ m is covered in 24.2 sec than the final avg velocity = 54.5 ft/sec
Dividing this velocity by a 1/5th of a second = 10.90 feet
If the final ¼ m is covered in 25.2 sec than the final avg velocity = 52.3 ft/sec
Dividing this velocity by a 1/5th of a second = 10.46 feet

By comparison, if you were to mistakenly think that the average velocity was based on just the overall final time of 72 sec (1min,12sec) for 3960 feet = 55 ft/sec which would equate to a length being equal to 11 feet.

Class Consideration

Without going any further, we have to also keep in mind that we’re just using an average Class of horse in this example. As the Class rating improves, so do the times at each point of call, which again would not only increase the average final velocity at each call, but the length equivalent as well. (And visa-versa)

This is one of the reasons why horses moving up and down the Class ladder at various times during their racing careers are affected by the new average velocities at each point of call that they must face in each race. Of course that’s both good and bad depending on how much the Class rating changes.

The Beaten Length Fantasy
The proposed clarification of a beaten length dimension is all well and good and should be considered. However, the genuine crux of the matter for attempting to establish truly accurate time figures for all entries in a race lies not only in utilizing the proper beaten length factor for a specific portion of the race, but an underlying factor that seems to be completely overlooked or maybe even ignored.

I call it the "Phantom Dynamic".

The easiest explanation for the recognition of this Phantom Dynamic is to think about what’s commonly being done to establish an accepted time figure for each entry that’s behind the horse at any point of call. Basically, a static snapshot is taken when any leading horse reaches a specific predetermined spot on the race course. Simultaneously an observed measured distance (beaten length) is established for each of the trailing entries. This stationary measurement completely disregards the running dynamic velocity of any individual entry at that moment. In point of fact, some of the entries may be accelerating while others are decelerating. This technique simply presumes (erroneously) that each entry being examined will pass that SAME point of call based on a static beaten length which is based on the velocity of the leading horse. This couldn’t be further from the truth because “acceleration” or “deceleration” is defined as a dynamic change in velocity over a specific period of time.

An example would be an entry, say 4 lengths behind a leading horse at the ½ mile pole. From the race times used above the leading horse has reached this position in 46.8 seconds and has a final average velocity of 54.4 ft/sec. If we accept that a beaten length during that portion of the race is 10.9 feet it would mean that this horse would cross the same point 0.8 seconds later or in 47.6 seconds. However, if this same pursuing horse happens to be accelerating during that portion of the race with an average velocity of 58 ft/sec. its actual time would be 47.5 seconds. This same scenario could also be applied to the final ¼ mile of any race. A 10th of second doesn’t sound like very much of a difference, but at 58 ft/sec it equates to 5.8 feet. How many races are won (or lost) by less of a margin (approx ½ a length).

Conclusion:
The actual measurement of a beaten length can vary at various fixed points in a race due to changes in the average velocities of the leading horse. However, computing the actual time of a trailing horse at any point of call should not be dependent on its beaten lengths in conjunction with the speed of the leading horse but rather on its own average velocity. The problem becomes actually measuring that velocity. Multiplying this Phantom Dynamic by the number of points of call can certainly alter the calculated Speed profile of any entry.

There is a more accurate method of deriving a horse's time at a given point of call though it takes a bit more work. It's based on the velocity of the that segment of the race. When you know how fast they are going at that segment you can simply calculate the horses time based on distance behind and that velocity. Of course there is still some error possible but only when the horse involved and leader are going at significantly different velocities.

Actually, now that more and more tracks are using satellite tracking and GPS systems that give each horse's position, ground loss and exact time at every given call down to each furlong, the entire problem of beaten length conversion to time goes away. You must realize that as the technology improves so does the accuracy. We are not back in the "bad old day" days before methods of calculating times and speed/pace figures were less accurate.

In addition modern research in exercise physiology has produced much valuable information on evaluating methods of energy distribution with differing pace patterns. The problem is that few handicappers are incorporating this into their handicapping, using methods based on outdated
traditional lore. Fortunately, the information is increasingly there for those willing to study and apply it, giving them an advantage.

Nitro 11-08-2018 11:14 AM

Quote:

Originally Posted by bobphilo (Post 2393468)
There is a more accurate method of deriving a horse's time at a given point of call though it takes a bit more work. It's based on the velocity of the that segment of the race. When you know how fast they are going at that segment you can simply calculate the horses time based on distance behind and that velocity. Of course there is still some error possible but only when the horse involved and leader are going at significantly different velocities.

There in lies the problem I’m talking about.
(It's not how fast "they are going" because at any point of call you only know how fast the lead horse is going.
You can’t assume that the horses trailing are traveling at the same average velocity of the horse arriving at any point of call.

Please re-read the last 3 paragraphs I posted.

cj 11-08-2018 11:36 AM

Making a mountain out of a molehill here in my opinion. Yes, it isn't perfectly accurate. It also isn't as bad as is being made out either. The problems can be minimized to an extent, and in the grand scheme of things, exact times at the fractional calls aren't overly important.

At the finish, beaten lengths as listed are irrelevant. I know what each horse ran to the hundredth. All you have to do is figure out the conversion used by the photo company at the track and it is easy math. Beaten lengths at this point (and also at the fractions for Trakus and the new GPS systems) are simply a conversion of time to a number representing the time gap. The actual times are available for each horse if you know how to look.

Nitro 11-08-2018 12:25 PM

Quote:

Originally Posted by cj (Post 2393618)
Making a mountain out of a molehill here in my opinion. Yes, it isn't perfectly accurate. It also isn't as bad as is being made out either. The problems can be minimized to an extent, and in the grand scheme of things, exact times at the fractional calls aren't overly important.

I’m actually surprised that one of few people who is recognized as someone who develops these speed figures and supports their reputation for many who use them doesn’t appreciate the significance of the early fractions.

I believe that everything about a race in general is about the start. During those early fractions as the horses are accelerating (in particular) lots of things can take place (like bumping, blocking, changing lanes, etc.) which certainly affect their speed during any portion of a race. How that speed (or energy) is rationed out during the race can very often determine its outcome.

In my estimation, the entire purpose of generating somewhat accurate speed figures was to have the ability to make speed comparisons between different entries having run in different races that are now competing against each other. I would think that they also might provide some insight as to how a race might shape up based on the running styles of each entry and how their past speed profile at each point of call would support that style against the competition.

Dave Schwartz 11-08-2018 12:36 PM

Quote:

Originally Posted by cj (Post 2393618)
Making a mountain out of a molehill here in my opinion. Yes, it isn't perfectly accurate. It also isn't as bad as is being made out either. The problems can be minimized to an extent, and in the grand scheme of things, exact times at the fractional calls aren't overly important.

At the finish, beaten lengths as listed are irrelevant. I know what each horse ran to the hundredth. All you have to do is figure out the conversion used by the photo company at the track and it is easy math. Beaten lengths at this point (and also at the fractions for Trakus and the new GPS systems) are simply a conversion of time to a number representing the time gap. The actual times are available for each horse if you know how to look.

CJ,

My view is:

Quote:

"At the end of the day, speed and pace ratings are simply metrics. Some will be better than others and have a greater correlation to reality than others."
What do you think of that statement?

Of course, the logical question is "How does one improve them?"

Dave

bobphilo 11-08-2018 12:56 PM

Quote:

Originally Posted by Nitro (Post 2393601)
There in lies the problem I’m talking about.
(It's not how fast "they are going" because at any point of call you only know how fast the lead horse is going.
You can’t assume that the horses trailing are traveling at the same average velocity of the horse arriving at any point of call.

Please re-read the last 3 paragraphs I posted.

You did not read the rest of my paragraph when I noted, "Of course there is still some error possible but only when the horse involved and leader are going at significantly different velocities.". No this method is not perfect but is one improvement over just using the more traditional method and yields very good results.
In fact, you can determine a horses velocity even more accurately by comparing it's lengths gained or lost between calls in relation to the leader. Actually even the quick and dirty method of just multiplying the beaten lengths by .167 yielded good results for me for many previous years.

As l also noted there are even more accurate methods to measure precise times using GPS and satellite technology to get exact position, ground loss and times for every horse at every point of call. In addition, cj uses a precise method to that is accurate to the hundreth of a second, but apparently either you ignore these or will just never be satisfied. There is no such thing as perfection in the real world. Something does not have to be perfect to be effective.

The main point is that speed and pace figures using the above methods are accurate, though not perfect, assessments of a horse's performance and most importantly the single best method of rating performances. Particularly when combined with other handicapping factors. This will reduce the small imperfections that you point out when one uses speed/pace figures alone. By implying that figure handicappers are assuming using their figures alone, because they think they are perfect, you are constructing a straw man that is not typical of a good figure handicapper.

jay68802 11-08-2018 01:17 PM

Quote:

Originally Posted by Dave Schwartz (Post 2393658)
CJ,

Of course, the logical question is "How does one improve them?"

Dave

Accurate timing for each horse, including the run up in opening fraction and total distance of the race.

Nitro makes some valid points. I would say nothing new, at least to those that have studied speed figures. There are flaws in any handicapping method, including tote board methods.

Speed and pace figures are basically just a still frame of a movie that has been run in the past. But, to me they give you a general idea of the talent level of the horse. How that horse will run to day is a guess. Clues to that can be found in the past performance as well as the odds board.

Nitro 11-08-2018 01:32 PM

Quote:

Originally Posted by bobphilo (Post 2393669)
The main point is that speed and pace figures using the above methods are accurate, though not perfect, assessments of a horse's performance and most importantly the single best method of rating performances. Particularly when combined with other handicapping factors. This will reduce the small imperfections that you point out when one uses speed/pace figures alone. By implying that figure handicappers are assuming using their figures alone, because they think they are perfect, you are constructing a straw man that is not typical of a good figure handicapper.

Bob, I agree with much of what you’ve mentioned, but I never implied that anyone should rely on just speed figures ALONE.
My critique was simply attempting to point out a hidden flaw in speed figure development. It’s one that is very difficult to solve unless of course you can actually time every horse individually in a race at each point of call instead of backing into the time using beaten lengths off of the horse hitting the point of call first.

JJMartin 11-08-2018 01:46 PM

Nitro, excellent post. Your detailed explanation shows indisputable mathematical proof.

Tom 11-08-2018 02:01 PM

All that is missing is how good do the figures work to predict winners.

Seems to be the ONLY criteria that matters.
And it is ignored.
Nothing was prived here.
Lots of assumptions were made.
Nothing more.

Nitro 11-08-2018 02:12 PM

Quote:

Originally Posted by jay68802 (Post 2393674)
Accurate timing for each horse, including the run up in opening fraction and total distance of the race.

Nitro makes some valid points. I would say nothing new, at least to those that have studied speed figures. There are flaws in any handicapping method, including tote board methods.

Speed and pace figures are basically just a still frame of a movie that has been run in the past. But, to me they give you a general idea of the talent level of the horse. How that horse will run to day is a guess. Clues to that can be found in the past performance as well as the odds board.

Well said Jay!
I completely agree that we certainly don’t live in a perfect world, least of which is the horse racing game. How we individually apply our tools and assets to the betting arena in the end will determine our success or failure. It was tough decision to give up on the speed aspects of the game at the time, but I believe that as an Outsider I’ve now got more than just “clues” as to who the contenders in a race might be.

Jeff P 11-08-2018 02:30 PM

I'm glad Jay mentioned the runup.

You know --

The distance between the starting gate and the point in the race where official timing of the race actually begins.

Looking at runup data for the past 30 days --

I see a 1 1/16 mile turf race at Indiana Grand on 10-17-2018 with a runup listed at 229 feet.

I see a second race that same day same turf course same distance with a runup of 219 feet.

The Equibase data for the first race has the leader's opening 1/4 mile fraction listed at 21.95 seconds.

The Equibase data for the second race has the leader's opening 1/4 mile fraction listed at 22.04 seconds.

At first glance you might think "Wow the turf course was really fast that day."

(Maybe it was.)

But if you knew the horses were given 70 yards to get up to full speed before timing of the race began - you might have a better understanding of what actually happened that day.

The one thing that bothers me most about runups?

While runup distance is available in the Equibase data --

Time of the runup is not.




-jp

.


All times are GMT -4. The time now is 03:11 PM.

Powered by vBulletin® Version 3.8.9
Copyright ©2000 - 2024, vBulletin Solutions, Inc.
Copyright 1999 - 2023 -- PaceAdvantage.Com -- All Rights Reserved

» Advertisement
Powered by vBadvanced CMPS v3.2.3

All times are GMT -4. The time now is 03:11 PM.


Powered by vBulletin® Version 3.8.9
Copyright ©2000 - 2024, vBulletin Solutions, Inc.
Copyright 1999 - 2023 -- PaceAdvantage.Com -- All Rights Reserved
We are a participant in the Amazon Services LLC Associates Program, an affiliate advertising program
designed to provide a means for us to earn fees by linking to Amazon.com and affiliated sites.