View Full Version : Beyer speed chart question
Ever notice that using the Beyer charts, the track is ALWAYS faster for sprints than for routes. What I mean is his variant always shows the track faster for sprints, and by always I mean about 98% of the time, if not more.
Doesn't this mean there is something wrong with the charts in the first place? The chart was supposed to show equal performances at different distances.
Ever notice that using the Beyer charts, the track is ALWAYS faster for sprints than for routes. What I mean is his variant always shows the track faster for sprints, and by always I mean about 98% of the time, if not more.
Doesn't this mean there is something wrong with the charts in the first place? The chart was supposed to show equal performances at different distances.
Just guessing at possible reasons ... but maybe this has something to do with it.
"Fact": When a track is slower than par, it will be more slow per furlong as the distance of the race increases.
Assumption: Suppose that he uses track records as par times - in that case it would not be a huge surprise to find variants that are faster for sprints.
Greyfox
08-08-2011, 06:11 PM
The DRF variants, which a lot of people knock here, always show that the route variants are about 50 % higher than the sprint variants for the same day.
eg. Sprint variant 18
Route variant 27
Just guessing at possible reasons ... but maybe this has something to do with it.
"Fact": When a track is slower than par, it will be more slow per furlong as the distance of the race increases.
Assumption: Suppose that he uses track records as par times - in that case it would not be a huge surprise to find variants that are faster for sprints.
I understand, but the speed chart is supposed to take care of the difference in distances. A length at 6 furlongs is worth 2.5 points, while a length at 9 furlongs is only worth 1.8 points. He does not use track records for his times on the speed chart.
The DRF variants, which a lot of people knock here, always show that the route variants are about 50 % higher than the sprint variants for the same day.
eg. Sprint variant 18
Route variant 27
Those are based on the best time at the track in the last last three years, so that makes sense. It really doesn't apply to Beyer charts.
I'm wondering if routers have been getting worse and worse over the years, and the lower variants are an attempt to keep things similar.
aaron
08-08-2011, 07:32 PM
Those are based on the best time at the track in the last last three years, so that makes sense. It really doesn't apply to Beyer charts.
I'm wondering if routers have been getting worse and worse over the years, and the lower variants are an attempt to keep things similar.
I think routers have been getting worse. I don't remember the last real good horse that a full campaign or 2. It seems the routers have maybe one or two big races,so with that as a basis,I think they are slower. On the other hand,the thorograph sheets show more negative numbers now than they did 10-15 years ago,so Jerry Brown has them faster and he certainly knows more than I do.
I think routers have been getting worse. I don't remember the last real good horse that a full campaign or 2. It seems the routers have maybe one or two big races,so with that as a basis,I think they are slower. On the other hand,the thorograph sheets show more negative numbers now than they did 10-15 years ago,so Jerry Brown has them faster and he certainly knows more than I do.
He is the only one that has them faster, and it doesn't hurt his own interests to have horses getting faster either.
I'm not saying I know for sure that they are getting worse, I'm just trying to understand how the charts many learned to make figures on don't ever seem to mesh with reality.
maddog42
08-08-2011, 09:12 PM
The DRF variants, which a lot of people knock here, always show that the route variants are about 50 % higher than the sprint variants for the same day.
eg. Sprint variant 18
Route variant 27
Greyfox you are quite right. The deceleration curve is much steeper for routes.
One of the reasons I gave up speed handicapping many years ago, was that i could not make sense of the track variant. I wasn't very good at it, but I did notice that if the sprints were running fast 6, the routes would probably be fast 4. If the sprints were running slow 5 the routes would be slow 8. That is on the GOOD days when things made sense. I was trying to make pars for Remington and a few other tracks and the equitrack in 1990 that RP was using gave me fits. I believe Ragozin or someone mentions something like this.
Cj if you say the route variant is slower about 98% of the time, then my Hat is off to you . I was never this consistent, nor did I have that tight of a model. ( I love tight models!) It probably means you are doing things correctly. The nonlinear nature of racing and pars, was the reason I went to percentage of par while making Pace Constructs. It is definitely not perfect,
but gets closer to reality.
dnlgfnk
08-08-2011, 09:34 PM
"Greyfox you are quite right. The deceleration curve is much steeper for routes"...
And yet when I became frustrated and sensed something not right, it was from routes, but also 5f dashes where I surmised that the intense pace contributed to deceleration which threw the 6f relationship out of sync. I've arrived at a place where each different distance is a world unto itself.
"The nonlinear nature of racing and pars, was the reason I went to percentage of par while making Pace Constructs"...
It's a major reason I went to trip handicapping, allowing the crowd to set largely figure based odds on horses and trying to exploit public error with trips.
CJ, have the charts been ajar for months or over a year? It would seem the database guys would make corrections after a time. I know that they periodically review projected figures in order to avoid the inevitable "shrinking figure" problem that always plagued me, from not giving winners enough credit for a bigger number when projecting. With time, I was projecting based on lower than real figures, continually lowering the figures until May and September would look like 2 different horse colonies.
CJ, have the charts been ajar for months or over a year? It would seem the database guys would make corrections after a time. I know that they periodically review projected figures in order to avoid the inevitable "shrinking figure" problem that always plagued me, from not giving winners enough credit for a bigger number when projecting. With time, I was projecting based on lower than real figures, continually lowering the figures until May and September would look like 2 different horse colonies.
The variants have fallen this way for years now.
PhantomOnTour
08-09-2011, 12:45 AM
Just guessing at possible reasons ... but maybe this has something to do with it.
"Fact": When a track is slower than par, it will be more slow per furlong as the distance of the race increases.
Assumption: Suppose that he uses track records as par times - in that case it would not be a huge surprise to find variants that are faster for sprints.
Above is the answer imo.
Also, horses are able to run closer to par (or the TR) at shorter distances than longer ones.
Coming within 2 seconds of the world record for 6f is nothing special, but coming within 2 seconds of the world record at 2.5miles is quite an accomplishment.
I hear what CJ is saying about a flaw in the charts themselves...if the sprints and routes keep coming up different then maybe one or both of their base starting points/numbers is flawed...is the famous parallel time chart wrong?
A question regarding Beyers: is the 6f par fig for An1x 3+ the same as the 9f par fig? IOW, is the An1x par the same for all distances?
I should know this answer but it's been a long time since I messed with Beyer style figs.
dnlgfnk
08-09-2011, 01:37 AM
If the pars were different, that would be a departure from the approach Beyer has always described.
None of these problems should theoretically exist. The pars and equivalent times at different distances are all based upon large amounts of data from individual tracks under consideration. Sprints are almost universally similar, but two turn routes are track specific. Surely the comparative times are based upon the average time at that track, at that class, at that distance.
Apparently something goes amiss in the day to day variant assessment, however. I suspect that a large sampling of two turn route par times becomes less relevant when one is dealing with say, 2 or 3 routes and 6 sprints on a particular race day.
KingChas
08-09-2011, 02:29 AM
Interesting comment's from Steve Crist's Spa blogs on Sprints vs Routes.
One of his favorite betting angles.
"What did I like about Unknown Wonder in the 5th? Well, you know me and turnbacks (8f to 5.5f), but it wasn't just the fact of the turnback, it was that Unknown Wonder made a wild premature middle move in that mile race and was actually right there after 5.5f though beaten 11 at the finish. This kind of horse often gets ignored because people look only at the final margin and figure rather than the relevant portion of the race as it applies to today's distance."
So I would say the final Beyers are not comparable Sprint vs. Route.
proximity
08-09-2011, 05:43 AM
I'm wondering if routers have been getting worse and worse over the years, and the lower variants are an attempt to keep things similar.
perhaps modern chemistry is helping the sprinters get (relatively) better?
maybe today's juice is helping them decelerate slower by improving their cardio but doesn't have the same influence on the localized muscular endurance that the longer races call for?
just some theories.
Robert Goren
08-09-2011, 06:43 AM
This leads me to believe what I have suspected for a long a time. In races for lower claiming prices the winner of a sprint is a far better animal than the winner of a route race at least in the short run.
JeremyJet
08-09-2011, 06:55 AM
Ever notice that using the Beyer charts, the track is ALWAYS faster for sprints than for routes. What I mean is his variant always shows the track faster for sprints, and by always I mean about 98% of the time, if not more.
Doesn't this mean there is something wrong with the charts in the first place? The chart was supposed to show equal performances at different distances.
What Beyer variant? The one you can figure out from Simulcast Weekly/Daily? I'm surprised you ask this question if that's what you're referring to.
When I do Beyer pars for a track, I always get a different number for the same class at sprints and routes. DRF always gives one par for both and I never see it in the real world.
Quinn's book, On Track Off Track has a Quirrin -style class chart with Beyer pars in it and and he has built in a slowing down by distance. I have used that model to "equalize" the numbers, say $50,000 claiming is a 96 at 6 and only a 92 at a mile and a sixteenth, I would use a +4 or -4 adjustment in my handicapping.
Perhaps a model like that could be used to re-set the basic numbers on the chart.
What Beyer variant? The one you can figure out from Simulcast Weekly/Daily? I'm surprised you ask this question if that's what you're referring to.
Anyone can figure out the variant with PPs in hand and the Beyer charts which are readily available in books and on the web.
aaron
08-09-2011, 09:18 AM
He is the only one that has them faster, and it doesn't hurt his own interests to have horses getting faster either.
I'm not saying I know for sure that they are getting worse, I'm just trying to understand how the charts many learned to make figures on don't ever seem to mesh with reality.
I don't know why having horses run faster numbers is in Brown's interest,but I do know that the classic horses of the past were better horses than the classic horses of today. A horse like Forego never lost unless he was giving weight. Cigar was not considered fast on the numbers,but I find it hard to believe that overall he was not better than anything running today.Skip Away at his best was probably better than today's horses.
What I see from today's horses is the occasional fast race and then they are generally done for a while.
If there was a way to do numbers for races of many years ago,you would probably have your answer.
Valuist
08-09-2011, 11:00 AM
According to the theory, I thought Beyer espoused having different charts for every track, as the relationship between a sprint like 6f and 1 1/16 miles will vary widely track to track.
I think it makes more sense just to make different variants for sprints and routes. I've never been a believer that you can accurately use both 6 furlong races and 9 furlong races to make a single variant. Wind and track maintenance will always make those relationships shaky at best.
According to the theory, I thought Beyer espoused having different charts for every track, as the relationship between a sprint like 6f and 1 1/16 miles will vary widely track to track.
I think it makes more sense just to make different variants for sprints and routes. I've never been a believer that you can accurately use both 6 furlong races and 9 furlong races to make a single variant. Wind and track maintenance will always make those relationships shaky at best.
I agree about making separate variants. I'm not saying they should be the same and usually they aren't. However, if the data he used to make the original charts was accurate, sometimes routes would be faster than sprints. That virtually NEVER happens.
PhantomOnTour
08-09-2011, 11:56 AM
So you are getting variants like this?
sprints: F6
routes: F2
Looking at my Mth figs and Sar figs for the current meets (skipped Bel cuz of the one turn route thing) and found most days to be in sync.
I will have final time variants like:
Spr: F4
Rte: F7
I dug up a regular day at Mth with 4 sprints and 4 routes on a fast dirt track...July 17th.
The pace calls and final times are clearly faster for routes than sprints, no?
I know this is just one day, but most follow this dynamic.
So you are getting variants like this?
sprints: F6
routes: F2
Looking at my Mth figs and Sar figs for the current meets (skipped Bel cuz of the one turn route thing) and found most days to be in sync.
I will have final time variants like:
Spr: F4
Rte: F7
I dug up a regular day at Mth with 4 sprints and 4 routes on a fast dirt track...July 17th.
The pace calls and final times are clearly faster for routes than sprints, no?
I know this is just one day, but most follow this dynamic.
Are you using his speed charts, or your own?
I don't store Beyers variants, but I would be very surprised if the routes were rated faster than sprints. I'll look it up later and see, but it is a very rare event when it happens.
The whole point of both of these threads (this and the Carroll one) is that I'm trying to find out if the logic used in making these "speed charts" matches up to the real world. When they were made, the data was much tougher to gather and study than it is now. So far what I'm finding is that they have a lot of flaws, yet lots of figure systems are built on them.
I am not knocking either guy, or any others. They worked with the tools they had at the time.
PhantomOnTour
08-09-2011, 12:09 PM
Using my own.
Horses used to 'crack' or step on their pedigree when trying to get that 10th furlong, maybe now they crack trying to get that 9th furlong???
Are they just slowing down faster than they used to slow down?
According to the theory, I thought Beyer espoused having different charts for every track, as the relationship between a sprint like 6f and 1 1/16 miles will vary widely track to track.
I almost forgot about this part. I would say "why?" Sure, if tracks are built differently or have big differences in run ups, it might make sense. But if you have two mile tracks with pretty much the same type turns, there shouldn't be a big difference. I think it might have been a case of making the data fit rather than using it objectively.
classhandicapper
08-09-2011, 12:18 PM
IMO there are several possibilities.
1. We have been breeding speed to speed for so long we may be producing a higher percentage of faster horses that aren't as good at stretching out.
2. Perhaps some of the tweaking of surfaces to make them safer by giving them deeper cushions or other characteristics is changing the relationships between distances. This is somewhat related to point someone made earlier in the thread that as surfaces slow or speed up they may not change equally from distance to distance.
3. The Beyer charts were never right to begin with. This is probably the least likely, but it is possible. We may just be noticing it now because so many people have access to way more data to check.
4. The relationships between distances might be different for Grade 1 horses with deeper reserves of stamina than for cheaper horses like low level claimers and maiden claimers that are often staggering after a half mile. At some tracks, the quality has clearly declined.
#1 and #2 are becoming a bit of pet peeve of mine. Most modern handicappers tend to judge horses almost entirely on speed (namely speed figures), but stamina/racing reserve, determination etc... is what often separates the very best horses in the biggest races.
A lot of horses can run really fast when they face an easy field, get the perfect pace, shake loose, catch a racing strip that accents their speed etc... But when you throw a bunch of graded stakes horses together there typically are no easy trips. The pace is usually competitive, lots of horses make moves so they have to be used to get, maintain and improve their position, and the finish also usually involves several horses making strong bids. Those conditions don't just test speed. They test what you actually have in the tank. Perhaps by breeding speed to speed for so long we are getting a lot of horses that can run really fast when conditions suit them, but we aren't getting as many GREAT horses using a broader definition.
Since a speed point represents a different percentage of the total by distance, should we change the charts to have each point represent, say, 1% of the total race instead of a fifth?
Wish I could figure where I put my Beyer book since I think I would have some thoughts about this.
Since a speed point represents a different percentage of the total by distance, should we change the charts to have each point represent, say, 1% of the total race instead of a fifth?
There are a lot of ways to do it, but what is the "total race" that we use? How do we find equal performances at different distances. That is the biggest dilemma I think.
Robert Goren
08-09-2011, 01:10 PM
One of things that I was looking at before life got in the way of horse racing many years ago was beaten lengths by the median horse at each distance to decide what a point should be worth at each distance.
classhandicapper
08-09-2011, 01:15 PM
I almost forgot about this part. I would say "why?" Sure, if tracks are built differently or have big differences in run ups, it might make sense. But if you have two mile tracks with pretty much the same type turns, there shouldn't be a big difference. I think it might have been a case of making the data fit rather than using it objectively.
This could also be a surface issue.
We know that tracks can sometimes be speed favoring or bad for speed. It doesn't take much of a leap to think the relationship between distances might change at the extremes also.
If two different tracks have slightly different speed favoring profiles due to differences in their surfaces, the relationships between distances might be slightly different even if the turns, run up etc... are the same.
We might not notice it much when comparing 6F to 7F because it's very small (even though it's there), but as distances stretch out to routes it might be large enough to be clear.
JeremyJet
08-09-2011, 01:15 PM
Anyone can figure out the variant with PPs in hand and the Beyer charts which are readily available in books and on the web.
I understand what you're trying to do, but you're missing something.
I'm surprised you ask this question.
Good luck to you and your subscribers.
I understand what you're trying to do, but you're missing something.
By all means, fill me in. I'm not claiming to have all the answers. I'd love some input. If I didn't want it I wouldn't have started the thread.
classhandicapper
08-09-2011, 01:30 PM
I understand what you're trying to do, but you're missing something.
Good luck to you and your subscribers.
I think what CJ is hinting at in a diplomatic way is that the relationship between sprints and routes may have changed over time. However, most speed figure makers are still using the same charts they used a decade or longer ago. Therefore they may not be making speed figures that accurately equalize abilities across distances the way they are supposed to.
In addition, how can you tell how fast a track really was if all your figures are out of sync by distance?
You'd have to make a separate variant for every distance.
I realize that some people actually try to do that because tracks change speed and other factors impact time at different distances differently, but ultimately someone like that is NOT making speed figures. They are making "performance figures" that also build pace, race development, and other things that impact time right into the speed figure via the track variant even though a change in track speed may have had nothing to do with the time.
IMO that is a process fraught with peril because those other impacts on final time do not impact all horses equally - even inside the same race.
There is almost no way you can make an accurate track variant off a single race in some (maybe many) circumstances without a very deep understanding of the impact of pace, trip, etc... on time and exactly how each individual horse was impacted in that race.
I think what CJ is hinting at in a diplomatic way is that the relationship between sprints and routes may have changed over time. However, most speed figure makers are still using the same charts they used a decade or longer ago. Therefore they may not be making speed figures that accurately equalize abilities across distances the way they are supposed to.
That is what I am thinking exactly. I think most figure makers do a good job of adapting to this, but it is getting harder and harder as time goes on and putting those figures made using the existing charts on shaky ground.
jasperson
08-09-2011, 08:28 PM
I think the answer is our modern training methods. Routes are trained like sprinters and you can't get a horse ready to do his best at of a mile with works or 3f or 4f. When have you seen a mile work for a horse lately? Present day trainers are afraid to condition their horse to run routes for fear they will break down.
maddog42
08-09-2011, 10:09 PM
There are a lot of ways to do it, but what is the "total race" that we use? How do we find equal performances at different distances. That is the biggest dilemma I think.
"How do we find equal performances at different distances ?"
"How do we find equal performances at different distances ?"
I thought the question warranted repeating. The best questions usually are. This is 1/3 of the dilemma.
First of all lets assume there are "equal performances".
Then let us assume that this performance can be repeated at a different distance.
Then let us assume that we know what the odds are that he will repeat this performance today. I can't tell you how many times I have been RIGHT about a horse and see him not win, because of another Out of the blue performance by another horse. I digress.
Equal performances at different distances. You must have some sort of baseline, Par or World record to project it to.
I have been experimenting with different kinds of Pars. Not just PPF Pars.
Compounded Pace rating Pars. Anyhow....
Carroll talks about handicapping in the days before the Beyer Figs came along
and the outstanding Overlays he was getting. He would get $20 horses that should have been 3/1. His method wasn't better than Beyers ( I think it was clearly inferior) but it was different and good enough to pay dividends.
My point is that your method of Judging these "equal performances" doesn't have to be that much better, but it damn sure needs to be original. And of course it needs to reflect a subset of reality that pays high mutuels.
I looked at some old Beyer pars compared to newer ones, for a point of reference. Not sure if this is going to help a lot, but I'll throw it out ( and then you can throw it out if it doesn't help!
I tried to pick two classes that might not vary as much as others, ie, claiming races today are a different animal than back then. MSW and NW1 seemed to be the two that would be the most static over time, class-wise.
Aqueduct inner dirt, fast, older males, no state breds, MSW and NW1 allowance races, 2000 - 2004
MSW 6.0F par = 83.9
MSW Route par = 79.6
NW1 6.0 par = 87.9
NW1 route par = 88.9
Aqueduct inner dirt, fast, older males, no state breds, MSW and NW1 allowance races, 2006 - 2011
MSW 6.0F par = 80.0
MSW Route par = 84.3
NW1 6.0 par = 92.3 (92.7 - not using AOC races w/NW1 clause)
NW1 route par = 91.2 (93.0 - not using AOC races w/NW1 clause)
Can't explain the MSW numbers at all, but the NW1 make more sense.
"How do we find equal performances at different distances ?"
"How do we find equal performances at different distances ?"
Just to be clear, I'm not talking about 'equal' for one horse. I am talking what is equal for the same class of horse, be it G1 stakes for older males or 10k claimers.
I looked at some old Beyer pars compared to newer ones, for a point of reference. Not sure if this is going to help a lot, but I'll throw it out ( and then you can throw it out if it doesn't help!
I tried to pick two classes that might not vary as much as others, ie, claiming races today are a different animal than back then. MSW and NW1 seemed to be the two that would be the most static over time, class-wise.
Aqueduct inner dirt, fast, older males, no state breds, MSW and NW1 allowance races, 2000 - 2004
MSW 6.0F par = 83.9
MSW Route par = 79.6
NW1 6.0 par = 87.9
NW1 route par = 88.9
Aqueduct inner dirt, fast, older males, no state breds, MSW and NW1 allowance races, 2006 - 2011
MSW 6.0F par = 80.0
MSW Route par = 84.3
NW1 6.0 par = 92.3 (92.7 - not using AOC races w/NW1 clause)
NW1 route par = 91.2 (93.0 - not using AOC races w/NW1 clause)
Can't explain the MSW numbers at all, but the NW1 make more sense.
I like the idea, but the inner dirt course had a pretty big shift in purses over the last few years. I'm not sure that is the best example.
I didn't want to use Belmont because of the one turn routes, and I wanted to use as much older data I could find.
If it looked promising, I can find other tracks to use, but I have a long term DB for NYRA tracks.
I have lots for Finger Lakes, that might be better to use. Any suggestions on a track - I might be able to get enough races out of older dbs to put one together.
maddog42
08-10-2011, 12:55 AM
Just to be clear, I'm not talking about 'equal' for one horse. I am talking what is equal for the same class of horse, be it G1 stakes for older males or 10k claimers.
It seems to me you are looking for the Holy Grail of Pars. A very worthy undertaking. And like the Holy Grail it may not exist. You are looking for more accuracy and consistency and that almost certainly does exist.
Back to your original question:
Considering deceleration, don't routes by definition have to run disproportionately slower than sprints ? Percentage wise it might be equal, but not timewise. The beaten length chart wouldn't have much say in that. Am i misunderstanding something here?
Native Texan III
08-10-2011, 06:52 AM
Ever notice that using the Beyer charts, the track is ALWAYS faster for sprints than for routes. What I mean is his variant always shows the track faster for sprints, and by always I mean about 98% of the time, if not more.
Doesn't this mean there is something wrong with the charts in the first place? The chart was supposed to show equal performances at different distances.
I would agree with that and to my mind it has been going on a long time because the basic USA method is badly flawed in concept. The method was acceptable in the days before widespread computing and you had to have such rough methods to do things by hand. Doing it by hand you could easily spot bad answers and double check back. Now it may be the worst of all worlds, they seem to use the same oversimplified, flawed methods but computerise them without there being a human thinking hang on a minute that looks wrong or no the race was not run that way. Not knowing there are serious errors means the errors get compounded as the current "truth" used in formulating future blind "truth".
The UK methods are far more firmly fixed into reality and they base things on the standard horse concept and a universal rating system for the horse population. What optimum time would this standard super horse take to run that distance in those conditions. Here we remain in the dark ages and have fudge upon fudge, with no real standardisation. Of course, even those UK methods are being left behind by those preparing their own figures based on modern science and race statistics, and benefiting greatly from the misinformation others are unwittingly using.
Why is the method flawed?
Why is the method flawed?
The lengths-to-points conversion, for example, is deeply flawed.
There should not be two formula's for it (i.e. sprint or route), there should be one for each course/distance/going.
If you use the winner of the race, lengths are not part of it.
TrifectaMike
08-10-2011, 09:57 AM
If you use the winner of the race, lengths are not part of it.
Tom,
The entire process is flawed.
Although, I don't use speed ratings in the normal sense of handicapping, I believe I have a method which is a vast improvement over the current approach and no more complicated.
Tom, you appear to have database capability. I can PM you the method. If you choose to, you can post the charts, but not the method.
If you agree, let me know.
Mike (Dr Beav)
Well, I will certainly give it a try, if I have the right data.
Of course all will be confidential.
I may have questions, but I will do my best and be objective.
classhandicapper
08-10-2011, 10:36 AM
Not knowing there are serious errors means the errors get compounded as the current "truth" used in formulating future blind "truth".
That's a serious potential problem, but I'm not sure it's a cause of any potential issues with the relationship between route and sprint times.
Alright, I tried Mike's ideas and here is Saratoga dirt.
I think 6.5 furlongs might be out of whack due to small number of races, but here is my first pass at it.
TrifectaMike
08-10-2011, 05:00 PM
Alright, I tried Mike's ideas and here is Saratoga dirt.
I think 6.5 furlongs might be out of whack due to small number of races, but here is my first pass at it.
Tom,
It looks reasonable. You could have extended the lower portion of the chart. Can you send me the stats.
Mike (Dr Beav)
I did extend it, but had to crop the jpeg to get the size down for posing.
I will send you the whole file in the morning - left my thumb drive at work!
delayjf
08-11-2011, 12:13 AM
However, most speed figure makers are still using the same charts they used a decade or longer ago
Could the answer be as simple as contructing your own charts??
What if you just took the variants for the past year, say, and find out how much off the routes are to the sprints and then just adjust the route chart by that much?
What if you just took the variants for the past year, say, and find out how much off the routes are to the sprints and then just adjust the route chart by that much?
I've got something in the works that seems to be working nicely. When the time comes you'll know here, not playing the silly PM game. Everybody that has contributed to these threads is appreciated.
DISTANCE par time
5 1/2 FURLONGS 64.57
6 FURLONGS 71.24
6 1/2 FURLONGS 77.50
7 FURLONGS 84.37
1 1/8 MILES 112.78
1 3/16 MILES 118.99
Here are my pars for Saratoga dirt. I use the usual points per length but a make some distance dependent adjustments that may account for any deficiencies in the points scale.
DISTANCE par time
5 1/2 FURLONGS 64.57 72
6 FURLONGS 71.24 69
6 1/2 FURLONGS 77.50 73
7 FURLONGS 84.37 72
1 1/8 MILES 112.78 71
1 3/16 MILES 118.99 69
Here are my pars for Saratoga dirt. I use the usual points per length but a make some distance dependent adjustments that may account for any deficiencies in the points scale.
Just out of curiosity, I looked these up on the chart I'm working on to see how they line up. It looks like they aren't far from my baseline chart.
If you want any other tracks posted for comparison let me know.
If you want any other tracks posted for comparison let me know.
Thanks, maybe in a few days, still in the very rough stages.
TrifectaMike
08-11-2011, 04:31 PM
I've got something in the works that seems to be working nicely. When the time comes you'll know here, not playing the silly PM game. Everybody that has contributed to these threads is appreciated.
CJ, using a single point as a baseline is absurd. The method I described to Tom (which is an approximation) uses an entire distribution at a distance as a baseline. All other distances are calculated from determining equal probabilities for each distance using one distance as a baseline. Essentially compare probability densities at each distance to a baseline distance by sliding the densities over each other until you find comparable probabilities.
Mike (Dr Beav)
CJ, using a single point as a baseline is absurd. The method I described to Tom (which is an approximation) uses an entire distribution at a distance as a baseline. All other distances are calculated from determining equal probabilities for each distance using one distance as a baseline. Essentially compare probability densities at each distance to a baseline distance by sliding the densities over each other until you find comparable probabilities.
Mike (Dr Beav)
That's very funny, I do something similar. Not the same exactly, but in the same spirit.
I'm slightly baffled by this thread to be honest. Surely, the serious speed figure makers can be assumed to update their base charts at least once a year.
PhantomOnTour
08-11-2011, 05:05 PM
That's very funny, I do something similar. Not the same exactly, but in the same spirit.
I'm slightly baffled by this thread to be honest. Surely, the serious speed figure makers can be assumed to update their base charts at least once a year.
Every winter.
It's a lot of work but necessary imo.
Aside from my starting number of a 100 fig for all distances, every class level has a different par fig for sprints and routes...ie:
BELMONT
6f: An1x 3+ is a 103-103-104
Mdn20 3+ is a 94-94-93
9f: An1x 3+ is a 104-104-105
Mdn20 3+ is a 93-93-92
PhantomOnTour
08-11-2011, 05:36 PM
Beyer has the most popular figs out there today, but he doesn't deal with pace and that's where he lost me...i chose William Quirin and his way of making figs.
The first speed fig book you will hear anyone mention is Beyer On Speed or Picking Winners but the book that spelled it all out for me was Quirin's Thoroughbred Handicapping: State Of The Art, circa 1985 or so.
He covers all aspects of betting the ponies (class, pedigree, trainer intent, etc...)but spends a few chapters on making figs and applying them.
His Clm50 3+ pars from SA:
6f:106-106
7f:106-107
8f:108-109-109
9f:108-109-110
note:i make 2 pace figs for sprints(2f and 4f)but he only made a 4f pace fig and a final time speed fig. His route figs were for 4f-6f-final time, as are mine.
Also note how Clm50 has different par figs for different distances.
That, and the way I calculate beaten lengths (ie-convert them to actual time) is the whole ball of wax for me.
Things aren't perfect...my GP figs all seemed a bit too high when they began to go back to NY and face the locals who had been laid up for the winter or racing over the Aqu inner....but you go back and tweak...next season will be better. Truth was that I bought my GP pars instead of researching them myself; and the fact that many good horses leave the grounds after the Fla Derby and the rest of the meet (used to be at Hialeah) is pretty low level stuff.
The better MSW and entry level Alw horses are gone...i am seriously considering a separate chart for post FlaDby GP races. I will have a closer look this winter and compare the times from before and after the FlaDby.
dnlgfnk
08-11-2011, 07:16 PM
What if you just took the variants for the past year, say, and find out how much off the routes are to the sprints and then just adjust the route chart by that much?...
...and gm10..."I'm slightly baffled by this thread to be honest. Surely, the serious speed figure makers can be assumed to update their base charts at least once a year"...
Which is why I suggested in post #9...
"CJ, have the charts been ajar for months or over a year? It would seem the database guys would make corrections after a time. I know that they periodically review projected figures in order to avoid the inevitable "shrinking figure" problem that always plagued me, from not giving winners enough credit for a bigger number when projecting"...
delayjf
08-11-2011, 09:58 PM
It would seem the database guys would make corrections after a time.
I know the guys at HDW use a big database to calculate their figures - Ron Tiller where are you.
I recalculate all my variants every 2 or 3 weeks and update the pars as part of the process.
I recalculate all my variants every 2 or 3 weeks and update the pars as part of the process.
That's quite a lot! Any specific reason? You wouldn't think that pars can change that fast?
(I update mine twice a year, after the Derby, and after the Breeders Cup.)
I don't think the pars change in any significant way. What I am after is revising the variants based on projecting back from horses that run in subsequent races.
In order to have a stable process I need to update the class pars and time pars along with the variants.
It takes the computer an hour or two much not much of my time involved.
I'm slightly baffled by this thread to be honest. Surely, the serious speed figure makers can be assumed to update their base charts at least once a year.
I do, but I don't think Beyer does. Actually, I don't think it has been updated in years.
CJ, using a single point as a baseline is absurd. The method I described to Tom (which is an approximation) uses an entire distribution at a distance as a baseline. All other distances are calculated from determining equal probabilities for each distance using one distance as a baseline. Essentially compare probability densities at each distance to a baseline distance by sliding the densities over each other until you find comparable probabilities.
Mike (Dr Beav)
By all means, give us an example.
As far as being absurd, I'm willing to listen. I think determining the absolute fastest a horse can run each distance is a great way to start, if it can be done. You have to have some baseline to be able to compare horses across many different racetracks. There are just so many things to consider if you use all classes of horses. For example, cheap horses don't run certain distances very often. Very good horses also don't run certain distances very often. You have to be able to find a way to tie those distances together. You have the same dilemma with younger horses and older horses.
I could probably make a fabulous chart for any individual race track. It would work incredibly well making numbers for that one track. However, if I do that for every track, it will take many years to accurately determine how the tracks compare to each other, and by the time I'm finished things will have changed drastically at many tracks. The game is not static.
Here are the steps I think you should use to make speed charts:
Establish the times at each distance that are similar
Establish the value of time at each distance
Adjust for aberrations at individual tracks due to layout, runup, etc.
I think many times people are too quick to adjust for individual tracks. There should be a reason that can be pinpointed before making a change. For example, at Pimlico, all 6f races are slow in relation to the other distances because there is virtually no run up. I would certainly adjust my chart for that. However, doing it without a reason just to make the data fit is going to give a lot of bad ratings in my opinion.
By all means, give us an example.
As far as being absurd, I'm willing to listen. I think determining the absolute fastest a horse can run each distance is a great way to start, if it can be done. You have to have some baseline to be able to compare horses across many different racetracks. There are just so many things to consider if you use all classes of horses. For example, cheap horses don't run certain distances very often. Very good horses also don't run certain distances very often. You have to be able to find a way to tie those distances together. You have the same dilemma with younger horses and older horses.
I could probably make a fabulous chart for any individual race track. It would work incredibly well making numbers for that one track. However, if I do that for every track, it will take many years to accurately determine how the tracks compare to each other, and by the time I'm finished things will have changed drastically at many tracks. The game is not static.
It isn't that hard to achieve with Mike's approach.
Suppose you have the distribution of running times over each C&D, then you can quite easily verify how tracks relate to each other. For example, you may find that if an average horse can beat 70% of horses at Calder, it will only beat 55% of horses at Gulfstream (based on the distributions of running times at both tracks).
Since these %'s represent running times, they are quite easily converted into speed figure points.
The hardest part is automating the process as you don't want to be doing this manually every time you think it's necessary.
It isn't that hard to achieve with Mike's approach.
Suppose you have the distribution of running times over each C&D, then you can quite easily verify how tracks relate to each other. For example, you may find that if an average horse can beat 70% of horses at Calder, it will only beat 55% of horses at Gulfstream (based on the distributions of running times at both tracks).
Since these %'s represent running times, they are quite easily converted into speed figure points.
The hardest part is automating the process as you don't want to be doing this manually every time you think it's necessary.
I don't do anything manually.
There is always more to it than meets the eye. Do you use all Gulfstream horses, or do you use the ones before the "elite" part of the meet ends? What do you do with the slow horses that stick around the last month? As I said elsewhere, the game isn't static. It is constantly changing. About the only thing that stays the same is the best horses at the time win the best races.
TrifectaMike
08-12-2011, 12:02 PM
By all means, give us an example.
As far as being absurd, I'm willing to listen. I think determining the absolute fastest a horse can run each distance is a great way to start, if it can be done.
Creating a population of horses at a distance from all (major, and one minor) tracks is much better. Then split the population into three equal sub populations (Best, Middle, Worst). Derive statistics for each of the three populations.
Use the statistics from one distance and another distance to fill the chart. Continue the process from one distance to another.
The process (which is actually a good approximation ) is very similar to what Tom did for a single track.
Tom, can you send the PM I sent you and share the your data files with CJ?
After your review, I'll attempt to answer your questions.
Mike (Dr Beav)
KingChas
08-12-2011, 12:08 PM
I could probably make a fabulous chart for any individual race track. It would work incredibly well making numbers for that one track. However, if I do that for every track, it will take many years to accurately determine how the tracks compare to each other, and by the time I'm finished things will have changed drastically at many tracks. The game is not static.
I make the individual chart for my main track(NY).
Those baseline values are where I start.
Then I have a chart for every track in the US with every distance.
which I (+ or -) when shipping in.
I construct the average baseline 1st for every track at every distance(including odd distances,inner turf,even 4.5F.) using only the latest meet, I make the class adjustments later.
Should make this job quicker.
My baseline values are updated after the spring Belmont meet each year.
Sprints tend not to vary much at most tracks,but routes,odd distances and turf races do meet to meet.
I don't do anything manually.
There is always more to it than meets the eye. Do you use all Gulfstream horses, or do you use the ones before the "elite" part of the meet ends? What do you do with the slow horses that stick around the last month? As I said elsewhere, the game isn't static. It is constantly changing. About the only thing that stays the same is the best horses at the time win the best races.
I would consider the overall running times of Gulfstream. I don't go into such detail to be honest. You can spend years on such possible nuances, but most of them will just come out in the wash.
Anyhow, just pointing out that it isn't that hard to get a reasonably accurate track-to-track adjustment.
I would consider the overall running times of Gulfstream. I don't go into such detail to be honest. You can spend years on such possible nuances, but most of them will just come out in the wash.
Anyhow, just pointing out that it isn't that hard to get a reasonably accurate track-to-track adjustment.
Reasonably accurate is I guess where we will differ.
Tom, can you send the PM I sent you and share the your data files with CJ?
After your review, I'll attempt to answer your questions.
Mike (Dr Beav)
OK, done.
Reasonably accurate is I guess where we will differ.
Well no disrespect but that is very easy to say. You should present your own practical approach wrt track-to-track adjustments, so that we can judge for ourselves.
Well no disrespect but that is very easy to say. You should present your own practical approach wrt track-to-track adjustments, so that we can judge for ourselves.
That is what I'm working on and will post when the time comes. If you think it isn't that hard, I would say you aren't doing enough work, but that is just my opinion. There are tons of tracks to deal with in North America, all at different class levels and with different surfaces and configurations.
That is what I'm working on and will post when the time comes. If you think it isn't that hard, I would say you aren't doing enough work, but that is just my opinion. There are tons of tracks to deal with in North America, all at different class levels and with different surfaces and configurations.
If you've got nothing to show for it, you shouldn't claim better accuracy.
I can't help but wonder what adjustments you are making for while you are developing this super method.
If you've got nothing to show for it, you shouldn't claim better accuracy.
I can't help but wonder what adjustments you are making for while you are developing this super method.
It never takes long for you true colors to appear. Where did I claim I had better accuracy? I said I'm trying to make better charts, nothing more, and I know is isn't "not too hard", at least with the precision I'm looking to get.
If you've got nothing to show for it, you shouldn't claim better accuracy. Myself, and scores of other Pace Figure users have MUCH to show for his efforts to date.
I can't help but wonder what adjustments you are making for while you are developing this super method. Me too. If they are half as good as some of the other stuff he has given us over the years, I will dance in the streets!
Much to show......
It never takes long for you true colors to appear. Where did I claim I had better accuracy? I said I'm trying to make better charts, nothing more, and I know is isn't "not too hard", at least with the precision I'm looking to get.
"Reasonably accurate is I guess where we will differ."
Listen, you ducking every question by sending another vague accusation my way.
You've criticized and questioned a lot of people here (including Beyer), but you're not telling us how you will actually deal with this problem or how you are dealing with it right now. You haven't even given us any evidence that the problem actually exists!!
"Reasonably accurate is I guess where we will differ."
Listen, you ducking every question by sending another vague accusation my way.
You've criticized and questioned a lot of people here (including Beyer), but you're not telling us how you will actually deal with this problem or how you are dealing with it right now. You haven't even given us any evidence that the problem actually exists!!
I didnt' realize I was a professor giving a lecture. I thought it was a discussion board. I've never claimed anything, though you ducked that question. I'm trying to stimulate a discussion which was going well. You seem to have an uncanny ability to be a thread killer though.
JeremyJet
08-12-2011, 02:58 PM
You haven't even given us any evidence that the problem actually exists!!
Bingo!
This all came about because CJ thought he noticed something wrong with the route variants. The problem is that what he was looking at wasn't just the variant.
Bingo!
This all came about because CJ thought he noticed something wrong with the route variants. The problem is that what he was looking at wasn't just the variant.
That isn't the only thing I've noticed. I have seen problems for a few years, and the variants were more proof of the problem than the problem themselves.
JeremyJet
08-12-2011, 03:18 PM
That isn't the only thing I've noticed. I have seen problems for a few years, and the variants were more proof of the problem than the problem themselves.
Why are you so concerned with what Beyer is doing? Unless, of course, you're useing his information to complete yours.
Why are you so concerned with what Beyer is doing? Unless, of course, you're useing his information to complete yours.
I have posted here before that I track what he does with his speed charts and even posted the adjustments I've found here on this site. It isn't a secret. However, I'm getting away from that. I think I can do it better. Time will tell if I'm right.
JeremyJet
08-12-2011, 03:25 PM
Why are you so concerned with what Beyer is doing? Unless, of course, you're useing his information to complete yours.
And why, all of a sudden, is there a concern about the accuracy of your numbers? Tom's on record saying they're a "gold mine." :rolleyes:
Because he never rests on his laurels - he is known for continual improvements to his products.
And why, all of a sudden, is there a concern about the accuracy of your numbers? Tom's on record saying they're a "gold mine." :rolleyes:
I see a trend I don't like and I'd rather catch it before it goes too far. It isn't a "sudden" concern, I'm always concerned.
JeremyJet
08-12-2011, 03:29 PM
And why, all of a sudden, is there a concern about the accuracy of your numbers? Tom's on record saying they're a "gold mine." :rolleyes:
Don't get me wrong, i'm not suggesting your product isn't any good, I just think Tom is over-the-top with his enthusiasm.
PhantomOnTour
08-12-2011, 03:40 PM
I see a trend I don't like and I'd rather catch it before it goes too far. It isn't a "sudden" concern, I'm always concerned.
Hello!
I heard that
Because he never rests on his laurels - he is known for continual improvements to his products.
Can someone finally come up with any evidence of this alleged problem?? Is this a real problem or a marketing ploy?
(Just kidding CJ :p )
Can someone finally come up with any evidence of this alleged problem?? Is this a real problem or a marketing ploy?
(Just kidding CJ :p )
Well obviously I think there is a problem. I do very little marketing, and this certainly isn't any of it. If there wasn't a problem, someone that understands Beyers could have answered the original question with a satisfactory answer. That hasn't happened yet.
JeremyJet
08-12-2011, 03:55 PM
If there wasn't a problem, someone that understands Beyers could have answered the original question with a satisfactory answer. That hasn't happened yet.
Speaking for myself ...
Maybe the answer isn't forthcoming for the same reasons you cite for not handicapping on this board.
Speaking for myself ...
Maybe the answer isn't forthcoming for the same reasons you cite for not handicapping on this board.
I've done plenty of handicapping on this board, just not on demand. Maybe that is the reason, but if so, there is no need to comment at all really.
JeremyJet
08-12-2011, 04:04 PM
Is this a real problem or a marketing ploy?
Well, if it is a marketing ploy, one thing is clear ... CJ uses a different marketing agency than Dave Schwartz!!! ;)
I've done plenty of handicapping on this board, just not on demand. Maybe that is the reason, but if so, there is no need to comment at all really.
Reading this again, it doesn't sound good. What I'm trying to say is if you think you know something others don't, but you don't want to share, why post at all? It just looks like you think are smarter than everyone else.
JeremyJet
08-12-2011, 04:29 PM
Reading this again, it doesn't sound good. What I'm trying to say is if you think you know something others don't, but you don't want to share, why post at all? It just looks like you think are smarter than everyone else.
I tried to give you a clue and it went right over your head. gm10 is right, there is no problem.
I tried to give you a clue and it went right over your head. gm10 is right, there is no problem.
If that were true, there wouldn't be constant adjustments, and I'm not just talking routes. Also, if there is no problem, why do we never see fast horses any more? I can't buy that our horses just suddenly stink.
JeremyJet
08-12-2011, 05:05 PM
If that were true, there wouldn't be constant adjustments, and I'm not just talking routes. Also, if there is no problem, why do we never see fast horses any more? I can't buy that our horses just suddenly stink.
I don't know what adjustments you're referring to, but don't you think their (Beyer) techniques have become more sophisticated over the past 30/40 years? Not the methodology as a whole, but rather the technique in solving the variant puzzle. No doubt influenced by other, more advanced, figure makers (Jerry Brown & Len Ragozin).
What do you consider fast?
I don't know what adjustments you're referring to, but don't you think their (Beyer) techniques have become more sophisticated over the past 30/40 years? Not the methodology as a whole, but rather the technique in solving the variant puzzle. No doubt influenced by other, more advanced, figure makers (Jerry Brown & Len Ragozin).
What do you consider fast?
I can tell you what isn't fast....105s routinely winning G1 races.
As for the adjustments, if you don't know then maybe you don't know as much as you think. Beyer sees the same things I'm seeing, but I think he is trying to hold it together with band aids rather than fixing it.
I think they have become more sophisticated of course, but not nearly to the level they could be.
Just to be clear, I think the same things are probably troubling most figure makers. It is easier to talk about Beyer because his stuff is public.
maddog42
08-12-2011, 10:56 PM
"Reasonably accurate is I guess where we will differ."
Listen, you ducking every question by sending another vague accusation my way.
You've criticized and questioned a lot of people here (including Beyer), but you're not telling us how you will actually deal with this problem or how you are dealing with it right now. You haven't even given us any evidence that the problem actually exists!!
Damn Cj. You stir up more shit over here in the handicapping section than I do over in off-topic when I call Rush a junkie. How do you get all these enemies?
Why don't you tell us EVERYTHING you know even when we act like a bunch of assholes ? Just because you spent years developing this stuff, we don't care.
We don't want to do any work. We want it served on a silver platter so we can rip it to shreds. What Gives?
cj's dad
08-12-2011, 11:27 PM
I know I am biased but some of you are acting foolishly. The sons #s work.
If you incorporate them with the info you are using they work even better.
Damn Cj. You stir up more shit over here in the handicapping section than I do over in off-topic when I call Rush a junkie. How do you get all these enemies?
Why don't you tell us EVERYTHING you know even when we act like a bunch of assholes ? Just because you spent years developing this stuff, we don't care.
We don't want to do any work. We want it served on a silver platter so we can rip it to shreds. What Gives?
On the contrary, I have been working on this for 10 years so I am very interested in such issues. It's just that nobody has so far managed to turn this suspicion into something that looks like it's a real problem.
(Not that there haven't been any some nice hypotheses put forward).
Personally, the more I think about it, the less convinced I am. There's very little chance that the serious figure makers, guys like Beyer who have been doing this for decades, aren't aware of the need to keep updating their fundamental reference data.
For my own program I decided years ago that I cannot try to improve individual pieces (such as the accuracy of the speed figures) and expect to have a beneficial effect on returns.
If you are using adjusted speed figures it may well be that in the process of making the adjustment number you are compensating for the shortcomings in the speed figure.
Like a ramshackle structure where the walls are not straight but the parts hold each other up; I don't try to straighten one piece because that would upset the balance.
maddog42
08-13-2011, 09:33 AM
On the contrary, I have been working on this for 10 years so I am very interested in such issues. It's just that nobody has so far managed to turn this suspicion into something that looks like it's a real problem.
(Not that there haven't been any some nice hypotheses put forward).
Personally, the more I think about it, the less convinced I am. There's very little chance that the serious figure makers, guys like Beyer who have been doing this for decades, aren't aware of the need to keep updating their fundamental reference data.
"serious figure makers". I am pretty sure CJ is very serious. From the posts on this forum, He has been doing it for more than a decade. I have never used his figures but a few of my friends have. I don't think you get as many customers as he has, without being at least competent. If he says there is a problem with the way conventional Beyer Figure makers compute them, then there probably is. I think he wants to find a better way. Why don't we get back to the topic of why there is this discrepancy between route and sprint figures ?
I apologize for indirectly calling anyone names.
Personally, the more I think about it, the less convinced I am. There's very little chance that the serious figure makers, guys like Beyer who have been doing this for decades, aren't aware of the need to keep updating their fundamental reference data.
I've already told you he doesn't update it. Anyone with a DRF and his speed charts can figure this out, but you would rather just think I'm lying. Why would I make this up?
TrifectaMike
08-13-2011, 10:56 AM
CJ, assuming Tom has sent you the latest PM. This thread no longer has any purpose for continuation.
One simple equation and proper conditioning of the data addresses both problems: statistically sound track specific speed charts and track to track adjustments. In addition periodic updating is rather simple to implement.
I suggest you take a serious look at it.
Mike (Dr Beav)
Saratoga_Mike
08-13-2011, 11:06 AM
CJ, assuming Tom has sent you the latest PM. This thread no longer has any purpose for continuation.
Mike (Dr Beav)
Yes CJ, stop wasting our time with a thread that is actually useful and informative.
TrifectaMike
08-13-2011, 11:11 AM
Yes CJ, stop wasting our time with a thread that is actually useful and informative.
Useful and informative? Have you read the majority of the posts? Point out the useful and informative portions.
Mike (Dr Beav)
P.S. Gm10's posts are spot on.
Relwob Owner
08-13-2011, 11:19 AM
Useful and informative? Have you read the majority of the posts? Point out the useful and informative portions.
Mike (Dr Beav)
P.S. Gm10's posts are spot on.
DR,
Point out the potions that arent useful and informative.
TrifectaMike
08-13-2011, 11:30 AM
DR,
Point out the potions that arent useful and informative.
The last 30 or so post are nothing more than guys engaging in a circle jerk.
Mike (Dr Beav)
Relwob Owner
08-13-2011, 12:37 PM
The last 30 or so post are nothing more than guys engaging in a circle jerk.
Mike (Dr Beav)
Dr,
Not very specific.
If you want to criticize a thread and actually be correct, I would direct you to the "Handicapping angle-The Pretend Act Up" thread and do a "diagnosis" of that one.
Useful and informative? Have you read the majority of the posts? Point out the useful and informative portions.
Mike (Dr Beav)
P.S. Gm10's posts are spot on.
Oh, like the one where he says I'm lying about the Beyer speed chart not being updated? That is spot on?
What I have found is many of the people that commented don't really understand the Beyer methodology. If they did, the comments would not have been made.
CJ, assuming Tom has sent you the latest PM. This thread no longer has any purpose for continuation.
One simple equation and proper conditioning of the data addresses both problems: statistically sound track specific speed charts and track to track adjustments. In addition periodic updating is rather simple to implement.
I suggest you take a serious look at it.
Mike (Dr Beav)
I'll certainly read them after the weekend. If you think it is simple though, I suspect you know more about math than racing. We'll see, I'll go in with an open mind.
classhandicapper
08-13-2011, 12:56 PM
If I were doing a project like this I would look at time/distance relationships across the entire class spectrum at each track. I think you have to look at both extremes because it could easily be that slow cheap horses slow down faster and worse as distances stretch out than expected by Beyer type formulas that try to value a 1/5th of a second or a length (and vice versa).
Once you get the distance relationships down at each track, then you can start with various assumptions about the market for horses being fairly efficient and try to equalize the tracks by common classes.
That will wind up being inaccurate because there are strong and weak groups in various parts of the country and because purses and costs are also not necessarily in sync. However, it should at least be in the ballpark.
Once you have ballpark figures nationally, you have to monitor every single shipper on an ongoing basis so you can constantly tweak the figures until the entire national database is in sync and can be kept in sync.
You would more or less have to keep doing this on an ongoing basis to keep all relationships up to date.
All that said, IMO people need to realize that an entire book could be written on all the complexities and problems with making accurate figures. A second book could be written on why even when they are accurate they don't reflect the abilities and performances of the horses very well.
Final time is a function of pace, race development, the timing of moves, day to day changes in the surface other than track speed that impact the distance relationships and running style advantages, horses getting favorable racing position etc...
So IMO you don't want to get too bogged down in very small inaccuracies. IMO figures are best used to identify contenders, not separate them.
I happen to agree with CJ that there may be more than a minor problem with some of the current sprint/route relationships for a variety of reasons. So I think what he is doing is well worthwhile and I trust him to correct any problems (I am a long time customer).
Saratoga_Mike
08-13-2011, 01:00 PM
Useful and informative? Have you read the majority of the posts? Point out the useful and informative portions.
Mike (Dr Beav)
P.S. Gm10's posts are spot on.
Have you read your own threads? Wow.
Saratoga_Mike
08-13-2011, 01:04 PM
Classhandicappper,
I'm a big fan of speed and pace figures, but I always enjoy your posts, including your most recent one in this thread. They're always well thought out and there's no agenda.
illinoisbred
08-13-2011, 01:05 PM
Classhandicappper,
I'm a big fan of speed and pace figures, but I always enjoy your posts, including your most recent one in this thread. They're always well thought out and there's no agenda.
Here too!!
classhandicapper
08-13-2011, 01:31 PM
Thanks guys. I'm just a huge fan of the sport and handicapping for a long time. I like to add my 2 cents once in awhile. ;)
If I were doing a project like this I would look at time/distance relationships across the entire class spectrum at each track. I think you have to look at both extremes because it could easily be that slow cheap horses slow down faster and worse as distances stretch out than expected by Beyer type formulas that try to value a 1/5th of a second or a length (and vice versa).
This is something I'm considering also. For example, if you set a baseline of a certain number of points per fifth second like Beyer does, it can easily be made variable instead of static.
For example, Beyer used (I think, but it is close) 1:11 4/5 as his baseline for 6f. He breaks that down to 359 fifths, and makes each fifth worth 1/359 X 1000 for a value of 2.7855 points per fifth of a second.
How would this change if you varied it by race time? Well, lets look at two extremes, 1:05 4/5 and 1:17 4/5, 6 seconds in either direction.
1:05 4/5 = 3.0395 points per fifth.
1:17 4/5 = 2.5707 points per fifth.
Obviously, those are two very extreme times and probably the limits you would ever see. Is this difference worth utilizing? Everyone can judge for themselves, but here is the effect. If a horse was beaten by one second at each time, these would be the adjustments from the winner's speed figure:
1:05 4/5: -15 points
1:11 4/5: -14 points
1:17 4/5: -13 points
At 9f, it would look like this. I used 8 second gaps because routes are more spread out obviously:
1:44 4/5: 1.9084 per fifth, -10 points at 1 second behind
1:52 4/5: 1.7730 per fifth, - 9 points at 1 second behind
2:00 4/5: 1.6556 per fifth, - 8 points at 1 second behind
Given the lack of precision in reported lengths behind, I'm not sure it matters at all, but it wouldn't hurt to use a variable beaten lengths chart as opposed to a static one. Would it make a difference in results...highly unlikely.
Oh, like the one where he says I'm lying about the Beyer speed chart not being updated? That is spot on?
What I have found is many of the people that commented don't really understand the Beyer methodology. If they did, the comments would not have been made.
Nobody has called you a liar - I merely suspect that you are wrong. I'm not a big fan of BSF because they are so one-dimensional, but I do not buy into your accusation that he is guilty of such fundamental errors.
Regarding the relationship course-surface-distance-going-DTV. It is indeed a very complex one. There was an excellent chapter on it in "Bioenergetics and Racehorse Ratings", but imo the best practical way to deal with the complexity of the problem is by estimating the distribution of running times of each course-surface-distance-going combination. You want to look at the standard deviation of each distribution, and you want to look at the family of the distribution. A linear time/lengths-to-points suggests that it is a Uniform distribution which it is patently not.
Regarding the track-to-track adjustments, what you need is set of raw, unadjusted figures and then a map of how shippers do figure-wise for each duo of tracks (this is not realistic however, so I suggest a pyramid of tracks with the best tracks forming the top layer of the pyramid, and then comparing the layers). It is important that your raw figures are governed by a probability distribution and not some manually compiled baseline. Trying to determine the time effect of things like a run-up at each track is a waste of time (pun). That's why statistics was invented, to estimate things that are too complex to explain otherwise. Once you've got the process automated, it becomes kind of a fun instrument to use. You can see what kind of impact slot money (or lack of it) is having, for example.
Nobody has called you a liar - I merely suspect that you are wrong. I'm not a big fan of BSF because they are so one-dimensional, but I do not buy into your accusation that he is guilty of such fundamental errors.
I've already stated I know for a fact he is using the same speed charts, and I've told you how you can check it yourself since you obviously think I just make stuff up. Apparently, you would rather just assume I'm wrong than do the work (about 5 minutes worth) and figure it out for yourself.
As I've said many times there is no bigger Beyer fan than me. He is a true pioneer. I just think he is not adapting to the sport as it changes with his figures.
maddog42
08-13-2011, 03:26 PM
I've already stated I know for a fact he is using the same speed charts, and I've told you how you can check it yourself since you obviously think I just make stuff up. Apparently, you would rather just assume I'm wrong than do the work (about 5 minutes worth) and figure it out for yourself.
As I've said many times there is no bigger Beyer fan than me. He is a true pioneer. I just think he is not adapting to the sport as it changes with his figures.
For what its worth. Charles Carroll says that Beyer is "the Man", and lavishes praise on his methods and his writing calling Picking Winners one of the great
How-to books of all time.
classhandicapper
08-13-2011, 03:32 PM
This is something I'm considering also. For example, if you set a baseline of a certain number of points per fifth second like Beyer does, it can easily be made variable instead of static.
I don't see any downside to that approach. It's probably a good idea.
I'm getting at is something else. Over the years I seem to recall a lot of very cheap maidens and others that could run figures in the 40s and 50s in sprints, but when they stretched out the same horses would run in the 30s and 40s and finish in a predictable way. Yet none of the other cheap horses would step up and improve a lot at the distance to beat them. It was almost as if once horses are bad enough, most have trouble stretching out at all. There may be some selective memory at work on my part because I haven't researched it, but I think it's worth looking at the extremes at both ends of the class spectrum to see if things break down.
PhantomOnTour
08-13-2011, 03:38 PM
Given the lack of precision in reported lengths behind, I'm not sure it matters at all, but it wouldn't hurt to use a variable beaten lengths chart as opposed to a static one. Would it make a difference in results...highly unlikely.
That's exactly what I have. In other words, considering the following:
Horse A: beaten 5 lengths in a 6f race run in 22.4-46.6-112.2
Horse B: beaten 5 lengths in a 6f race run in 22.1-45.6-112.2
The above two horses will most likely not receive the same final figure, because they most likely did not run the same final time. One ran thru a 26.0 last quarter and the other ran through a 27.0 last quarter.
Throw in lengths gained/lost during that last quarter and it's highly unlikely that it took these two horses the same amount of time to cover those 5 lengths after the winner hit the wire.
I've already stated I know for a fact he is using the same speed charts, and I've told you how you can check it yourself since you obviously think I just make stuff up. Apparently, you would rather just assume I'm wrong than do the work (about 5 minutes worth) and figure it out for yourself.
OK - tell me how to test this in 5 minutes. ("this" = the theory a relationship that has changed in the last decade, and the base charts do not reflect that change.)
classhandicapper
08-13-2011, 03:54 PM
Regarding the relationship course-surface-distance-going-DTV. It is indeed a very complex one. There was an excellent chapter on it in "Bioenergetics and Racehorse Ratings", but imo the best practical way to deal with the complexity of the problem is by estimating the distribution of running times of each course-surface-distance-going combination. You want to look at the standard deviation of each distribution, and you want to look at the family of the distribution. A linear time/lengths-to-points suggests that it is a Uniform distribution which it is patently not.
I hope I understand you. If not, disregard my comments.
The variations you seem to be taking about (especially from surface to surface) are at least partially related to the adjustments jockeys make in pace in order to compensate for the impact of the surface on the horse. Paces tend to be slower on synthetic/turf and that tends to bring the horses tighter together.
The question then becomes do you want figures to reflect real times or "possible" abilities. I say possible because each of the horses is different and changes in pace will have different impacts.
I suspect this even extends to same surface on different days on a more subtle level.
On a day that the track is carrying speed well, the paces are going to be faster than when it is not and vice versa. (it takes several races for the jockeys to figure out what's going on). I think these subtle changes from day to day impact the relationships between distances also.
I'm a numbers oriented guy. But the more I think about these problems the more I steer clear of pure numbers based handicapping. I prefer a more subjective comparative class/trip method and to use the strengths of the numbers to fill in the holes in my own method.
classhandicapper
08-13-2011, 04:16 PM
("this" = the theory a relationship that has changed in the last decade, and the base charts do not reflect that change.)
If you looks at the speed figures earned by the leading 3YOs in recent classic races, on average they seem to be shrinking (especially in the Belmont). Someone posted an article on this recently.
If you look at Grade 3 routes around the country these days they often earn figures slower than the Classified Alw PAR and routine ALW sprints even though the horses are more seasoned, have gone through their conditions, have competed at higher levels etc....
It's not that it's impossible to earn a huge route figure, but on average they seem to be shrinking over time while the best sprints seem to be the same or have even gotten faster over time.
It could be random or even selective memory, but it needs to be researched because something "seems" to be changing.
That's exactly what I have. In other words, considering the following:
Horse A: beaten 5 lengths in a 6f race run in 22.4-46.6-112.2
Horse B: beaten 5 lengths in a 6f race run in 22.1-45.6-112.2
The above two horses will most likely not receive the same final figure, because they most likely did not run the same final time. One ran thru a 26.0 last quarter and the other ran through a 27.0 last quarter.
Throw in lengths gained/lost during that last quarter and it's highly unlikely that it took these two horses the same amount of time to cover those 5 lengths after the winner hit the wire.
The difference would be minimal, would it not? The finish beaten lengths are taken from the ACTUAL finish time of the horse and converted to lengths via a formula.
Cratos
08-13-2011, 05:43 PM
Ever notice that using the Beyer charts, the track is ALWAYS faster for sprints than for routes. What I mean is his variant always shows the track faster for sprints, and by always I mean about 98% of the time, if not more.
Doesn't this mean there is something wrong with the charts in the first place? The chart was supposed to show equal performances at different distances.
I don’t want to get into a p----- contest about Beyer Speed Figures (I don’t use them), but I will suggest plotting the curve; it is very simple because all of the variables are continuous. I don’t see anything that is mystifying or not understandable about the Beyer speed charts except that their underlining premise is based on simple statistics and not scientific facts.
Then by all means, plot it and post it.
Teach us.
I don’t want to get into a p----- contest about Beyer Speed Figures (I don’t use them), but I will suggest plotting the curve; it is very simple because all of the variables are continuous. I don’t see anything that is mystifying or not understandable about the Beyer speed charts except that their underlining premise is based on simple statistics and not scientific facts.
This has absolutely nothing to do with the premise of the thread, but thanks for stopping by.
Saratoga_Mike
08-13-2011, 06:03 PM
Then by all means, plot it and post it.
Teach us.
Could be as good as one of Dr Beav's teachable moments...ahh, we can only hope.
I hope I understand you. If not, disregard my comments.
The variations you seem to be taking about (especially from surface to surface) are at least partially related to the adjustments jockeys make in pace in order to compensate for the impact of the surface on the horse. Paces tend to be slower on synthetic/turf and that tends to bring the horses tighter together.
The question then becomes do you want figures to reflect real times or "possible" abilities. I say possible because each of the horses is different and changes in pace will have different impacts.
I suspect this even extends to same surface on different days on a more subtle level.
On a day that the track is carrying speed well, the paces are going to be faster than when it is not and vice versa. (it takes several races for the jockeys to figure out what's going on). I think these subtle changes from day to day impact the relationships between distances also.
I'm a numbers oriented guy. But the more I think about these problems the more I steer clear of pure numbers based handicapping. I prefer a more subjective comparative class/trip method and to use the strengths of the numbers to fill in the holes in my own method.
Classhandicapper, I'm not sure how to answer your post. I agree with what you are saying, and all of it contributes to the speed handicapping having its limits.
Sometimes things get so complex, that you need to revert to statistics. One of the main theorems of statistics, the Central Limit Theorem, states that when you add a lot of factors together, combined they actually start following a Normal distribution. This is a very neat result. The Normal distribution is well understood, and statistical inference from it is reliable and powerful.
Applied to the problem at hand, you want to get the distribution of running times right before you do anything else (this isn't easy - see PhantomOnTour's post for example). When you do, you'll have a sound base for drawing conclusions and the perfect vehicle for translating time into ability.
Anyway, I agree with you that there are limits. To give a common European problem: when the turf is heavy, jockey's will usually go slowly to conserve energy, resulting in a even slower DTV than was the case. When there is an exception on the card where a jockey goes a decent pace to win wire-to-wire, it will result in a hopelessly inflated speed figure.
I cannot find the book but it was my starting point years ago in thinking about making figures.
What I recall was to make a par time by taking an average of adjusted times (adjusted back to 80 using the pars for the track/class combination and the daily track variant) for all races at a distance over say the last 3 years.
I would expect the difference between distance pars to vary from track to track.
1 Mile pars - 6 furlong pars:
TRACK DIFF
ALB 28.18
AP 26.60
AQI 28.04
AQU 27.27
ARP 28.79
ASD 27.81
BEL 26.94
BEU 28.52
BRD 27.60
CBY 28.26
CD 26.43
CNL 26.09
CRC 29.23
DEL 28.29
DMR 27.58
ELP 27.24
EMD 27.31
EVD 28.56
FG 28.16
FMT 27.39
FNO 27.44
FON 27.47
FP 28.71
GG 28.43
GP 27.58
HOO 27.49
HOU 28.28
HPO 27.42
IND 27.69
LNN 26.69
LRL 27.64
LS 28.68
MED 27.09
MNR 28.52
MTH 28.06
NP 27.64
OP 28.23
PEN 28.62
PHA 28.41
PID 27.64
PM 28.48
PNL 28.45
PRM 28.15
RD 27.90
RET 28.27
RP 28.09
RUI 29.43
SA 27.67
SOL 27.61
SR 27.30
STK 27.72
SUF 28.53
SUN 28.21
TDN 28.00
TP 28.02
TUP 27.62
WRD 27.84
YAV 27.86
ZIA 28.31
HUSKER55
08-14-2011, 09:05 AM
just so we are clear on this, a circle jerk is when a guy takes 3 or 4 women to bed with him. Right?
maddog42
08-14-2011, 09:47 AM
The difference would be minimal, would it not? The finish beaten lengths are taken from the ACTUAL finish time of the horse and converted to lengths via a formula.
Assuming that a length is 8ft. horse A ran the race in 72.97.
Horse B ran than the race in 73.00. Carroll would figure the race time for both animals as 72.92. This might not seem like much, but the actual difference is about half a length between Carrolls time and horse B's time. The actual difference in time .03 is about one fifth of a length.
Assuming that a length is 8ft. horse A ran the race in 72.97.
Horse B ran than the race in 73.00. Carroll would figure the race time for both animals as 72.92. This might not seem like much, but the actual difference is about half a length between Carrolls time and horse B's time. The actual difference in time .03 is about one fifth of a length.
My point was if two different races reported the same final time of 1:12.00, and each had a horse reported as 5 lengths backs, those two horses ran the exact same final time. If the photo finish formula is a length = .17 seconds, then both horses finished in exactly 1:12.85. It doesn't matter which was moving faster at the finish.
PhantomOnTour
08-14-2011, 01:30 PM
My point was if two different races reported the same final time of 1:12.00, and each had a horse reported as 5 lengths backs, those two horses ran the exact same final time. If the photo finish formula is a length = .17 seconds, then both horses finished in exactly 1:12.85. It doesn't matter which was moving faster at the finish.
But that's the problem I have.
A length =0.17 seconds regardless of the time of the last fraction.
I don't wanna open another can of worms, but turf races have final fractions that vary wildly...i've seen a 1m race finish in 22.6 and 25.6...surely 5 BL is not equal to the same amount of time in both cases. Just can't be.
Again, I apologize by introducing turf as I'm sure Beyers BL method is different from dirt, but you get my analogy.
But that's the problem I have.
A length =0.17 seconds regardless of the time of the last fraction.
I don't wanna open another can of worms, but turf races have final fractions that vary wildly...i've seen a 1m race finish in 22.6 and 25.6...surely 5 BL is not equal to the same amount of time in both cases. Just can't be.
Again, I apologize by introducing turf as I'm sure Beyers BL method is different from dirt, but you get my analogy.
The point is it doesn't matter. They aren't measuring how long it takes the horse to run the last five lengths. Finish distances aren't reported as how far back the horse was when the winner hit the wire. They are a time converted to a fictional distance. All finish positions are the final time of the horse subtracted from the final time of the race and converted to lengths.
I agree the .17 value or whatever is used is probably inaccurate in many cases. However, it doesn't change the fact that the horses finished in the exact same time. The five lengths is just a representation of time, it is not actual distance.
PhantomOnTour
08-14-2011, 01:43 PM
So they roll the film forward and calculate time; then assign BL based on their time formula?
So they roll the film forward and calculate time; then assign BL based on their time formula?
Well, sort of. They don't roll the film backwards, each horse is recorded with its exact final time. Why those aren't reported I have no idea. They report them for quarterhorses.
They can't report the way you were thinking. Horses pass each other all the time after the winner hits the wire.
GameTheory
08-14-2011, 02:44 PM
Well, sort of. They don't roll the film backwards, each horse is recorded with its exact final time. Why those aren't reported I have no idea. They report them for quarterhorses.
They can't report the way you were thinking. Horses pass each other all the time after the winner hits the wire.And of course then they round off the lengths. I tried using the video and frame by frame calculating the exact finish time of the losers to come up with the value they use for conversion to lengths (realizing there will never be an exact value because of the rounding), but could never find a single value that was consistently right. It must be different from track to track or at least from distance to distance (I think I did read somewhere they used bigger values for longer races). I had concocted a method to determine times for each horse at each call taking into account the velocity they were going (more or less) at each call -- a method that shouldn't have made sense to use at the finish because those should be exact times converted to lengths with an exact number, but I ended up using it anyway because it still seemed more accurate than any constant number I could plug in (although I believe .18 was the best fixed value I could find in terms of minimizing error). That was 7-8 years ago though -- maybe the digital systems they've got now they use more consistently. You could probably figure it out from the tracks that have Trakus data...
steveb
08-14-2011, 06:45 PM
My point was if two different races reported the same final time of 1:12.00, and each had a horse reported as 5 lengths backs, those two horses ran the exact same final time. If the photo finish formula is a length = .17 seconds, then both horses finished in exactly 1:12.85. It doesn't matter which was moving faster at the finish.
not sure this is anything to do with your thread, but it's a very important topic nevertheless.
because they take a measure of time and convert it into a measure of distance(idiotically) then it DOES matter how fast they are moving.
in your example it may not matter, but could i suggest a little test.
try to reverse the margins back to time, and see how close you come to reality.
it won't matter what the formula is they use, it CAN'T be accurate in ALL cases.
also i hope you are not assuming the formula they use is a constant at all tracks under all conditions, because if it's anything like over here, then it's all over the shop.
guys like carroll, whoever he is, are wrong when they try to figure the physical length of a horse, because it's totally irrelevant in this context, because it is a measure of time, NOT distance.
not sure this is anything to do with your thread, but it's a very important topic nevertheless.
because they take a measure of time and convert it into a measure of distance(idiotically) then it DOES matter how fast they are moving.
in your example it may not matter, but could i suggest a little test.
try to reverse the margins back to time, and see how close you come to reality.
it won't matter what the formula is they use, it CAN'T be accurate in ALL cases.
also i hope you are not assuming the formula they use is a constant at all tracks under all conditions, because if it's anything like over here, then it's all over the shop.
guys like carroll, whoever he is, are wrong when they try to figure the physical length of a horse, because it's totally irrelevant in this context, because it is a measure of time, NOT distance.
I don't see how it can matter how fast they are moving at the finish. The final time is known, then converted. If you know the conversion, you get the exact finishing time. If you don't, you won't. I can't speak for your country, but here it is pretty consistent. There are a few companies and they are very close.
steveb
08-14-2011, 07:46 PM
I don't see how it can matter how fast they are moving at the finish. The final time is known, then converted. If you know the conversion, you get the exact finishing time. If you don't, you won't. I can't speak for your country, but here it is pretty consistent. There are a few companies and they are very close.
you have no way of knowing it's close, unless you can get the actual times the individual horses ran.
and if you can do that then you don't need the margins anyway, because you have now got an accurate measurement.
instead of using an example where the margins are the same, imagine if horse 'a' was beaten 6 lengths by 'b'
i put it to you that 6 lengths is dependent on how fast it is travelling.
is it going 4 or 5 or 6 or 7 or lengths per second???
has it occurred to you that to get inaccurate margins, that they have taken accurate measures of time, and converted them to something less so?
the obvious question then becomes....why don't they supply the times in the first place that they already have?
GameTheory
08-14-2011, 07:49 PM
it won't matter what the formula is they use, it CAN'T be accurate in ALL cases.
Understand they are taking a known time and converting it to a fictional beaten lengths. So the beaten lengths is not accurate at all because it is just made up. But if they use a fixed value for the conversion, you should be able to convert the beaten lengths back to the exact time. Problem is the beaten lengths are rounded off, and like I said I could never find that elusive fixed value even though it should have been easy. I suspect there is an adjustment for the distance or maybe they actually use a dynamic value based on the actual speed they are going (fixed for each race, but not for all races). Maybe I will take a look at the trakus data and see how it matches up...
The whole idea of using lengths is stupid. All that matters is time.
The way they do it now makes no sense at all. As it is with much of this game.
you have no way of knowing it's close, unless you can get the actual times the individual horses ran.
and if you can do that then you don't need the margins anyway, because you have now got an accurate measurement.
instead of using an example where the margins are the same, imagine if horse 'a' was beaten 6 lengths by 'b'
i put it to you that 6 lengths is dependent on how fast it is travelling.
is it going 4 or 5 or 6 or 7 or lengths per second???
has it occurred to you that to get inaccurate margins, that they have taken accurate measures of time, and converted them to something less so?
the obvious question then becomes....why don't they supply the times in the first place that they already have?
I've timed more races on my computer with software than I could ever count, so I know it is close. Of course I don't want to do that for every race, and since I understand the system it isn't necessary anyway. Why do you say I have no way to know?
I can't tell you why they don't supply the actual times. Racing is always about 80 years behind on the technology front.
steveb
08-14-2011, 08:22 PM
Understand they are taking a known time and converting it to a fictional beaten lengths. So the beaten lengths is not accurate at all because it is just made up. But if they use a fixed value for the conversion, you should be able to convert the beaten lengths back to the exact time. Problem is the beaten lengths are rounded off, and like I said I could never find that elusive fixed value even though it should have been easy. I suspect there is an adjustment for the distance or maybe they actually use a dynamic value based on the actual speed they are going (fixed for each race, but not for all races). Maybe I will take a look at the trakus data and see how it matches up...
how could they make an adjustment for the speed that fits ALL??
and i don't care what formula they use it CAN'T be exact, simply because time and distance are two different things, and they are deriving one from the other.
oh, i'm 5'11'' tall, what's that in seconds. :lol:
i live 20 miles from flemington, home of the melbourne cup, now let's see how quickly i can get there.
i am a slow driver, my wife on the other hand is a maniac when it comes to putting the foot down.
will we arrive at the same time because some formula has it that speed = 40 miler per hour?
steveb
08-14-2011, 08:24 PM
I've timed more races on my computer with software than I could ever count, so I know it is close. Of course I don't want to do that for every race, and since I understand the system it isn't necessary anyway. Why do you say I have no way to know?
I can't tell you why they don't supply the actual times. Racing is always about 80 years behind on the technology front.
ok, i will leave you to it.
you're not the only guy with software, but that has no relevance here.
i am saying that UNLESS you have the OFFICIAL times, then you have no way of knowing how accurately you can convert margins back to time.
i simply say you wouldn't be able to do it.
go to singapore site, or HKJC, or many other places, where they will give you margins and times, and try it. bet you can't.
how important it is, is up the individual i suppose.
GameTheory
08-14-2011, 08:33 PM
how could they make an adjustment for the speed that fits ALL??
and i don't care what formula they use it CAN'T be exact, simply because time and distance are two different things, and they are deriving one from the other.
oh, i'm 5'11'' tall, what's that in seconds. :lol:
i live 20 miles from flemington, home of the melbourne cup, now let's see how quickly i can get there.
i am a slow driver, my wife on the other hand is a maniac when it comes to putting the foot down.
will we arrive at the same time because some formula has it that speed = 40 miler per hour?I don't think you get what we are saying. The horse is timed exactly with a teletimer. You subtract the exact known horse's time from the exact known winner's time and you get the difference. This difference is then translated to a FICTIONAL beaten lengths via a number ala
time_difference_in_seconds / .18 = fake beaten lengths
So if you know the fake beaten lengths, which is what they put in the past performances instead of the actual time, you should be able to take the winner's actual time (which they do publish), multiply the beaten lengths by .18 and add that to the winners time to get the exact time of the horse in question. That's what were are talking about in terms of exactness -- that if you know that constant (.18 or whatever) then you can convert beaten lengths back to time exactly (if they didn't round the beaten lengths, which they do). No one is saying the beaten lengths reflect reality in any way -- they are just made up as a by-product of whatever constant they choose. Get it?
steveb
08-14-2011, 08:41 PM
I don't think you get what we are saying. The horse is timed exactly with a teletimer. You subtract the exact known horse's time from the exact known winner's time and you get the difference. This difference is then translated to a FICTIONAL beaten lengths via a number ala
time_difference_in_seconds / .18 = fake beaten lengths
So if you know the fake beaten lengths, which is what they put in the past performances instead of the actual time, you should be able to take the winner's actual time (which they do publish), multiply the beaten lengths by .18 and add that to the winners time to get the exact time of the horse in question. That's what were are talking about in terms of exactness -- that if you know that constant (.18 or whatever) then you can convert beaten lengths back to time exactly (if they didn't round the beaten lengths, which they do). No one is saying the beaten lengths reflect reality in any way -- they are just made up as a by-product of whatever constant they choose. Get it?
i know EXACTLY what i am saying, and the bit i have made bold, is 100% wrong.
as i say in a previous post...try it for yourself.
if they are not all travelling the same speed then it can't work.
ok, i will leave you to it.
you're not the only guy with software, but that has no relevance here.
i am saying that UNLESS you have the OFFICIAL times, then you have no way of knowing how accurately you can convert margins back to time.
i simply say you wouldn't be able to do it.
go to singapore site, or HKJC, or many other places, where they will give you margins and times, and try it. bet you can't.
how important it is, is up the individual i suppose.
I have no interest in Singapore or Hong Kong racing, sorry.
I don't even need the official times. What I need is to able to time the difference between the winner and the also rans, then comparing that difference in time to the reported beaten lengths. You are making it out to be some impossible task, but it is not.
GameTheory
08-14-2011, 09:07 PM
i know EXACTLY what i am saying, and the bit i have made bold, is 100% wrong.
as i say in a previous post...try it for yourself.
if they are not all travelling the same speed then it can't work.But you don't know what I am saying. I tried.
If you can't see that if you have 1.0 and you divide it by .20 to get 5, and that if you have 5 and you know X was divided by .20 to get that 5, then it is simple algebra to realize that if you take the 5 and you multiply by .20 then you get 1.0 again (X), then I don't know what else to say. It is that simple. It has nothing to do with accuracy in terms of beaten lengths or relative speeds or whatever. You are talking about something else.
THAT'S HOW BEATEN LENGTHS are derived. They take times and divide by a constant. IT DOESN'T MATTER WHAT THE VALUE OF THE CONSTANT IS. This isn't theory. That's how it is done.
So to get back to times you reverse the process, using the same constant they do. Simple.
steveb
08-14-2011, 09:20 PM
I have no interest in Singapore or Hong Kong racing, sorry.
I don't even need the official times. What I need is to able to time the difference between the winner and the also rans, then comparing that difference in time to the reported beaten lengths. You are making it out to be some impossible task, but it is not.
i am not making it out to be anything other than what it is.
i have no interest in singapopre or hkjc either, but it does demonstrate why i know i am right, and you're not.
perhaps you are just worried you may discover something!! :)
your last paragraph is a contradiction, but i digress.
so you don't want accurate, inaccurate will suffice.
if you had accurate official, then you would not need to time yourself, and likely have no need of an inaccurate margin.
but whatever rocks your boat........
steveb
08-14-2011, 09:34 PM
But you don't know what I am saying. I tried.
If you can't see that if you have 1.0 and you divide it by .20 to get 5, and that if you have 5 and you know X was divided by .20 to get that 5, then it is simple algebra to realize that if you take the 5 and you multiply by .20 then you get 1.0 again (X), then I don't know what else to say. It is that simple. It has nothing to do with accuracy in terms of beaten lengths or relative speeds or whatever. You are talking about something else.
THAT'S HOW BEATEN LENGTHS are derived. They take times and divide by a constant. IT DOESN'T MATTER WHAT THE VALUE OF THE CONSTANT IS. This isn't theory. That's how it is done.
So to get back to times you reverse the process, using the same constant they do. Simple.
well carry on then, because you can not reverse it in a way that would mimic the official individual times constantly, and you are saying that you can with some simple formula. i say rubbish.
it doesn't take einstein to understand why that is so.
the way to prove it is by attempting to do it from places that give margin and time, and the procedure(if not the specific formula) will be the same as USA.
now i know you guys are too busy to want to know or understand, and maybe it's too much like hard work anyway.
as usual i am hoping to learn something, but this is going nowhere it appears, so this is last for me on subject.
i'm perfectly happy for you guys to believe what you like.
i am not making it out to be anything other than what it is.
i have no interest in singapopre or hkjc either, but it does demonstrate why i know i am right, and you're not.
perhaps you are just worried you may discover something!! :)
your last paragraph is a contradiction, but i digress.
so you don't want accurate, inaccurate will suffice.
if you had accurate official, then you would not need to time yourself, and likely have no need of an inaccurate margin.
but whatever rocks your boat........
I have taken two tours of the photo finish booth at the track. I have had the system explained to me by the men that do the work. Both told me the same things and they were 2000 miles apart. Maybe things are done differently in Hong Kong, but I know how they are done here. Good luck to you mate, this is going nowhere.
it doesn't take eintstein to understand why that is so.
FTFY
GameTheory
08-15-2011, 01:18 AM
well carry on then, because you can not reverse it in a way that would mimic the official individual times constantly, and you are saying that you can with some simple formula. i say rubbish.You can if the way they do things is the way we've been explaining. And it is, so you can. (Except you actually can't because of the rounding errors, as I keep saying. And it is unclear whether they use the same constant for all races or not, as I also have said.) I can't figure out why you are so stubborn about this -- maybe you can't believe how stupid the process actually is here in the U.S. But it is -- it really is that stupid.
Native Texan III
08-15-2011, 06:43 AM
The whole idea of using lengths is stupid. All that matters is time.
The way they do it now makes no sense at all. As it is with much of this game.
You are absolutely right.
The problem the authorities had was a customer base with "lengths" being embedded in their minds as an essential part of handicapping. Even though a length in horse racing is AFAIK has never been defined officially. The accepted concept of a horse being "beaten" is another daft one - as if the horses had a clue what that meant and got upset.
Racing being wrapped up in the long gone past perhaps they did not want to cause riots as caused by the missing 11 days when the calendar changed.
On September 2, 1752, an odd happening occurred that's still keeping genealogists on their toes. On that day, the British Isles and all the English colonies, including America, lost 11 days--September 3 through 13. People went to sleep and when they awoke the next morning, the date had changed to September 14. There were riots in rural villages since the people thought the government was trying to cheat them out of 11 days of their lives. Though these days disappeared in English lands in 1752, a number had already vanished in other places--France in 1582, Austria in 1584, and Norway in 1700.
Despite the official calendar, people in England and the colonies began to use the Gregorian system as early as the 16th century. Thus, many early colonial records include double dates, written as "12 February 1661/1662," indicating that, although it was officially 1661, some considered it to be 1662."
http://www.genealogytoday.com/columns/everyday/030902.html
maddog42
08-17-2011, 09:25 AM
I agree with most of your post except:
The problem the authorities had was a customer base with "lengths" being embedded in their minds as an essential part of handicapping. Even though a length in horse racing is AFAIK has never been defined officially. The accepted concept of a horse being "beaten" is another daft one - as if the horses had a clue what that meant and got upset.
Some horses get very upset when headed. I have ridden at least one that did.
Carroll talks about a very old horse that did also.
But, sicne they know BOTH lengths and actual times, why not proviode BOTH.
I know, I know, asking anyone in the racing game to use their heads is a fool's errand,, but come on guys, as customer, WE have to let them ALL know what a pathetic job they do day in and day out. By mistake, they might have hired someone who gives a shit recently who would actually listen to customers.
To put this back on topic, let me suggest another thing regarding the original post. There are probably times (there must be actually) where sprints and routes don't need to be split. However, with the chart possibly not accurate, that is going to be impossible to know.
maddog42
08-19-2011, 09:47 AM
To put this back on topic, let me suggest another thing regarding the original post. There are probably times (there must be actually) where sprints and routes don't need to be split. However, with the chart possibly not accurate, that is going to be impossible to know.
Have you ever figured routes and sprints slow or fast by percentage?
Have you ever figured routes and sprints slow or fast by percentage?
If I were going to do that, it wouldn't be by total time as that can be deceiving. It would be using a range of times that horses can run.
To put this back on topic, let me suggest another thing regarding the original post. There are probably times (there must be actually) where sprints and routes don't need to be split. However, with the chart possibly not accurate, that is going to be impossible to know.
For what it's worth I routinely make a single variant for both sprints and routes and only split them on rare occasions.
I do measure whether the first fraction of two turn routes is more or less fast than expected and make an adjustment to the route figures based on that measurement but it is a modest adjustment.
For what it's worth I routinely make a single variant for both sprints and routes and only split them on rare occasions.
I do measure whether the first fraction of two turn routes is more or less fast than expected and make an adjustment to the route figures based on that measurement but it is a modest adjustment.
I tend to agree that is the best way to go. Sample sizes are small enough as it is! I think the split thing is overdone. If you evaluate the fractions and which part of the track over which they are run, things tend to clear up in that regard.
TrifectaMike
08-22-2011, 12:39 PM
To put this back on topic, let me suggest another thing regarding the original post. There are probably times (there must be actually) where sprints and routes don't need to be split. However, with the chart possibly not accurate, that is going to be impossible to know.
CJ, what are your conclusions thus far?
Mike (Dr Beav)
Cratos
08-23-2011, 02:51 PM
CJ, what are your conclusions thus far?
Mike (Dr Beav)
This is one conclusion that I am anxious to see; particularly when I have been told that curve fitting has nothing to do with this issue.
CJ, what are your conclusions thus far?
Mike (Dr Beav)
I have concluded so far that the Beyer chart needs some serious updating. Not only is the difference between routes and sprints bad, but turf/synthetic to dirt is not very good either. I am not saying a horse should move between surfaces and run the best figures, but the best horses on each should be similar.
As for "the curve" it is possible that in some situations races that run too far away from the preferred curve would distort the data.
maddog42
12-18-2011, 10:05 PM
Ever notice that using the Beyer charts, the track is ALWAYS faster for sprints than for routes. What I mean is his variant always shows the track faster for sprints, and by always I mean about 98% of the time, if not more.
Doesn't this mean there is something wrong with the charts in the first place? The chart was supposed to show equal performances at different distances.
From Nick Mordin's book ON TIME:
.....It's probably obvious, but if a horse is, say, one second faster at a mile than another horse, it should be two seconds faster at 2 miles , three seconds faster at three miles and so on. The same goes for the various classes of race. Selling platers can get quite close to Group race times at five furlongs, but at two miles the difference on the clock is huge.
As I say, it's pretty obvious that the time difference between horses expands in line with the increase in distance- but, amazingly, most published speed ratings don't take this into account. That is why they rarely give high numbers for horses racing at long distances, except in top class races, where the compilers invariably adopt an extremely convoluted interpretation of the race times in order to produce ratings as high as the horses seem to merit.
Cj I know you have read this book, but I thought this might be informative for
some of the skeptics that we encountered on this thread.
vBulletin® v3.8.9, Copyright ©2000-2024, vBulletin Solutions, Inc.