Speed Pars

[upthecreek]

I got into today via Twitter and mentioned on air with/by Rich Perloff today He misspoke about how the DRF computes their speed rating He wrongly said it was based on the track record which I quickly pointed out he was wrong, that it was based on the best time in last 3 years at each distance, which in turn he said he had no use for a 3 year old speed fig, he prefers figs based on pars Which got me thinking, how often is the Beyer, or Bris, or Timeform info updated ? Are the Beyer pars, speed fig/variants based on 5 yo , 10 yo data? Is Beyer still making figs for Saratoga on info he gathered in 1975? When the old/news Garden State open in 85, I made my own speed figs a la Quirin If they were still running today , I'm sure I would need new data

[cj]

I got into today via Twitter and mentioned on air with/by Rich Perloff today He misspoke about how the DRF computes their speed rating He wrongly said it was based on the track record which I quickly pointed out he was wrong, that it was based on the best time in last 3 years at each distance, which in turn he said he had no use for a 3 year old speed fig, he prefers figs based on pars Which got me thinking, how often is the Beyer, or Bris, or Timeform info updated ? Are the Beyer pars, speed fig/variants based on 5 yo , 10 yo data? Is Beyer still making figs for Saratoga on info he gathered in 1975? When the old/news Garden State open in 85, I made my own speed figs a la Quirin If they were still running today , I'm sure I would need new data

I can only speak for TimeformUS, but I don't use pars at all. I think the value is very limited given the vast assortment of race conditions there are today.

[upthecreek]

Thanks CJ I Tweeted that to Perloff,i like rattling his cage
Your right about the conditions It was much easier 30 years ago to keep pars for different classes I did well with my #'s in part because beside the raw drf speed rating ,I think the only other #'s were the Equibase #:s in the track program

[AltonKelsey]

Pars were always a crude method vs projection. I never used it, even back in the 70's

[sjk]

I use about 90% projection and 10% pars with the role of pars being to mitigate the effect of outliers.

The main need for me to have pars is to create par times; I don't know how else to make them. It does take some effort. I see that I have over 15,000 pars from Alb C1000 to Zia Str8000 (although I don't think I have ever made a bet at either of those tracks).

It would be a daunting challenge to create 15,000 pars all at once but when something builds over 20+ years it can get large unnoticed.

I too still have pars for GS but don't think they will ever be of use.

[NorCalGreg]

"Par" is just another term for "Baseline"...is it not?

How would anyone determine if any rating is poor, avg, or good without something to compare it with/against?

[sjk]

If it helps you make money it must be on balance good

[pandy]

I got into today via Twitter and mentioned on air with/by Rich Perloff today He misspoke about how the DRF computes their speed rating He wrongly said it was based on the track record which I quickly pointed out he was wrong, that it was based on the best time in last 3 years at each distance, which in turn he said he had no use for a 3 year old speed fig, he prefers figs based on pars Which got me thinking, how often is the Beyer, or Bris, or Timeform info updated ? Are the Beyer pars, speed fig/variants based on 5 yo , 10 yo data? Is Beyer still making figs for Saratoga on info he gathered in 1975? When the old/news Garden State open in 85, I made my own speed figs a la Quirin If they were still running today , I'm sure I would need new data

The tables that Beyer uses have been updated often. Speed figure making is not just about the par times, or projections. The parallel time charts are extremely important and it can take years of tweaking to get them to the point where they produce accurate ratings.

[RonTiller]

Jim Cramer does not use pars in his speed ratings, either for variants or track to track.

In the mid 90s, when we were relatively new in the business, we put a 3 year best speed rating in our files (still do), not because its a good number (it is not) but because some people wanted it.

We regularly got calls that our 3yb speed rating was wrong, because it did not match the one printed in the DRF. In paticular, this happened in Jan & Feb and Mar. The reason they didn't match was the DRF did not move up a year on Jan 1st - they still used the table of values they used the previous year. This got quite exasperating, defending our correct version of the 3yb speed rating against the "gold standard" incorrect 3yb number printed by the DRF.

The matter got even worse when, in mid summer, the numbers started mismatching again, after apparently the DRF did recalculate the 3 year best times for the previous 3 years. After wasting hours researching it, it became obvious that in mid summer, they had reverted back to the times they had used the previous years, after having used the correct 3 year best times for several months. Somebody probably had to restore a backup from somewhere and restored the previous year's values. So, the calls started all over again about how WE (AKA HDW) had the incorrect 3yb speed ratings in our files. Sigh...

The ultimate irony is that it probably made no difference whatsoever, as the number is not good to begin with.

Ron Tiller
HDW

[upthecreek]

Jim Cramer does not use pars in his speed ratings, either for variants or track to track.

In the mid 90s, when we were relatively new in the business, we put a 3 year best speed rating in our files (still do), not because its a good number (it is not) but because some people wanted it.

We regularly got calls that our 3yb speed rating was wrong, because it did not match the one printed in the DRF. In paticular, this happened in Jan & Feb and Mar. The reason they didn't match was the DRF did not move up a year on Jan 1st - they still used the table of values they used the previous year. This got quite exasperating, defending our correct version of the 3yb speed rating against the "gold standard" incorrect 3yb number printed by the DRF.

The matter got even worse when, in mid summer, the numbers started mismatching again, after apparently the DRF did recalculate the 3 year best times for the previous 3 years. After wasting hours researching it, it became obvious that in mid summer, they had reverted back to the times they had used the previous years, after having used the correct 3 year best times for several months. Somebody probably had to restore a backup from somewhere and restored the previous year's values. So, the calls started all over again about how WE (AKA HDW) had the incorrect 3yb speed ratings in our files. Sigh...

The ultimate irony is that it probably made no difference whatsoever, as the number is not good to begin with.

Ron Tiller
HDW
If you know how to use the drf #, it can be as godd as any other I use a variation of Quirin method, but I only make the #'s for Tam & Mth And I don't use 1 length = 1/5 of a second

[cj]

The tables that Beyer uses have been updated often. Speed figure making is not just about the par times, or projections. The parallel time charts are extremely important and it can take years of tweaking to get them to the point where they produce accurate ratings.


They were updated in the last year or so, but I certainly wouldn't say they have been updated often. He did tweak the tables for synthetics eventually and also tweaked the beaten lengths chart for turf, but for dirt they went basically unchanged for decades.

[AltonKelsey]

"Par" is just another term for "Baseline"...is it not?

How would anyone determine if any rating is poor, avg, or good without something to compare it with/against?

For purposes of betting, you only need accurate relative figures. If you're writing for the DRF and you're going for hyperbole, then you need pars.


Though I found it a little odd, Jerry Brown seemed to be saying he didn't care what the actual time of the Pegasus was, since he went off the other horses. Not sure I'm happy with that , but I believe that's what was said.

[classhandicapper]

Though I found it a little odd, Jerry Brown seemed to be saying he didn't care what the actual time of the Pegasus was, since he went off the other horses. Not sure I'm happy with that , but I believe that's what was said.

That's what he said.

He believes that tracks change speed multiple times a day due to the surface drying out, having water added, having maintenance done etc..

This is how I understand his thinking based on his comments over the year.

He tries to create a separate variant for every race. However, other races on the card help inform what he thinks about any individual race and stakes races with very consistent horses are easier to break out like that than horses with a less to go on. So for a race like Pegasus, the final time was almost irrelevant. He made a figure based on the horses' prior figures and the margins between horses, weight carried, and ground loss in the Pegasus.

[classhandicapper]

I don't check many tracks other than NYRA, but Beyer has done some tweaks to his published parallel time charts in NY.

[Tom]

6.5 furlongs at Aqu looks like it was changed a couple of ticks.

[classhandicapper]

6.5 furlongs at Aqu looks like it was changed a couple of ticks.

Yep. That one for sure. That may be a run up issue, but I don't have the data in front of me now.

[AltonKelsey]

That's what he said.

He believes that tracks change speed multiple times a day due to the surface drying out, having water added, having maintenance done etc..

This is how I understand his thinking based on his comments over the year.

He tries to create a separate variant for every race. However, other races on the card help inform what he thinks about any individual race and stakes races with very consistent horses are easier to break out like that than horses with a less to go on. So for a race like Pegasus, the final time was almost irrelevant. He made a figure based on the horses' prior figures and the margins between horses, weight carried, and ground loss in the Pegasus.

Problem with that is, he then assigned his fig based on pure projection and not on the actual performance. What's the point in that? I'd think a 4/5 time adjustment warranted some kind of tweak.

[AltonKelsey]

To expand on that, he's saying that no matter what the horses actually did on the race track , + or - a second, he would give the same figure.

Doesn't seem reasonable on the face of it, but he's no fool so maybe there's some subtlety I'm not seeing.

[cj]

To expand on that, he's saying that no matter what the horses actually did on the race track , + or - a second, he would give the same figure.

Doesn't seem reasonable on the face of it, but he's no fool so maybe there's some subtlety I'm not seeing.

I think he is way too liberal with the "changing speeds" thing personally. It covers up a lot of what he doesn't try to rate, mostly pace---again in my opinion.

[classhandicapper]

Doesn't seem reasonable on the face of it, but he's no fool so maybe there's some subtlety I'm not seeing.

There are trade offs in all these things.

If the track actually changed speeds, maybe you have a more accurate figure for a specific race. However, IMO, there are trip/pace related things going on that impact time. So you risk building those things into the figure via the variant and that can really throw off a trip handicapper.

[AltonKelsey]

There are trade offs in all these things.

If the track actually changed speeds, maybe you have a more accurate figure for a specific race. However, IMO, there are trip/pace related things going on that impact time. So you risk building those things into the figure via the variant and that can really throw off a trip handicapper.

Well, here's a pretty clear cut case where the reported time was wrong. Brown already had a somewhat different time from his clockers, so that somewhat dampens the difference, but still. You'd think even 2 lengths would change the fig, unless you say "well, I think the fig should be this, so that's what I'll give it"

[cj]

Well, here's a pretty clear cut case where the reported time was wrong. Brown already had a somewhat different time from his clockers, so that somewhat dampens the difference, but still. You'd think even 2 lengths would change the fig, unless you say "well, I think the fig should be this, so that's what I'll give it"

Does anyone really believe this on track clocker stuff?

[Twin Double]

I love the idea of perfect par times they just don't exist. When I think of par times I think of a perfect stair-step timetable that always shows cheaper horse run a little slower than the class above it. All the way from MCL to G1 races. Anyone who has made a serious effort to make par times knows the data never turns out like that. Anyone marketing par times in a perfect stair-step format is selling extrapolated data. Really isn't anything wrong with that because that's about all you can do to end up with data that makes sense.

I found the problem with par times is you can't make accurate par times without an accurate variant and you can't have an accurate variant without accurate par times. The perfect catch22. To have things make sense you have to start getting artsy with the numbers and while doing this you will start hearing that phrase in the back of your mind we all learned as kids - Garbage in Garbage out.

[Exotic1]

Does anyone really believe this on track clocker stuff?

The "trackman" missed the timing even after he knew what to look for. Does the same "trackman" do the ground work (ground loss)?

[classhandicapper]

I love the idea of perfect par times they just don't exist. When I think of par times I think of a perfect stair-step timetable that always shows cheaper horse run a little slower than the class above it. All the way from MCL to G1 races. Anyone who has made a serious effort to make par times knows the data never turns out like that. Anyone marketing par times in a perfect stair-step format is selling extrapolated data. Really isn't anything wrong with that because that's about all you can do to end up with data that makes sense.

Another problem is that in order to even estimate the relationships you have to use averages, but there is no guarantee that the relationships between fractions and final times, between sprints and routes, or between cheap horses and stakes horses all fit nicely into a single par time package based on averages. It could easily all vary day to day based on conditions that go beyond speed of track. We are all working with rough estimates of reality.

[rsetup]

There are any number of reasons for all these 'problems'.

One is that, as best as I can determine, no one is producing figures using even some semblance of statistical rigor. It's about projecting and then projecting on projections; and so on. Of course, all the followers (i.e. payers) all swear by the math of it all.

You can't get clean pars because the required variables to do so are not found in the splits. They are found in the beaten lengths, however. Doing the figures/pars/whatever is the simple part.

The hard part is categorizing the different race types. I'm confident that once the different types are identified, all these issues will be resolved. NOTE: I am not referring to PACE. Pace is better than simple speed figures but is a SUBSET of race type.

[Cratos]

There are any number of reasons for all these 'problems'.

One is that, as best as I can determine, no one is producing figures using even some semblance of statistical rigor. It's about projecting and then projecting on projections; and so on. Of course, all the followers (i.e. payers) all swear by the math of it all.

You can't get clean pars because the required variables to do so are not found in the splits. They are found in the beaten lengths, however. Doing the figures/pars/whatever is the simple part.

The hard part is categorizing the different race types. I'm confident that once the different types are identified, all these issues will be resolved. NOTE: I am not referring to PACE. Pace is better than simple speed figures but is a SUBSET of race type.

Excellent observation

[Cratos]

Another problem is that in order to even estimate the relationships you have to use averages, but there is no guarantee that the relationships between fractions and final times, between sprints and routes, or between cheap horses and stakes horses all fit nicely into a single par time package based on averages. It could easily all vary day to day based on conditions that go beyond speed of track. We are all working with rough estimates of reality.
Really? An estimate is a value that is inferred for a population based on data collected from a sample of units from that population. An estimate is not a guess, it is a value based on sampled data which has been adjusted using statistical estimation procedures.

[Exotic1]

There are any number of reasons for all these 'problems'.

One is that, as best as I can determine, no one is producing figures using even some semblance of statistical rigor. It's about projecting and then projecting on projections; and so on. Of course, all the followers (i.e. payers) all swear by the math of it all.

You can't get clean pars because the required variables to do so are not found in the splits. They are found in the beaten lengths, however. Doing the figures/pars/whatever is the simple part.

The hard part is categorizing the different race types. I'm confident that once the different types are identified, all these issues will be resolved. NOTE: I am not referring to PACE. Pace is better than simple speed figures but is a SUBSET of race type.

Yes, that could be a problem. There is certainly a built in and unavoidable margin of error. I don't mean to defend or indict any methodology but when we hear about "quarter point moves" on a graph, what the -- does that mean? Like the number before it was absolutely correct and the small move forward is absolutely correct?

[Nitro]

There are any number of reasons for all these 'problems'.

One is that, as best as I can determine, no one is producing figures using even some semblance of statistical rigor. It's about projecting and then projecting on projections; and so on. Of course, all the followers (i.e. payers) all swear by the math of it all.

You can't get clean pars because the required variables to do so are not found in the splits. They are found in the beaten lengths, however. Doing the figures/pars/whatever is the simple part.

The hard part is categorizing the different race types. I'm confident that once the different types are identified, all these issues will be resolved. NOTE: I am not referring to PACE. Pace is better than simple speed figures but is a SUBSET of race type.
I generally don’t involved in discussions like this only because I really don’t handicap anymore.
However, many years ago when I did, I also created very detailed Par values for all the races in NY. It did take an inordinate amount of time to not only produce, but to maintain as well. To be honest, I thought they were extremely accurate to the point where I could often predict the final times of upcoming races, once I witnessed how the first couple of races were run on any given day.

I think my method was a bit more involved than what you typically see being discussed on threads like this. When I would evaluate any previous race that was run I would determine the speed numbers based on Par values off of each time split. In other words, my Pars were not just based final times. They were based on each ¼ mile time from start to finish. I felt that this method was much more accurate because it automatically incorporated a Pace value into the final speed figures themselves. Because at the time I also kept individual speed profiles on every horse in training I was also able to make any necessary speed figure adjustments by comparing those profiles with each time split in the race.

I don’t envy anyone doing similar work today, especially doing multiple race jurisdictions and with so many Class variations it must be a real task.

[Twin Double]

I generally don’t involved in discussions like this only because I really don’t handicap anymore.
However, many years ago when I did, I also created very detailed Par values for all the races in NY. It did take an inordinate amount of time to not only produce, but to maintain as well. To be honest, I thought they were extremely accurate to the point where I could often predict the final times of upcoming races, once I witnessed how the first couple of races were run on any given day.

I think my method was a bit more involved than what you typically see being discussed on threads like this. When I would evaluate any previous race that was run I would determine the speed numbers based on Par values off of each time split. In other words, my Pars were not just based final times. They were based on each ¼ mile time from start to finish. I felt that this method was much more accurate because it automatically incorporated a Pace value into the final speed figures themselves. Because at the time I also kept individual speed profiles on every horse in training I was also able to make any necessary speed figure adjustments by comparing those profiles with each time split in the race.

I don’t envy anyone doing similar work today, especially doing multiple race jurisdictions and with so many Class variations it must be a real task.

You must have had a team of employees or literally compiled horse profiles like it was a full-time job. I wouldn't even know how to attempt this.

[classhandicapper]

Really? An estimate is a value that is inferred for a population based on data collected from a sample of units from that population. An estimate is not a guess, it is a value based on sampled data which has been adjusted using statistical estimation procedures.

I believe you missed the point of my note.

1. The charts never come out nice and neat when you look at the actual data, averages, and relationships. So figure makers have to make estimates.

2. But beyond that, one day horses of perfectly equal quality that run 6F in 113 might run 9F in 154. On another day, it might be 155. On another day it might be 153.

So if some figure maker averaged 153, 154, and 155 to be equal to 154 when trying to calculate the relationship between 6F and 9F, he would be ridiculously wrong on 2 of the 3 days despite the data (of course he would use larger samples, but still run into the same issue).

Many figure makers realize this and get around it by making separate variants for sprints and routes, but the same is true to a lesser degree between various sprint distances or various route distances. It's just much harder to notice because it might be a 1/5 or 2/5.

Now apply that to relationships between the fractions and final times and you have quite a mess on your hands even though you used data to arrive at your conclusions.

[classhandicapper]

When I would evaluate any previous race that was run I would determine the speed numbers based on Par values off of each time split. In other words, my Pars were not just based final times. They were based on each ¼ mile time from start to finish. I felt that this method was much more accurate because it automatically incorporated a Pace value into the final speed figures themselves.

If you were doing this many years ago, you were ahead of your time (at least in terms of publicly available information)

That's what CJ does.

People understand the potential benefit of adjusting figures for extreme paces, but I don't think many grasp the benefit when it comes to making the figures themselves. If you have a more accurate appraisal of how well a horse ran, you have a better appraisal of how fast the track was to begin with.

Let's say some horse ran 6F in 110 and you were expecting him to run 111. That says the track was 1 second fast. But if he ran 4f in 43 3/5 you might come to different conclusion about how well he ran than if the fraction was 45. That would change how fast you thought the track was.

[Cratos]

I believe you missed the point of my note.

1. The charts never come out nice and neat when you look at the actual data, averages, and relationships. So figure makers have to make estimates.

2. But beyond that, one day horses of perfectly equal quality that run 6F in 113 might run 9F in 154. On another day, it might be 155. On another day it might be 153.

So if some figure maker averaged 153, 154, and 155 to be equal to 154 when trying to calculate the relationship between 6F and 9F, he would be ridiculously wrong on 2 of the 3 days despite the data (of course he would use larger samples, but still run into the same issue).

Many figure makers realize this and get around it by making separate variants for sprints and routes, but the same is true to a lesser degree between various sprint distances or various route distances. It's just much harder to notice because it might be a 1/5 or 2/5.

Now apply that to relationships between the fractions and final times and you have quite a mess on your hands even though you used data to arrive at your conclusions.
If I missed your point I apologize, but I wasn’t responding to how “speed figures” are calculated, but to your assertion: “Another problem is that in order to even estimate the relationships you have to use averages.”

That prompted me to question your use and understanding of “statistical estimation procedures.”

[Nitro]

You must have had a team of employees or literally compiled horse profiles like it was a full-time job. I wouldn't even know how to attempt this.No unfortunately that's not the case. It was just my brother and I doing all the work. Like I said it was very tedious to not only set up, but to maintain as well. Fortunately, they only run 1 track at a time in NY. I couldn’t even imagine doing multiple tracks!

[Nitro]

If you were doing this many years ago, you were ahead of your time (at least in terms of publicly available information)

That's what CJ does.

People understand the potential benefit of adjusting figures for extreme paces, but I don't think many grasp the benefit when it comes to making the figures themselves. If you have a more accurate appraisal of how well a horse ran, you have a better appraisal of how fast the track was to begin with.

Let's say some horse ran 6F in 110 and you were expecting him to run 111. That says the track was 1 second fast. But if he ran 4f in 43 3/5 you might come to different conclusion about how well he ran than if the fraction was 45. That would change how fast you thought the track was.That’s absolutely correct.

Well we certainly didn’t publish our information, and you’re right! At the time (early 80’s), we weren’t aware of anyone spending that much time breaking a race down like that. We actually started doing it because we recognized how a simple thing like the directional Wind factor could impact these values. The Pace factor actually just fell into overall equation by accident. It could only be appreciated when making comparisons to the individual horse profiles.

[classhandicapper]

That’s absolutely correct.

Well we certainly didn’t publish our information, and you’re right! At the time (early 80’s), we weren’t aware of anyone spending that much time breaking a race down like that. We actually started doing it because we recognized how a simple thing like the directional Wind factor could impact these values. The Pace factor actually just fell into overall equation by accident. It could only be appreciated when making comparisons to the individual horse profiles.

Why did you stop handicapping?

I know it's not the most lucrative endeavor for the vast majority of people but if you were that far along 30-35 years ago there's no telling where you be now.

I started tinkering with my own pace figures on the Beyer scale at around the same time as you but didn't start putting it all together until much later. It's very complex. My thinking is still evolving.

[Nitro]

Why did you stop handicapping?

I know it's not the most lucrative endeavor for the vast majority of people but if you were that far along 30-35 years ago there's no telling where you be now.

I started tinkering with my own pace figures on the Beyer scale at around the same time as you but didn't start putting it all together until much later. It's very complex. My thinking is still evolving.Well to be perfectly honest, no matter how well we scrutinized everything about a future race, we always came across some results that defied the numbers. Sometimes I think we went a bit overboard because as thorough as our evaluations might have been, we were in fact treating these animals like precision running machines. Don’t get me wrong we had our fair share of winning plays, but at times we felt like there was something missing.

About 8 years ago I came across a guy who I initially thought was off the wall when came to playing the horses. Today I’m proud to call him my mentor because he’s taught me just about everything I know about how the tote analysis works and how to take advantage of what it produces. It’s also allowed me to focus more on the betting and money management part of the game which I personally believe is even more important. From time-to-time I enjoy sharing my live picks in the Selection forum. So, I guess the best way I can answer your question and justify my decision is to use a recent example of just how powerful a tool it can be.
http://www.paceadvantage.com/forum/showthread.php?t=136288&page=1
(BTW because I’m not the developer, I have no idea how it analyzes all of the betting pools at the same time)

[Twin Double]

Well to be perfectly honest, no matter how well we scrutinized everything about a future race, we always came across some results that defied the numbers. Sometimes I think we went a bit overboard because as thorough as our evaluations might have been, we were in fact treating these animals like precision running machines. Don’t get me wrong we had our fair share of winning plays, but at times we felt like there was something missing.

About 8 years ago I came across a guy who I initially thought was off the wall when came to playing the horses. Today I’m proud to call him my mentor because he’s taught me just about everything I know about how the tote analysis works and how to take advantage of what it produces. It’s also allowed me to focus more on the betting and money management part of the game which I personally believe is even more important. From time-to-time I enjoy sharing my live picks in the Selection forum. So, I guess the best way I can answer your question and justify my decision is to use a recent example of just how powerful a tool it can be.
http://www.paceadvantage.com/forum/showthread.php?t=136288&page=1
(BTW because I’m not the developer, I have no idea how it analyzes all of the betting pools at the same time)

I think this is just part of the natural evolution of a handicapper. I'm still guilty of this at times. It sounds like common sense, of course, their not machines but when involved in that personal pursuit of perfect handicapping it's easy to lose sight of the basic truths of the game.

[steveb]

I love the idea of perfect par times they just don't exist. When I think of par times I think of a perfect stair-step timetable that always shows cheaper horse run a little slower than the class above it. All the way from MCL to G1 races. Anyone who has made a serious effort to make par times knows the data never turns out like that. Anyone marketing par times in a perfect stair-step format is selling extrapolated data. Really isn't anything wrong with that because that's about all you can do to end up with data that makes sense.

I found the problem with par times is you can't make accurate par times without an accurate variant and you can't have an accurate variant without accurate par times. The perfect catch22. To have things make sense you have to start getting artsy with the numbers and while doing this you will start hearing that phrase in the back of your mind we all learned as kids - Garbage in Garbage out.


first bold bit....
sorry to disappoint you, but it does actually turn out exactly like that, provided you have the data AND the know how.

i have no idea if usa has official ratings or not, but that would be one way of defeating the problem of the lack of races of a particular type.

your expected speed could be figured by simple linear regression with the independent variable being the official rating average of the horses in any particular race.
as per attachment(hk because i was already doing something there,but works the same anywhere, even australia where there is 7 zillion tracks)

is it exact,.......probably not.
does it work........definitely


second bold bit....
if you were to make time standards(pars)/classes/speeds/variants all at the same time then it works just fine.
each defines the other.
trial and error is your friend.

[cj]

first bold bit....
sorry to disappoint you, but it does actually turn out exactly like that, provided you have the data AND the know how.

i have no idea if usa has official ratings or not, but that would be one way of defeating the problem of the lack of races of a particular type.

your expected speed could be figured by simple linear regression with the independent variable being the official rating average of the horses in any particular race.
as per attachment(hk because i was already doing something there,but works the same anywhere, even australia where there is 7 zillion tracks)

is it exact,.......probably not.
does it work........definitely


second bold bit....
if you were to make time standards(pars)/classes/speeds/variants all at the same time then it works just fine.
each defines the other.
trial and error is your friend.

Very sharp post here.

[Twin Double]

first bold bit....
sorry to disappoint you, but it does actually turn out exactly like that, provided you have the data AND the know how.

i have no idea if usa has official ratings or not, but that would be one way of defeating the problem of the lack of races of a particular type.

your expected speed could be figured by simple linear regression with the independent variable being the official rating average of the horses in any particular race.
as per attachment(hk because i was already doing something there,but works the same anywhere, even australia where there is 7 zillion tracks)

is it exact,.......probably not.
does it work........definitely


second bold bit....
if you were to make time standards(pars)/classes/speeds/variants all at the same time then it works just fine.
each defines the other.
trial and error is your friend.

First off you're not disappointing me I'm not anti par times. I use them myself. Second I do like the data table you posted. Not sure about racing in Australia here in the states you have so many different class structures and track changes within a circuit you can't get a huge sample so you rely on extrapolated data. The whole point of my post was that you can't just record a bunch of times throw out the outliers and get a perfect stair-step table. It takes additional manipulation and without the true variant for each of your samples it clouds things further

[Twin Double]

first bold bit....
sorry to disappoint you, but it does actually turn out exactly like that, provided you have the data AND the know how.

i have no idea if usa has official ratings or not, but that would be one way of defeating the problem of the lack of races of a particular type.

your expected speed could be figured by simple linear regression with the independent variable being the official rating average of the horses in any particular race.
as per attachment(hk because i was already doing something there,but works the same anywhere, even australia where there is 7 zillion tracks)

is it exact,.......probably not.
does it work........definitely


second bold bit....
if you were to make time standards(pars)/classes/speeds/variants all at the same time then it works just fine.
each defines the other.
trial and error is your friend.

If they ran the same race types all day I guess this would work. I'm not exactly the sharpest crayon in the box. I was just sharing my personal experience with the issue. I'm definitely no handicapping authority.

[steveb]

First off you're not disappointing me I'm not anti par times. I use them myself. Second I do like the data table you posted. Not sure about racing in Australia here in the states you have so many different class structures and track changes within a circuit you can't get a huge sample so you rely on extrapolated data. The whole point of my post was that you can't just record a bunch of times throw out the outliers and get a perfect stair-step table. It takes additional manipulation and without the true variant for each of your samples it clouds things further


too may preconceptions.
you need to change the word 'you' to 'me' or 'I'

personally, i make no judgements myself.
i let my methods do it.
for instance, i may have some simple formula to make a time for any distance anywhere.
i may make all the classes(from the very best G1 to the lowest of the low), exactly the same.

iterate through all the date once and see what it throws up......
bloody hell, that's not too good.

computer then makes some adjustments to times and classes, and iterates through it all again.

wow, it just got a lot closer to the classes=equalling the speeds at each distance at each track.

lets make another adjustment, and iterate through the data again.
crikey it's closer still.

make another adjustment, and check.
jeepers creepers, all the classes equal the speeds at every track over every distance, to 3 decimal points no less!!
no more iteration needed, it simply can't get it any better.

not a variant in sight either.

[Twin Double]

too may preconceptions.
you need to change the word 'you' to 'me' or 'I'

personally, i make no judgements myself.
i let my methods do it.
for instance, i may have some simple formula to make a time for any distance anywhere.
i may make all the classes(from the very best G1 to the lowest of the low), exactly the same.

iterate through all the date once and see what it throws up......
bloody hell, that's not too good.

computer then makes some adjustments to times and classes, and iterates through it all again.

wow, it just got a lot closer to the classes=equalling the speeds at each distance at each track.

lets make another adjustment, and iterate through the data again.
crikey it's closer still.

make another adjustment, and check.
jeepers creepers, all the classes equal the speeds at every track over every distance, to 3 decimal points no less!!
no more iteration needed, it simply can't get it any better.

not a variant in sight either.

Like I said earlier, I was only posting my personal issues embarking on the par time journey. My data never worked out that consistent without hard manipulation by myself. I guess I need to tackle my pars using a different method.

[Twin Double]

too may preconceptions.
you need to change the word 'you' to 'me' or 'I'

personally, i make no judgements myself.
i let my methods do it.
for instance, i may have some simple formula to make a time for any distance anywhere.
i may make all the classes(from the very best G1 to the lowest of the low), exactly the same.

iterate through all the date once and see what it throws up......
bloody hell, that's not too good.

computer then makes some adjustments to times and classes, and iterates through it all again.

wow, it just got a lot closer to the classes=equalling the speeds at each distance at each track.

lets make another adjustment, and iterate through the data again.
crikey it's closer still.

make another adjustment, and check.
jeepers creepers, all the classes equal the speeds at every track over every distance, to 3 decimal points no less!!
no more iteration needed, it simply can't get it any better.

not a variant in sight either.

It sounds like you're able to create your pars applying the same algorithm to refine the data. I haven't been able to do that. I guess it's nice to know it can be done :)

[Nitro]

first bold bit....
sorry to disappoint you, but it does actually turn out exactly like that, provided you have the data AND the know how.

is it exact,.......probably not.
does it work........definitely

second bold bit....
if you were to make time standards(pars)/classes/speeds/variants all at the same time then it works just fine.
each defines the other.
trial and error is your friend.
If they ran the same race types all day I guess this would work. I'm not exactly the sharpest crayon in the box. I was just sharing my personal experience with the issue. I'm definitely no handicapping authority.To some extent you're correct.
Don’t let it bother you Twin Double because all of the time standards mentioned are interrelated and totally dependent upon one another from an evaluation point of view. Unfortunately that evaluation is based on the past and is pure speculation for the future. Only the Insiders have the knowledge to health and condition of their horses on race day.

I’m not sure how anyone can use just the numbers without individual horse profiles. The profiles I used indicated the actual individual split times for each at race at each distance, best distances run, preferred running style, optimum pace…etc., etc.

If you’re interested in this sort of stuff, it’s another good reason for playing the races in HK. The HKJC provides all of that information and more free of charge.

[Twin Double]

To some extent you're correct.
Don’t let it bother you Twin Double because all of the time standards mentioned are interrelated and totally dependent upon one another from an evaluation point of view. Unfortunately that evaluation is based on the past and is pure speculation for the future. Only the Insiders have the knowledge to health and condition of their horses on race day.

I’m not sure how anyone can use just the numbers without individual horse profiles. The profiles I used indicated the actual individual split times for each at race at each distance, best distances run, preferred running style, optimum pace…etc., etc.

If you’re interested in this sort of stuff, it’s another good reason for playing the races in HK. The HKJC provides all of that information and more free of charge.

Yeah, HK seems to have an interesting product. Personally never tried to handicap any of their races but would watch it on TV late at night sometimes. Always seems to be nice full fields and wide open races at least according to the tote board.

[steveb]

To some extent you're correct.
Don’t let it bother you Twin Double because all of the time standards mentioned are interrelated and totally dependent upon one another from an evaluation point of view. Unfortunately that evaluation is based on the past and is pure speculation for the future. Only the Insiders have the knowledge to health and condition of their horses on race day.

I’m not sure how anyone can use just the numbers without individual horse profiles. The profiles I used indicated the actual individual split times for each at race at each distance, best distances run, preferred running style, optimum pace…etc., etc.

If you’re interested in this sort of stuff, it’s another good reason for playing the races in HK. The HKJC provides all of that information and more free of charge.


i am sorry to be the bearer of bad news, but it's not the insiders that have the knowledge, it's the computer teams there.
that's whose money you are likely following, not insiders.


anyway nitro, i would love to respond to your post, but i can't follow it.
all i put up was how i find time and class.....ie: BASELINES
nothing more.
individual horses/ sections/running styles/ and whatever else is another subject completely.



while the hkjc is very very good at supplying info, there would be many people that make lots of money, praying to imaginary deities, that all their opponents(other gamblers), used the par/class/section times on their site.

they are colour by numbers stuff, and anybody with even a rudimentary knowledge of same, could pick up glaring errors, with simply a glance at them

[Twin Double]

i am sorry to be the bearer of bad news, but it's not the insiders that have the knowledge, it's the computer teams there.
that's whose money you are likely following, not insiders.


anyway nitro, i would love to respond to your post, but i can't follow it.
all i put up was how i find time and class.....ie: BASELINES
nothing more.
individual horses/ sections/running styles/ and whatever else is another subject completely.



while the hkjc is very very good at supplying info, there would be many people that make lots of money, praying to imaginary deities, that all their opponents(other gamblers), used the par/class/section times on their site.

they are colour by numbers stuff, and anybody with even a rudimentary knowledge of same, could pick up glaring errors, with simply a glance at them

Your one smug ass SOB aren't you

[steveb]

Your one smug ass SOB aren't you

as you wish.
some other person might say that i helped them, but whatever you like.
not a problem.

[Nitro]

i am sorry to be the bearer of bad news, but it's not the insiders that have the knowledge, it's the computer teams there.
that's whose money you are likely following, not insiders.

while the hkjc is very very good at supplying info, there would be many people that make lots of money, praying to imaginary deities, that all their opponents(other gamblers), used the par/class/section times on their site.

they are colour by numbers stuff, and anybody with even a rudimentary knowledge of same, could pick up glaring errors, with simply a glance at them
Don’t get me wrong, just because I’m saying that the HKJC offers a substantial amount of speed and profile information that I’m promoting its use. Its relevancy is for those who might be interested to determine.

Regarding the source of well intentioned betting, I beg to differ with you. After carefully following the racing in HK in combination with the tote analysis for the last 4 years you’ll never convince me that I’m following just those relying on computers to make large wagers. I can state unequivocally that there is certainly insider action. The key is being able to differentiate it by knowing when and where that money is going. From the perspective of the analysis output it’s actually much easier to identify those specific activities. That’s primarily because of the large race fields which generally average 10 to 14 entries at Sha Tin and 12 entries at Happy Valley. In spite of the huge Win, Place and Quinella pools, the analysis can consistently discriminate between those entries of interest and those getting just ordinary play.
But even if I’m 100% wrong, your news is certainly not “bad”, because either way it’s provided me with some very valuable insight into great game halfway around the world.

[steveb]

Don’t get me wrong, just because I’m saying that the HKJC offers a substantial amount of speed and profile information that I’m promoting its use. Its relevancy is for those who might be interested to determine.

Regarding the source of well intentioned betting, I beg to differ with you. After carefully following the racing in HK in combination with the tote analysis for the last 4 years you’ll never convince me that I’m following just those relying on computers to make large wagers. I can state unequivocally that there is certainly insider action. The key is being able to differentiate it by knowing when and where that money is going. From the perspective of the analysis output it’s actually much easier to identify those specific activities. That’s primarily because of the large race fields which generally average 10 to 14 entries at Sha Tin and 12 entries at Happy Valley. In spite of the huge Win, Place and Quinella pools, the analysis can consistently discriminate between those entries of interest and those getting just ordinary play.
But even if I’m 100% wrong, your news is certainly not “bad”, because either way it’s provided me with some very valuable insight into great game halfway around the world.

hk is where it all begun.
i guess computer teams is not the right word, the massive betting syndicates, is what i am referring to, and they will dwarf the tote turnover of all the other players.
that's where they cut their teeth, and as far as i know still operating.
their computers pumping out the probs for races there, but i guess most of them could well be computer illiterate for all i know.
only those that compile the models need know computers i suppose.

i know it is not 'bad' for you, i just like to take a bit of licence sometimes!!
i am probably one of the few that knows it can be done, the way you appear to do it, not that i would do it that way myself, if i still was interested.
personally i have never been interested in what other players(insider or not) do
but even those massive 'betting syndicates':) will have factors based on what the tote board is saying.
and depending on where, what betfair is saying.

actually i don't really believe in 'insiders' as such, but i do the syndicates, because, well, just because.

[classhandicapper]

Like I said earlier, I was only posting my personal issues embarking on the par time journey. My data never worked out that consistent without hard manipulation by myself. I guess I need to tackle my pars using a different method.

I don't think you are wrong. There are ways of fudging the answers, but they are all wrong some of the time. The bottom line is that we no matter what methodology you are using you are dealing with approximations of reality that are often wrong.

[Tom]

When it come to making figures, those who can, do, and those who can't say that those who can can't.

Rationalizing ones limitations is common.