Projected Final Times for use in making Variants

[osophy_junkie]

Originally Posted by cj

Here is my point on all of this...

If I had a machine that launched 1000 baseballs a day the same way every time, and on one day it averaged 400 feet, and on day two 380 feet, and day 3 420 feet, I could set the variants as 0, -20 feet, and +20 feet with no fear of being wrong.

Would it be possible to figure out why the days were different, and what factors contributed each day? Possibly. Would it help me get a better answer? Not really. Same goes for measuring track speed in my opinion.

[classhandicapper]

Originally Posted by raybo

Wouldn't the same apply to class pars? "Class" levels/groupings, depending on how you determine them, and how tightly you group them, are also subject to changes in racing, from surface changes/maintenance and weather, to purse levels and other artificial manipulations including ever increasing sets of qualifications for races for carding different, supposed, degrees of "class" horses, to more and more frequent movements of horses between tracks, surfaces, distances, class levels, etc., the list goes on and on. Determining class levels is a minefield, full of obstacles that will, intuitively and calculatively (sp?), force significant misrepresentations of "class".

I should have been clearer.

What you are saying is true, but this is not really about class. This is just to have something to use as a flag against my speed figure projections so I don't accidentally go way off track on a figure.

I would start by looking at someone else's speed figures for the last few years (call it 3 years) to create speed figure based Class Pars for as many classes as I could. If there were changes due to casino purses etc.. in the last few years, it would add a complication, but I'd have to start somewhere.

Once I had some Speed Figure Class PARs, they would simply be a backstop check against my projection variants.

Let's say I give a figure of 105 to some maiden that won by 5 lengths and the PAR was 82, there would be a red flag saying "are you sure about that 105, that's a 23 point difference?"

Then I'd know to take a better look at that race and that day.

Anything close, would not got flagged.

Once I was rolling with my own projections, I could recalculate the Class PARs every year using my figures on perhaps a rolling 3 year basis to stay current enough but still give me decent sample sizes.

For a lot of classes I'd have no PAR due to sample size or it being some whacky combination class, but that's OK. Where I have the data it will help me make figures and where I don't I'd still be making projection figures like everyone else.

[cj]

Originally Posted by classhandicapper

I should have been clearer.

What you are saying is true, but this is not really about class. This is just to have something to use as a flag against my speed figure projections so I don't accidentally go way off track on a figure.

I would start by looking at someone else's speed figures for the last few years (call it 3 years) to create speed figure based Class Pars for as many classes as I could. If there were changes due to casino purses etc.. in the last few years, it would add a complication, but I'd have to start somewhere.

Once I had some Speed Figure Class PARs, they would simply be a backstop check against my projection variants.

Let's say I give a figure of 105 to some maiden that won by 5 lengths and the PAR was 82, there would be a red flag saying "are you sure about that 105, that's a 23 point difference?"

Then I'd know to take a better look at that race and that day.

Anything close, would not got flagged.

Once I was rolling with my own projections, I could recalculate the Class PARs every year using my figures on perhaps a rolling 3 year basis to stay current enough but still give me decent sample sizes.

For a lot of classes I'd have no PAR due to sample size or it being some whacky combination class, but that's OK. Where I have the data it will help me make figures and where I don't I'd still be making projection figures like everyone else.

Pars can be of some help, but experience is better. I'll give an example, only sprint of the day is for MSW. Your par figure is 90, the horses run a 100 raw time. Most would use the par and assign a 90.

But, what if the winner was a 3-5 from Todd Pletcher that won by 6? What if the winner won from way off the pace after a wicked three horse battle on the front and was from a 2% trainer and was 30-1 and bred to go long?

[thaskalos]

Originally Posted by cj

Pars can be of some help, but experience is better. I'll give an example, only sprint of the day is for MSW. Your par figure is 90, the horses run a 100 raw time. Most would use the par and assign a 90.

But, what if the winner was a 3-5 from Todd Pletcher that won by 6? What if the winner won from way off the pace after a wicked three horse battle on the front and was from a 2% trainer and was 30-1 and bred to go long?

Or, what if the winner was a new and private acquisition of a hot-shot trainer with 40% winners first off the claim?

[cj]

Originally Posted by thaskalos

Or, what if the winner was a new and private acquisition of a hot-shot trainer with 40% winners first off the claim?

No doubt...horses like that I tend to ignore prior figures and look at the other horses, let the figure fall where it may for the move up horse.

[PhantomOnTour]

I use projected variants in conjunction with traditional class pars, but I don't try to project a fig for every race.
A low level state bred Mdn for 3yr old fillies isn't likely to get my attention when assigning variant/figs for the day.
Conversely, a Clm20 for 4+ with numerous veteran runners having established form is likely to be weighed heavily in computing my figs.

Sometimes a single race with the above type runners has determined my variant for the day.

[thaskalos]

Originally Posted by cj

No doubt...horses like that I tend to ignore prior figures and look at the other horses, let the figure fall where it may for the move up horse.

That's why the figures derived from those "projected track variants" are so enigmatic to me. They often have as much to do with the competence of the figure-maker, as they do with the ability of the horses in question.

It appears that the horseplayer needs as much handicapping experience in judging the merits of the figure-maker, as the figure-maker needs in judging the performances of the horses.

[cj]

Originally Posted by thaskalos

That's why the figures derived from those "projected track variants" are so enigmatic to me. They often have as much to do with the competence of the figure-maker, as they do with the ability of the horses in question.

It appears that the horseplayer needs as much handicapping experience in judging the merits of the figure-maker, as the figure-maker needs in judging the performances of the horses.

I hear you, but I've also seen sets of figures that are made from rigid rules without much human intervention, like Equibase figures...no thanks.

[thaskalos]

Originally Posted by cj

I hear you, but I've also seen sets of figures that are made from rigid rules without much human intervention, like Equibase figures...no thanks.

True. The Equibase figures are, without a doubt, the worst speed figures known to man. Ever since Trackmaster got their hands on them...they've been the laughing stock of the speed figure industry.

[classhandicapper]

Originally Posted by cj

Pars can be of some help, but experience is better. I'll give an example, only sprint of the day is for MSW. Your par figure is 90, the horses run a 100 raw time. Most would use the par and assign a 90.

But, what if the winner was a 3-5 from Todd Pletcher that won by 6? What if the winner won from way off the pace after a wicked three horse battle on the front and was from a 2% trainer and was 30-1 and bred to go long?

You are misunderstanding my use.

I am suggesting people make projections as you suggest, but I would have a process that FLASHED any figure that was more than X points away from the PAR.

So if a single race (like in your example) got flagged off my projection, I would immediately review it and see that it was a high quality Pletcher horse that won by an above average margin. I'd assume that the horse ran an above average figure and that the figure was totally legit.

Same if it was claimed by some move up guy etc...

It's the same exact process.

If the horse was from some run of the mill barn, there was a 3 way photo for the win with a couple of long shots, and no large gap to the horses in the back, I would know to be highly suspicious of that figure specifically because there would be no evidence of a high quality field but the figure would be huge relative to PAR. Therefore, something could be amiss with my projection or a prior figure.

If I made figures for an entire day using projections and I came up with fast by 3/5ths but my PARS showed the races were slow by 1/5th, I'd immediately know that it was either a day with a lot of very strong fields relative to the class Pars or my projections were off. So I'd take another look at the day.

The general use would only be to flag the outlier figures and days relative to the class PARs for review. That way you can reduce the risk of bad figures or drift.

If you don't know that the PAR for a NW2L 20K claimer for 3+ is 71, then if you project a 58 or an 83 you won't know that it could be suspicious.

[Tom]

Originally Posted by thaskalos

It appears that the horseplayer needs as much handicapping experience in judging the merits of the figure-maker, as the figure-maker needs in judging the performances of the horses.

That doesn't take very long.
It quickly becomes what figs are good and what ones are not.

[DeltaLover]

Originally Posted by Tom

It quickly becomes what figs are good and what ones are not.

As I have said several times before in this forum, this is best question we can ask about two sets of figures..

How do we really decide that Beyer is better than Equibase or Ragozin is better than thorograph?

I do not know of any simple answer to this fundamental question, although during the years I have created some metrics trying to measure the accuracy (and even worse the quality) of a speed figure methodology...

There are a lot of arguments against using traditional DRF variant, any kind of a class - pars based methodology (like beyer) a pure computer based approach and the same applies preety much to performance figures (like ragozin or thorograph)...

I have come to the conclusion that the best metric to compare figures, should be based on the ROI of the best betting strategy we can create by each one!

For example lets say we have two set of figures( figure 1 and 2) where best ROI for both is earned by betting the top last figure (this is by no means the norm as there might be many variantions to this rule and this is what handicapping is all about)!

Even if figure 1 appears to be way more accurate than figure 2, resulting to let's say 33% winners versus only 22%, figure 2 will be superior if it happens to produce a higher return.

This probably means that most of the times fig1 is close to all the other figures used by the public but some times fig2 appears to be diametrical different than the consensus and it happens to be correct frequently enough to result to a higher ROI.

Having said this, the real value of a figure depends on the cases when it contradicts the others and is proven correct...

Of course the example I gave, is an oversimplification of howthe real world, since the input we receive from any set of figures can be interpreted in infinite ways, thus we can never be sure that we know the ultimate way to apply it to the race as a whole..





I really do not believe that class pars can be used as the basis of a successfu

[MJC922]

Originally Posted by DeltaLover

As I have said several times before in this forum, this is best question we can ask about two sets of figures..

How do we really decide that Beyer is better than Equibase or Ragozin is better than thorograph?

To me accuracy and profitability are two entirely different things. Win% for top ranked figure in the most recent start is a metric that I would find very hard to believe is not a reliable indicator of who generates the most accurate performance figure. Take a look at that over a few thousand races and you'll have a good idea which is best. I wouldn't muddy the water with selective application of it either, let everyone be at the same disadvantage there... just take the horse's last race fig and see what it hits, period.

After that type of study, if you have two sets of numbers at the top that are essentially equal, e.g. one like the Beyer figure is widely distributed and the other one is less known to the public, the ROI for the latter should result in a better ROI for the type of strategy you're building around it. My own numbers lose 15% on top fig, really doesn't bother me in the least bit, I know they're reliable.

[MJC922]

I would add too, that IMO we should be very careful with using ROI as a guide to build a strategy piecemeal as it's very easy to go off into the weeds with that kind of thought process. One could do a study on public odds rank, notice the favorite has the 'best' ROI of all ranks and conclude that the ideal way to profit is to build an entire strategy around the favorite exclusively. IMO that's not at all the case, you're wagging the dog here with ROI. Follow Benter, find what's most predictive first, then figure out how to make money with it later.

[DeltaLover]

Originally Posted by MJC922

To me accuracy and profitability are two entirely different things.
....

My own numbers lose 15% on top fig, really doesn't bother me in the least bit, I know they're reliable.

But that's exactly the point, how do you decide that a specfic set of figures is more accurate than another? Can you define an objective methodology to do it? Do you have one or you just rely on your experience?