using variants

[shoelessjoe]

This is something I came accross in Bill Olmsted's book Lone Speed Method,I dont think he will mind.Consider these 2 horses both are 20,000 claimers that ran 6 furlongs on different days.


HORSE A 21.4 46.1 59.1 112.1 1-NK 1-1 1-3 1-5

HORSE B 22.1 47.1 59.3 112.3 3-2 3-1 4-3 4-NK

At first you would think A is the faster horse but really B is because on the day A ran the track was lightning fast with cheap claimers running times that alw horses usually do.On the day B ran the track was deep and slow.

Comparing raw fractional times can at times be misleading because racing surfaces can changer radically fom day to day producing slower average times one day and faster times the next.


Being on the east coast I would say Bill hit it dead on.

[proximity]

Comparing raw fractional times can at times be misleading because racing surfaces can changer radically fom day to day producing slower average times one day and faster times the next..

the surface/conditions can also drastically OR gradually change DURING the same card too. if you play alot of east coast tracks, i recommend the henry kuck ratings from woodside associates.

[shoelessjoe]

good point

[cj]

1st call variants alone are very difficult to make. Using the "overall" track variant for a fraction is fraught with problems.

[bobphilo]

1st call variants alone are very difficult to make. Using the "overall" track variant for a fraction is fraught with problems.

On the other hand, there is also a serious problem with computing a seperate pace variant based on fractional times. You're introducing the confounding variable that the fractions may also be influenced by rating and jockey tactics rather than just the speed of the track.

Bob

[cj]

On the other hand, there is also a serious problem with computing a seperate pace variant based on fractional times. You're introducing the confounding variable that the fractions may also be influenced by rating and jockey tactics rather than just the speed of the track.

Bob

No doubt about it, which is why I don't do either stand alone. I have long maintained that they should be looked at in tandem to get the best overall picture. It certainly isn't a perfect picture even that way.

[proximity]

1st call variants alone are very difficult to make. Using the "overall" track variant for a fraction is fraught with problems.

i would agree with all that. many times in route races though you do get a better picture of the shape of the race by looking at the 1/2.

[proximity]

On the other hand, there is also a serious problem with computing a seperate pace variant based on fractional times. You're introducing the confounding variable that the fractions may also be influenced by rating and jockey tactics rather than just the speed of the track.

Bob

the fractions definitely SHOULD be influenced by rating and jockey tactics and this should show up as a "tightening" in your pace charts as race distances get longer. using proper charts along with medians (instead of all inclusive averages) and possibly ignoring horses who set fast paces but didn't contend at the finish will help get around these difficulties.

[raybo]

1st call variants alone are very difficult to make. Using the "overall" track variant for a fraction is fraught with problems.

CJ, since you make your own track variants, what is your personal opinion of the DRF track variant, as an overall variant not as a pace variant.

Is it adequate for general usage as a basic surface speed indication?

What other available adjustments could be made to the DRF for use in calculations? I've seen the "less than 16 - subtract 1 point per point of variant below 16, and if more than 19 - add 1 point per point of variant above 19, and between 16 and 19 no adjustment".

I'm speaking from the perspective of one who does not have an extensive database of races and results from which to arrive at my own variants.

[LottaKash]

HORSE A 21.4 46.1 59.1 112.1 1-NK 1-1 1-3 1-5

HORSE B 22.1 47.1 59.3 112.3 3-2 3-1 4-3 4-NK
.

I view this in different light....Basically I believe that the fractions of a race are very fluid, dynamic and ever changing, from day to day, week/month to week/month, and season to season as well as weather to weather....I think it is a very tall order to try and interpret a variant, because of the makeup of the contestants and the classifications that the horses are competing in are always in a state-of-flux...Thus, I have all but abandoned trying to make a variant, and I let Trackmaster take care of that worry for me and just loosely accept the numbers that they provide with their past perfomances....

So, how do I view this comparison ? (between A & B)...When viewing the internal fractions, It is more of an attempt to determine where an individual horse may be in his form cylcle,rather that how fast or slow his previous pace may have been, and this is usually evident by his reaction to the fractions, and how well he finished his race in spite of or despite the trip he may have had compared to the fractions that he may have had to contend with.... A horse's reaction and finish to a particular pace speaks volumes in where he may be in his FORM-CYCLE....Trying to predict how fast and in what fractions a horse may run next week is a very tall order, as the everchanging pace scenarios and it's resulting fractions, may yield highly volatile and/or unpredictable fractions on any given contest on race-day....So what is normal ? (variant wise) .... and what signifcance does it have on todays reality ?.......Over the years I believe I have wasted an inordinate amount of time on this particular aspect of handicapping....I just let the provided SR's do their thing.....

best,

[llegend39]

CJ, since you make your own track variants, what is your personal opinion of the DRF track variant, as an overall variant not as a pace variant.

Is it adequate for general usage as a basic surface speed indication?

What other available adjustments could be made to the DRF for use in calculations? I've seen the "less than 16 - subtract 1 point per point of variant below 16, and if more than 19 - add 1 point per point of variant above 19, and between 16 and 19 no adjustment".

I'm speaking from the perspective of one who does not have an extensive database of races and results from which to arrive at my own variants.

I use the form variant but I figure the avg sprint and route variant for the particular track. You can take 30-50 races and avg all sprint variants on a fast track and do the same for routes. (IE) at Phila Park i know that the avg sprint var=17 and rt=22 then using the Scott 2 for 1 on a day when the sprint variant is 15 lets say I assume the track was 1 fast.

[cnollfan]

CJ, since you make your own track variants, what is your personal opinion of the DRF track variant, as an overall variant not as a pace variant.

Is it adequate for general usage as a basic surface speed indication?

What other available adjustments could be made to the DRF for use in calculations? I've seen the "less than 16 - subtract 1 point per point of variant below 16, and if more than 19 - add 1 point per point of variant above 19, and between 16 and 19 no adjustment".

I'm speaking from the perspective of one who does not have an extensive database of races and results from which to arrive at my own variants.

In a pinch I will use 2/3 of the DRF variant, figuring 1/3 is due to the quality of the horses running that day.

[cj]

CJ, since you make your own track variants, what is your personal opinion of the DRF track variant, as an overall variant not as a pace variant.

Is it adequate for general usage as a basic surface speed indication?...



My personal opinion is no, it is not adequate. It is better than nothing at all.

[raybo]

My personal opinion is no, it is not adequate. It is better than nothing at all.

Fair enough CJ, I'll take you at your word and go from there. I guess I'll have to keep on "grinding" with creating a database (God, I hate my inabilities in this area!). I've been working on and with computers since 1970 and database creation and usage still bewilders me and this disgusts me to no end.

My kingdom for a racing database template!

ps. this, in no way, means that I will ever buy or subscribe to a "service" that has one. I'm far too cheap for that, I'll eventually find one or somehow create my own. If I can't figure out how to get the information I want without spending even more money than I already do, I'll just do without.

[raybo]

In a pinch I will use 2/3 of the DRF variant, figuring 1/3 is due to the quality of the horses running that day.

That 1/3 that you figure is due to the quality of horses running that day is important too, wouldn't you say? How do you utilize that information?

I know that it has nothing to do with the "true" surface speed that day, but it is an indication of what the horse had to contend with.

[cnollfan]

That 1/3 that you figure is due to the quality of horses running that day is important too, wouldn't you say? How do you utilize that information?

I know that it has nothing to do with the "true" surface speed that day, but it is an indication of what the horse had to contend with.

All I meant was that the DRF variant does not take into account the quality of the horses running that day (high class races are faster than low class races, which makes the raw variant lower on a classy day if class is ignored, as the DRF variant does). Also, DRF variant does not take into account weird track records at less frequently run distances. So, if DRF variant is 18, I will use 12 as the variant instead of 18 as a rough way to adjust for those intangibles. Very rough, I'll admit.

[raybo]

All I meant was that the DRF variant does not take into account the quality of the horses running that day (high class races are faster than low class races, which makes the raw variant lower on a classy day if class is ignored, as the DRF variant does). Also, DRF variant does not take into account weird track records at less frequently run distances. So, if DRF variant is 18, I will use 12 as the variant instead of 18 as a rough way to adjust for those intangibles. Very rough, I'll admit.

Understood.

[Pell Mell]

I know you will all think I'm crazy but back 40-50 yrs ago I was into pace and speed and tried various ways to get a decent track variant. I tried different adjustments to the DRF variants and I put a lot of labor into making pars for the NJ and Pa tracks. As some of you have mentioned, the class of the race and jockey tactics have a profound effect on pace. I had often thought that, unless the track was off due to weather conditions, the track didn't change all that much from day to day, especially fast tracks. Knowing that the pace of the race determined the final times and that the pace was determined by either the class of the horses or jockey tactics I thought that logic dictated that a variant that only considered the track condition could be obtained by getting pars from the only portion of the race where everyone, except the faders, was running full out. This of course, is the last 1/4 or 3/8ths of the race. For 2 yrs I charted the last 1/4s of all the races at PP. NOT the last 1/4 of the race but of the horse in each race that ran the last 1/4 the fastest. I found that there was a window of 2-4 seconds between the fastest and slowest, on average. I then made my variant based on this and it turned out to be the most accurate variant I had ever used, believe it or not. I no longer use figures in my handicapping because in my later years I just don't have the energy or interest I once had.

[DanG]

For 2 yrs I charted the last 1/4s of all the races at PP. NOT the last 1/4 of the race but of the horse in each race that ran the last 1/4 the fastest. I found that there was a window of 2-4 seconds between the fastest and slowest, on average. I then made my variant based on this and it turned out to be the most accurate variant I had ever used, believe it or not.
Very interesting PM; thanks for sharing that. :ThmbUp:

When I made numbers at Calder for example I was amazed at what a high % of the total variant was only in the stretch.

[bobphilo]

Fair enough CJ, I'll take you at your word and go from there. I guess I'll have to keep on "grinding" with creating a database (God, I hate my inabilities in this area!). I've been working on and with computers since 1970 and database creation and usage still bewilders me and this disgusts me to no end.

My kingdom for a racing database template!

ps. this, in no way, means that I will ever buy or subscribe to a "service" that has one. I'm far too cheap for that, I'll eventually find one or somehow create my own. If I can't figure out how to get the information I want without spending even more money than I already do, I'll just do without.

Ray,


Rather than struggle with the DRF variant which is filled with problems, you can also “reverse engineer” the variant that Beyer assigned to the race. In the appendix of “Beyer On Speed” there are tables for the ratings that each time would earn at every distance as well as a beaten lengths adjustment table. Lets say a horse wins a 2-turn 9-furlong race in 1.49.00 and earns a Beyer figure of 113. Go to the time rating table for what the unadjusted raw time of 1:49.00 would earn in a 2-turn 9-furlong track. It happens to be 113, which means the variant would be zero for 2-turn routes that day. If instead, the horse had earned a 122 Beyer for the same time, this indicates the track was slow by 9 Beyer points because 9 points had been added to the horse’s raw unadjusted time (122 -113 = 9.) If instead, our 1:49.00 horse had earned a 104 Beyer, the track was fast by nine points and nine points had been subtracted from his unadjusted time (104 -113 = -9). The variant is simply the difference between the unadjusted rating for a given time and the variant adjusted rating.


Suppose you want to calculate the variant when you only have the Beyer for an also-ran – this is where the beaten lengths table comes in. Of course the time listed is only for the winner so to find the rating that the winner would have earned by running that time simply use the beaten lengths adjustment table to convert the also-ran’s Beyer figure into the winner’s Beyer. Then you calculate the variant as above by comparing the winner’s raw figure with his actual earned figure.


Note: The variant that you calculate this way will include not just the variant for that particular track that day, but also include the adjustment from track to track. You only have to do this calculation once for each 1-turn and each 2-turn race once for each day for a given track. Eventually you can build a ready to use database for each day for each track you play.


Hope this helps.


Bob

[cj]

Ray,


Rather than struggle with the DRF variant which is filled with problems, you can also “reverse engineer” the variant that Beyer assigned to the race.

You can certainly do that, but there is a lot of work involved if you want to know if it is a true variant, or one that is broken out from the other races on the day.

Also, a lot of tracks have variations from the Beyer charts in the book at one or more distances.

[HUSKER55]

I have been giving this topic some thought and I have come to the conclusion that what they call a varient would probably come closer to describing the strength of the fields that day. That has nothing to do wwith calculating speed.

First, it is an average of the class types, age, track conditions and race conditions for that day. Can anyone really rely on that number unless the card today is identical to the one from which the number was created from?

Second, as stated previously, throw in jockey tactics and allow for a lopside card (fast and slow races) and how would you ever be able to apply that number to a race with a resemblence of accuracy.

Third, some of them variants end up making two or more horses having broken the track record and we know that can't happen. Take a look at how long some of the track records have been in place. It happens in almost every race.

There are easier, more accurate ways around using the DRF track variant.

Just MHO

husker55

[raybo]

I have been giving this topic some thought and I have come to the conclusion that what they call a varient would probably come closer to describing the strength of the fields that day. That has nothing to do wwith calculating speed.

First, it is an average of the class types, age, track conditions and race conditions for that day. Can anyone really rely on that number unless the card today is identical to the one from which the number was created from?

Second, as stated previously, throw in jockey tactics and allow for a lopside card (fast and slow races) and how would you ever be able to apply that number to a race with a resemblence of accuracy.

Third, some of them variants end up making two or more horses having broken the track record and we know that can't happen. Take a look at how long some of the track records have been in place. It happens in almost every race.

There are easier, more accurate ways around using the DRF track variant.

Just MHO

husker55

So, does that mean that CJ's, HTR's, HSH's, etc. variants are just as inaccurate?

Can't wait to see their replys to your post.:jump:

[DanG]

I have been giving this topic some thought and I have come to the conclusion that what they call a varient would probably come closer to describing the strength of the fields that day. That has nothing to do wwith calculating speed.

First, it is an average of the class types, age, track conditions and race conditions for that day. Can anyone really rely on that number unless the card today is identical to the one from which the number was created from?

Second, as stated previously, throw in jockey tactics and allow for a lopside card (fast and slow races) and how would you ever be able to apply that number to a race with a resemblence of accuracy.

Third, some of them variants end up making two or more horses having broken the track record and we know that can't happen. Take a look at how long some of the track records have been in place. It happens in almost every race.

There are easier, more accurate ways around using the DRF track variant.

A legitimate opinion, but it does reinforce the point of why we all don’t end up on the same animal.

IMHO: The DRF “variant” only serves to give professional quality variants a bad name. I don’t know where to start with its flaws.

Class par research (by the good fig makers I’m aware of) has gone the way of the dinosaur. It was a good starting point in developing parallel speed charts and various baselines, but as far as configuring a daily adjustment to it…far too much noise.

BTW: Class pars in terms of a variant base / not to be confused with “Pars” within a given race. A VERY underrated part of the capping process imo.

The speed figure process has become so refined; it’s no longer a question of is it a viable method…it’s a question of methodology among purveyors.

1. Ground loss.
2. Weight adjustments.
3. Back propagating of data.
4. The projection method and how far back in the field to apply it.
5. When to break a race out due to changing track condition.
6. Does the rating adjust for fluctuations in pace / position.
7. Wind or not to wind?
8. Are multiple internal variants used.
9. Gut feel…vs. computer generation / or both.

So many variables and yet so many very good figures around.

I will say one thing; thank goodness there are good fig makers willing to do the work. :jump: As anyone can testify to who has tried it (if you haven’t btw, you should imo; even if it’s only for a brief period.) they are a lot of work and when you find numbers that work for you, they pay for themselves in time savings alone many times over.

Especially true if you bet a circuit that has multiple feeder tracks…aka Saratoga, Keeneland etc…Far easier for a serious player to make So Cal #’s as they are on a virtual island. (Except during SA’s debacle with track surface changes of course.)

[raybo]

BTW: Class pars in terms of a variant base / not to be confused with “Pars” within a given race. A VERY underrated part of the capping process imo.

Glad you pointed that out. This is what I am in the process of doing.

[DanG]

This is what I am in the process of doing.
Time well spent! It’s huge Ray and soooo underutilized imo.

It should be a fundamental early move in the process if you even remotely use a numeric approach.

Ironically ~ often its greatest value is when all runners are below par…that can open the door for Siegfried from ‘Kaos :jump: and that is a multi-race player’s best friend. One of the worst bets in a lightly raced field is the “least slow” among proven commodities / or should I say unproven.

[HUSKER55]

Good points Dan. Considering the way the DRF computes their variant I think using theirs would be a mistake. I think you would be ahead to buy them from someone who does.

You are also correct on race pars being under-rated.

Just MHO

husker55

[Cangamble]

To Pell Mell,
That is an interesting way to calculate variant. I can see it working, and to be honest, never thought doing something like that. I do my own track variants for around 5 tracks at a time, and I trust them more than I do Beyer or even TSN ratings, but I find that races over a mile, the ratings become way less accurate.
I just wonder how you extrapolated the final 1/4 mile for a mile and a sixteenth race.

[Bobzilla]

Ray,


Rather than struggle with the DRF variant which is filled with problems, you can also “reverse engineer” the variant that Beyer assigned to the race. In the appendix of “Beyer On Speed” there are tables for the ratings that each time would earn at every distance as well as a beaten lengths adjustment table. Lets say a horse wins a 2-turn 9-furlong race in 1.49.00 and earns a Beyer figure of 113. Go to the time rating table for what the unadjusted raw time of 1:49.00 would earn in a 2-turn 9-furlong track. It happens to be 113, which means the variant would be zero for 2-turn routes that day. If instead, the horse had earned a 122 Beyer for the same time, this indicates the track was slow by 9 Beyer points because 9 points had been added to the horse’s raw unadjusted time (122 -113 = 9.) If instead, our 1:49.00 horse had earned a 104 Beyer, the track was fast by nine points and nine points had been subtracted from his unadjusted time (104 -113 = -9). The variant is simply the difference between the unadjusted rating for a given time and the variant adjusted rating.


Suppose you want to calculate the variant when you only have the Beyer for an also-ran – this is where the beaten lengths table comes in. Of course the time listed is only for the winner so to find the rating that the winner would have earned by running that time simply use the beaten lengths adjustment table to convert the also-ran’s Beyer figure into the winner’s Beyer. Then you calculate the variant as above by comparing the winner’s raw figure with his actual earned figure.


Note: The variant that you calculate this way will include not just the variant for that particular track that day, but also include the adjustment from track to track. You only have to do this calculation once for each 1-turn and each 2-turn race once for each day for a given track. Eventually you can build a ready to use database for each day for each track you play.


Hope this helps.


Bob


Precisely my approach, Bob.

I do this daily as I attempt to reconcile my own variants with those of the Beyer people. When I feel like I can take a stand against Beyer's variants, hence his figures, a rare occasion indeed but it does happen, I feel as though I might have obtained an edge for a series of races soon to come. In recent years it's been my feeling that the Beyer figures for graded stakes races have been somewhat generous, perhaps to make them more believable to speed figure consumers who might expect large numbers from fields with reputable and established names.

While looking at today's 7th race at Aqueduct I couldn't help note the astonishing variant the Beyer boys must have been using for Saturday October 25th at Belmont. As my own attention was diverted that day towards the Breeders' Cup festivities on the west coast I hadn't really gauged the races in NY that day all too closely, so I was interested to see what the track was like. Take a look at AQUINO. I multiplied his beated lengths of 12 1/2 by 2.25 and arrive at 28, more or less. I add 28 to his BSF of 79 and arrive at 107, the number which is probably in the neighborhood of Kodiak Kowboy's winning figure. With a 107 I might expect a final time of 1:21: 3/5 or in that neighborhod, but 1:24: 4/5....... wow, that was a pretty slow track that day in New York. Correct me if I'm wrong but that's about 16 above par. I can definetly say that's an uncommonly slow track. All of a sudden that 22 4/5 opening quarter doesn't look too bad. It will be interesting to see if the now "legged up" AQUINO moves forward from that race.

[Pell Mell]

To Pell Mell,
That is an interesting way to calculate variant. I can see it working, and to be honest, never thought doing something like that. I do my own track variants for around 5 tracks at a time, and I trust them more than I do Beyer or even TSN ratings, but I find that races over a mile, the ratings become way less accurate.
I just wonder how you extrapolated the final 1/4 mile for a mile and a sixteenth race.

Haven't done it in years but I think I broke it down to 1/16s and smoothed it out.