How Do You Determine Class?

[rubicon55]

IMHO, maybe I am rehashing what has been said since the beginning of time but I ask myself the question who has the horse in question beat and who has beaten him/her and under what conditions and for what kind of purse value. I discount dream trips and heavy biases and look for what I think are honest efforts and give bonuses for those who have overcome the pace, traffic and had to race wide. I imagine there could be a good discussion if lengths mattered at the finish in the final analysis. I cannot seem to find a method that works best for me that would assign on paper or a computer a number or a value that would be called class value. A single number to me is too static for me in a race dynamic. Too many variables confuses regarding the definition of class is not useful to me. I find it more important in observing how a horse rates during a race IMO which can be a better indicator of ability at different purse levels, this may hold the keys if that horse can run competitively at that purse level or not. We all can tell by the first or second call who rating and who does not. I pay more attention where my runner is at the first call and how the race unfolds between the first and second call (sound familiar?) If the horses rates and does not run rank or whither under pressure then I consider that is the class level of that horse for this race until today's race is run in which he may rise or fall per my definition of class rating. Seen too many trainers putting horses at the wrong purse level. Obviously the horse in question need to finish well, say at least mid pack at the finish unless it is a blanket finish just to make sure he is not a quitter type. It is sort of a hybrid of trip handicapping supported by pace, speed and past performance (form cycle) under similar purse conditions. I start with my contenders and then start reviewing. Takes a lot of time to review videos and not convenient like a computer but not sure how else I can separate the chaff from the wheat. As other posters have stated the charts are only as good as the one making them. This method of course limits the races I can cap due to the review time so I have to be selective - in that case like most folks I go to the tracks, surfaces, distances and purses that I do better at. Very difficult to define class IMO.

[traynor]

Originally Posted by rubicon55

IMHO, maybe I am rehashing what has been said since the beginning of time but I ask myself the question who has the horse in question beat and who has beaten him/her and under what conditions and for what kind of purse value. I discount dream trips and heavy biases and look for what I think are honest efforts and give bonuses for those who have overcome the pace, traffic and had to race wide. I imagine there could be a good discussion if lengths mattered at the finish in the final analysis. I cannot seem to find a method that works best for me that would assign on paper or a computer a number or a value that would be called class value. A single number to me is too static for me in a race dynamic. Too many variables confuses regarding the definition of class is not useful to me. I find it more important in observing how a horse rates during a race IMO which can be a better indicator of ability at different purse levels, this may hold the keys if that horse can run competitively at that purse level or not. We all can tell by the first or second call who rating and who does not. I pay more attention where my runner is at the first call and how the race unfolds between the first and second call (sound familiar?) If the horses rates and does not run rank or whither under pressure then I consider that is the class level of that horse for this race until today's race is run in which he may rise or fall per my definition of class rating. Seen too many trainers putting horses at the wrong purse level. Obviously the horse in question need to finish well, say at least mid pack at the finish unless it is a blanket finish just to make sure he is not a quitter type. It is sort of a hybrid of trip handicapping supported by pace, speed and past performance (form cycle) under similar purse conditions. I start with my contenders and then start reviewing. Takes a lot of time to review videos and not convenient like a computer but not sure how else I can separate the chaff from the wheat. As other posters have stated the charts are only as good as the one making them. This method of course limits the races I can cap due to the review time so I have to be selective - in that case like most folks I go to the tracks, surfaces, distances and purses that I do better at. Very difficult to define class IMO.

What you are doing is a lot like what I am doing. I think one of the important factors--as you point out above--is that there are often multiple"real" class levels within a conventional class level (often at identical or similar purses) that can be separated by race conditions. and by the composition of the horses competing in the individual races.

In essence, it is more useful to compare a horse against the other horses it was racing against, rather than using artificial categories like purse or claiming price. At some tracks there are multiple grades within a given class that create a hierarchy of horses that cycle through according to their current fitness and condition. Viewed superficially, or with simplistic categorizations, it seems the horse are all "running at the same class level." That is a long way from being the case in the real world.

[Capper Al]

Next couple of days Arlington is having thunderstorms. I avoid off tracks if I can. All my stats are based on fast and firm tracks.

As far as formulas for class go, there are 3 basic methods that come to mind:

• Earnings per start (ESP)
• Average purchase value (APV)
• Who beat who on a per race basis

Many books talk about these and provide stats for the first two. I am sure if most of us opened our code we would find that we are doing similar things.

[illinoisbred]

Al..the poly is always labeled fast. It does not become "off" when wet. If anything,if substantial rain hits,it will play faster. Seeya Tomorrow!

[Capper Al]

Originally Posted by illinoisbred

Al..the poly is always labeled fast. It does not become "off" when wet. If anything,if substantial rain hits,it will play faster. Seeya Tomorrow!

It's better than dirt. The only big thing to watch for then, when it comes to rain, is the turf to poly. Looking forward to seeing you.

[pandy]

The toughest thing is evaluating how good the fields were that the horse raced in. It requires not only keeping tabs on key races but knowing what types of races the winners from those key races won. If you have a computer system that can rate class in this manner you should get a good class rating.

Speed ratings are often used as part of a formula to create class ratings but are not good for this purpose for many reasons.

[traynor]

Originally Posted by Capper Al

Next couple of days Arlington is having thunderstorms. I avoid off tracks if I can. All my stats are based on fast and firm tracks.

As far as formulas for class go, there are 3 basic methods that come to mind:

• Earnings per start (ESP)
• Average purchase value (APV)
• Who beat who on a per race basis

Many books talk about these and provide stats for the first two. I am sure if most of us opened our code we would find that we are doing similar things.

One of the most useful class ratings using most of the factors you mention above was a compound: APV is divided by today's purse to get a percentile relationship--like 0.95 or 1.12 or whatever. That value is then combined with win% and/or win-place percentage to establish the basic figure. The formula I learned years ago included EPS, but in "layers"--meaning the resulting "rating" could not directly compare $2500 claimers to allowance runners, but was extremely accurate within either (or any other) type of race. To adjust EPS by class level takes a lot of record keeping, to establish an average for EPS at that class level at that track. That average value is set to "1.0" and the actual EPS are divided by the average EPS to get another value similar to APV figure (0.95 or 1.12 or whatever).

The EPS values are important--unless they are grade/track specific, they will tweak the resulting rating all over the place.

[traynor]

Originally Posted by pandy

The toughest thing is evaluating how good the fields were that the horse raced in. It requires not only keeping tabs on key races but knowing what types of races the winners from those key races won. If you have a computer system that can rate class in this manner you should get a good class rating.

Speed ratings are often used as part of a formula to create class ratings but are not good for this purpose for many reasons.

I might add that it is also important to evaluate the races other than key races, for the same purpose. That is, to establish some means of categorizing each race based on the relative strength or weakness of the entries in that race. There is a lot of work involved (at least initially) but the results can be impressive.

Most bettors look at a set group of values for comparing performances. In general, a $7500 claimer is considered a $7500 claimer, and that is the end of it. Exactly as in the key race example you mention, the field in one $7500 claimer may be tough as nails, and in another--same class level, same purse, same track--be a cake walk. Knowing which are key races is important. Just as important is knowing which are "soft" or "average" races, for subsequent evaluations of the entries that competed against each other in those races.

Again, it takes a LOT of work, but the efforts can be well rewarded.

[Capper Al]

Originally Posted by traynor

One of the most useful class ratings using most of the factors you mention above was a compound: APV is divided by today's purse to get a percentile relationship--like 0.95 or 1.12 or whatever. That value is then combined with win% and/or win-place percentage to establsihd the basic figure. The formula I learned years ago included EPS, but in "layers"--meaning the resulting "rating" could not directly compare $2500 claimers to allowance runners, but was extremely accurate within either (or any other) type of race. To adjust EPS by class level takes a lot of record keeping, to establish an average for EPS at that class level at that track. That average value is set to "1.0" and the actual EPS are divided by the average EPS to get another value similar to APV figure (0.95 or 1.12 or whatever).

The EPS values are important--unless they are grade/track specific, they will tweak the resulting rating all over the place.

So you know a little something about class.

[Capper Al]

Originally Posted by pandy

The toughest thing is evaluating how good the fields were that the horse raced in. It requires not only keeping tabs on key races but knowing what types of races the winners from those key races won. If you have a computer system that can rate class in this manner you should get a good class rating.

Speed ratings are often used as part of a formula to create class ratings but are not good for this purpose for many reasons.

First and foremost about making class figs is not to use speed in anyway in the calculations. What is being attempted by a class fig is a little independance from speed figures. Even with no speed figures involved, they'll still correlate to each other.

Focusing on winners and key races gets into who beat who. This is hard to follow and make sense of.

[traynor]

Originally Posted by Capper Al

So you know a little something about class.

The formula above was presented by Michael Pizzola at a Sartin seminar in New Orleans in the 1980s, in conjunction with a money management system used by a group of professional bettors in New York. According to Pizzola, they had developed their class rating to a point that they virtually ignored pace, speed, and all the rest--comparisons were made on the basis of their class ratings, with wagers made on the basis of "gaps" ("differences") between the ratings.

A group of professional bettors from Argentina hired me (way back) to code an app that provided a structured approach to including trip notes and body language observations in a "conventional" (speed/pace) application. They used a combination of factors to establish individual race ratings, that were in turn used somewhat like Beyer's projected speed figures to create an individual "race rating" that could be used to compare past performances from those races in subsequent races.

Both approaches were rewarding--at least at the time and for some years after. Much more sophisticated approaches are used now by serious bettors.

[rubicon55]

Originally Posted by traynor

What you are doing is a lot like what I am doing. I think one of the important factors--as you point out above--is that there are often multiple"real" class levels within a conventional class level (often at identical or similar purses) that can be separated by race conditions. and by the composition of the horses competing in the individual races.

In essence, it is more useful to compare a horse against the other horses it was racing against, rather than using artificial categories like purse or claiming price. At some tracks there are multiple grades within a given class that create a hierarchy of horses that cycle through according to their current fitness and condition. Viewed superficially, or with simplistic categorizations, it seems the horse are all "running at the same class level." That is a long way from being the case in the real world.

On this note, I saw somewhere a chart that "normalized" purses between tracks. The A Class tracks I think was the baseline and the mid class and minor tracks were normalized to balance out the actual purse values so each track could be compared. Do you recall seeing such a table? I am not sure if it was Cynthia Publishing, HANA or what. The tracks were given a numerical value and not based on purse value. This would be especially handy comparing invaders with questionable purse earnings. Can anybody comment?

[traynor]

Originally Posted by Capper Al

First and foremost about making class figs is not to use speed in anyway in the calculations. What is being attempted by a class fig is a little independance from speed figures. Even with no speed figures involved, they'll still correlate to each other.

Focusing on winners and key races gets into who beat who. This is hard to follow and make sense of.

I could not agree more. The advantage is that discrete (separate) figures can then be "meaningfully combined" in ways that are impossible with "mixed" values. That incorporates the notion of "confounding variables" that is thoroughly muddled with combined (speed and class, pace and class) figures. Specifically, how predictive is A when B is => than X, and how predictive is A when B is <= Y--and, most interesting of all, how predictive is A when B is not factored in at all?

I think it is possible to derive much more predictive (and much more profitable) values by keeping the factors of pace, speed, class, current form, etc. completely separate.

[traynor]

Originally Posted by rubicon55

On this note, I saw somewhere a chart that "normalized" purses between tracks. The A Class tracks I think was the baseline and the mid class and minor tracks were normalized to balance out the actual purse values so each track could be compared. Do you recall seeing such a table? I am not sure if it was Cynthia Publishing, HANA or what. The tracks were given a numerical value and not based on purse value. This would be especially handy comparing invaders with questionable purse earnings. Can anybody comment?

I have seen a number of such charts, (including the DRF), but all seemed based on comparisons of "average purse values" by track, with the implication that track-to-track class "adjustments" can be made using such a chart. I stopped trying to use them years ago, and decided it would be better (for me) to develop my own approach to individual race class rather than track class. That seemed (seems) to work better than track comparisons.

For example, on the southwest circuit, bottom level claimers at one track are "better" than allowance competitors at another track, simply because of the competition in the respective races. Purses and earnings tend to present a very different (and much less accurate) view of that competition.

[traynor]

Originally Posted by Capper Al

First and foremost about making class figs is not to use speed in anyway in the calculations. What is being attempted by a class fig is a little independance from speed figures. Even with no speed figures involved, they'll still correlate to each other.

Focusing on winners and key races gets into who beat who. This is hard to follow and make sense of.

Not as hard as you might think. It presents a problem if you apply it to individual races going forward. It is much less difficult if you handicap races every day (whether you bet on them or not) and generate class ratings for all the entries, that can in turn be used to create a "race class" rating for each of the races.

It takes some work, but the values generated are worthwhile. I sometimes forget that people still handicap on the basis of "fresh" data. The app(s) I use create files of "handicapped races" that are used to handicap subsequent races. For example, I ignore any race with less than three contenders (according to very specific criteria) as "noise" that does little more than introduce nonsense into otherwise useful data.

The reason--a race with one or two standouts is overbet on the front end. It is also unlikely that a "lesser" entry in that race--that would otherwise have a good shot at winning or placing--will do more than use the race as an extended workout, saving its run for a softer spot. There are dozens of similar scenarios that make uncritical acceptance of "all past races as equally significant" very hard on the bottom line. Class ratings--and, by extension--class rating of races based on those class ratings--could go a long way to improving that situation.