Cramer speed numbers vs. Beyer Speed Rating

[William]

I don't know if I am the only one who hesitates to ask questions in this forum for fear of showing their ignorance, but I finally worked up enuff courage to ask another one. A few days I ago I bought a Daily Racing Form just because I was feeling guilty for not having bought one in over a year. I was comparing the Beyer Speed Ratings with the Cramer Speed Numbers in the latest version of TLC. While there seemed to be a fairly good correlation between the two in that the horse with the highest BSR in a particular race would also have the highest Cramer Speed number, this was not always the case. And in general, while there was what I would estimate to be a 90% correlation of the relative rating of the two speed numbers from the highest to the lowest. My question is does it even make sense to expect a correlation of the two? Which would the gurus in the forum here rate as the more reliable, Beyer or Cramer? (Incidentally, my usage of the word "guru" was not meant in a tongue in cheek manner at all as I have a lot of respect for the opinions and views shared in this discussion group).

[Donnie]

William...first, don't ever hesitate to post a question here. For some strange reason we all seem to want to help others on this board (one of the most assinine things I believe I have ever witnessed!! Imagine trying to make your "enemy" stronger; after all, are we not after each other's dollar??) You will always get a straight forward answer to a straight asked question.

I almost bought a racing form once about a year ago. Couldn't bring myself to do it. I used to read it religously. Not any more. I used to buy one just for the articles concerning the Derby, but for the last 2 years have not and still had the winner. Previously I couldn't pick a derby winner out of the form.

Your question---I personally have a problem with BSF because my understanding is they are highly subjective. If the person "working" the number feels it is too high, then he has the right to lower it to a more believable number. Suppose you had a horse who ran 132 BSF. IT ACTUALLY RAN A 132!!! This number would probably be lowered because no horse on this earth can run that fast! How often does this happen and by who's subjectivity are you making your betting decision? I personally look more at the pace figures with a quick look-see at the final number. But I do not base my final decisions on the speed figures.

Should they correlate to one another? I guess they could if their computations are close to being the same. But with a final "adjustment" on the BSF, that may be where you see discrepencies.

Good luck!!

[Glen]

William, no need to fear showing ignorance...I do it daily and still enjoy coming back for more...

You bring up an excellent point regarding Speed figs...If you compare Cramer, Ken's PER, DRF, DRF with and w/o variant, BRIS, Beyer, Trackmaster, Equibase, and ITS figs...you will find that they are all different...Crazy but true...Just like you said, most will be similiar but some of it will be real wacky...

More KM Stuff on speed figs:
http://www.homebased2.com/km/pdf/HTRMonthlyReport-JAN2000.pdf

[hurrikane]

Hey William...I am the king of stupid questions and these folks still let me hang around.
I use to make my own 'Beyer' figs before they were made public, and the fact is they are very subjective. That was one of the good things...my numbers would not be the same as the guy standing beside me. In fact if you are not making your own you lose a lot of the value of the numbers because you can reference notes and special situations to explain the wacky stuff...if you are using someone elses numbers you do not know how they were created.

I think in the over use of the numbers you have to look at the big picture anyway and not just one number. This is one of the great things about prat. Those strange anomolies stand right out there and you can tell right away if it is a number you can ignore.

That is the opinion of the king of stupid questions.

[tomcat]

Last year, there was supposed to be a big study on whose speed figures were the best. All the bigs were supposed to send in their speed figures every day for a year I think. Then a big publication was going to be printed. But, when the time came, nothing happened. Maybe the threat of lawsuits, incomplete information, I have no idea.

I was really looking forward to seeing the comparisons. In my judgement, Jim Cramer will win hands down. I have seen Jim work on these figs every day and he is very dedicated to his results.

I understand many of the other figs are just computer generated with no checking for accuracy.

[Bruce]

Jim Bayle at Sports Stats was supposed to do it, but he got out of the business and I believe is now doing selections and his own betting, but the Vegas guys would know better than me.

Who knows what's better? A study 5 years ago pointed to Trackmaster as having good figures.

There's a major problem in all of these studies. They track stuff like "Best last figure" or "Best figure in last 3," but you don't always handicap like that. The last figure could be useless if it was the wrong distance, surface, etc. Or best of last 3 could be useless if there were layoffs involved or the horse went into a form tailspin after posting a giant figure 3 back.

I strongly believe this kind of study has to be qualitatively not quantitatively. That is, look at longshot winners, say 6-1 and over and see how each of the figure services would have had the winner and give a one-line description like: 2nd best last figure, 4th best of last two, OR "last figure on slop, two back was competitive with last race of others"


bruce

[tomcat]

Thanks Bruce it was Bayle. I think he got scared of the study.

The issue, I think, is reliable speed figs. Many times weird, don't make sense speed figs will show up and no body says anything. That is because they don't check their numbers. That's why I give credit to Jim Cramer, he does his homework.

Remeber last year when the weird speed figs showed up in the 1st 2 weeks of Gulfstream? It was Cramer who blew the whistle, then everyone knew there was a problem, and Equibase finally get involved and found the problem.

[RonTiller]

For the record about the Bayle study:

Cramer (HDW), Ragozin (The Sheets), Jerry Brown (Thorograph), Beyer and TrackMaster all agreed to send Jim Bayle a comprehensive set of speed numbers for each horse at a dozen or so tracks prior to each day's racing. BRIS/TSN and ITS elected not to participate. Eveybody sent Bayle data for 6 straight months and everybody went to considerable trouble and effort to provide Bayle this data.

A few weeks after the test period was over, Bayle stopped answering his phone, did not respond to messages, did not answer repeated attempts to contact him concerning the status of the study. We all subsequently discovered that Bayle had gone full time into the longshot selection business and he chose to completely blow off finishing the study, leaving everybody involved hanging in the air and more than a little perturbed.

Seeing as how a study of this type would never be done again, Cramer contacted Len Ragozin, Jerry Brown and David Siegel (CEO of Trackmaster) (Beyer could not be reached) to see if he could garner support to have a third party do the study. After all, the data was already collected. The crucial issue was to get a CD burned with the data on it and to insure that the data could not be tampered with. Everybody agreed that they wanted the study completed, as there was not only bragging rights involved but also a lot of work on everybody's part.

Cramer finally got ahold of Bayle and expressed these wishes on behalf of the rest of the group (not including Beyer, whose position on this is unknown). Result: nothing nada zilch.
Cramer reluctantly gave up after a few more attempts and a few more unfollowed through promises by Bayle.

We are obviously disappointed that Bayle bailed out. This will probably never be attempted again.

The methodological issues are obviously complex but everybody agreed that the numbers would be tested with a series of studies, like % #1 last race winners, % #1 best of last 10, etc. Whether or not the numbers are slavishly used like this, all the numbers would have been tested objectively and the RELATIVE rankings among the different numbers would have clearly meant SOMETHING.

Concerning the subjective element, Cramer uses a complex projection method in calculating track variant and never adjusts horses or races to what he thinks they should be. If he is confident that some horses numbers are coming up too big or small, he does detective work to find out what part of the methodology is producing these seemingly errant numbers and then adjusts the methodology. This is particularly important with lightly raced horses.

Beyer and Cramer both scale approximately the same and in many cases have similar relative values. One thing Cramer does is review in a database the cases where the Beyer and Cramer numbers do differ significantly, in an effort to both understand the strengths and weaknesses of both their numbers.

Ron Tiller
HDW

[Mike Moore]

Ron,

I am confused about your reply to the "Cramer speed numbers vs. Beyer Speed Rating". You probably have much better information of the "Bayle Study" than me. But, I purchased this report from Bayle a number of years ago. To the best of my memory, Barry Meadows ratings were rated the best.
But it was published, advertised and sent to buyers.
Are you referring to a subsequent report?

Mike

[hurrikane]

Hey Ron, you obviously have some great insight into all of this. Thanks for the info and the methodolgies Cramer uses. I'm not at allfamiliar with his process. I have over the last 2 weeks been comparing the number side by side and although I find as many higher ranked Beyers winning as higher ranked Cramers the mutual on the Cramers is higher (I suppose because they are not so over used by the public). I also have found that the Cramers don't have the large swings the Beyers go through...esp in young horses..as you noted by the way Cramer evaluates his data.

Want to say I am really happy with the data, the cramers, the rankings and the 90 day number Ken has broken out in the new pps. Anyone else using all this with success?

[Glen]

Hurrikane, don't know if you know this but there is an export for the 90 Cramer on HTR4MSAC. And yes, it is very good.

[hurrikane]

is this the nSpeed stat and rank is rSpeed on the last column? Didn't realize it was the 90 day stat. guess it is time to go back and reread everything again and see what I have forgotten. Played with it a while back but left it alone for a while. Will have to look closer as a contender piece of the puzzle. thanks....so much data..so little time.

[Glen]

Yeop, nSpeed is the actual fig.
rSpeed is the rank of Speed.

[RonTiller]

Jim Bayle did a small study about 6 years ago that I believe focused on a small sample of Hollywood races in the summertime, as well as a few other major tracks. As I recall, it had the TrackMaster Power Rating, Barry Meadow's Rating, DRF 3yb Speed Rating, Ragozin's speed rating and maybe Beyer. I don't have the study so I may be missing a few.

Several problems:

1) He was comparing apples to oranges, by mixing in speed ratings (which are just numbers assigned to a horse's race, an evaluation of the final time of that race) and power ratings (which are numbers assigned to a horse as an evaluation of the horse in the upcoming race, using whatever factors the power rating creator chooses to use).

2) The sample size was tiny by statistical standards. When Jim is evaluating a change in some part of the methodology he uses for making speed ratings, he uses several independent 20,000 race samples.

3) The time span was late spring and summer. Winter races really seperate the different ratings out, when track variants vary wildly, allowing more room for different interpretations and thus different speed ratings. In other words, in the wintertime, there will be many more disagreements. Jim has found that one can create speed ratings for Hollywood summertime with no daily variant or track to track and have a number quite close to the "real" speed ratings.

4) There was only a small number of tracks involved. Speed ratings differ among themselves due to several methodological issues:

a: Daily track variant - some use pars; some, like Jim, use a complex multitiered projection method to determine what these horses in this race are capable of doing. With lightly raced horses, this becomes especially critical, due to the lack of history and the inherent improvement many make. Sometimes the answer to a projection is 'I don't know. Pars work better than nothing.'

b: Speed points per length - how many lengths does 1 speed rating point equal. Jim and Beyer both use an exponential formula. I believe Beyer had his pencil and paper and Jack Daniels when he came up with his formula. Jim had several computers, several years of data for every race run in north america. Amazingly, their answers were very very close. Some speed numbers use a constant number, regardless of distance, so 1 point = 1/2 length in 6f races and 12f races.

c: Scaling - This goes beyond the obvious difference between Ragozin/Brown scaling (lower numbers are better). The breadth of the scale (possible values are 0 to 120, more or less, for Beyer and Cramer) and the number of ties are also important. These issues are related. If half the horses in every race are tied for #1 last race speed rating, this obviously affects both the evaluation and use of these numbers. The smaller the range of normal values, the more ties one will have.

d: Track to Track adjustments - The classic question. Many use pars. Jim doesn't. I don't understand how he does it, but several years of researching various methods proved to him that pars adjust horses in the wrong direction more often than not.

By making Hollywood the centerpiece of his earlier study, Bayle was basically not testing the differing interpretations of track to track that different speed ratings exhibit. Turf Paradise shippers ocassionally, Northern California shippers more often, and that is basically it. To really evaluate these different speed ratings, one needs to look at tracks like Oaklawn, Gulfstream, Keenland and several of the eastern seabord tracks that have horses coming from all over the place.

I could go on, but the basic point is, the new study to end all studies would have corrected all these issues: 6 months of continuous data, Nov to May, tracks from all over the country, major and minor tracks, etc. Plus power ratings were to be evaluated seperately.

One final point in an already too long post. The earlier study tested several categories and provided win% and roi. Once it was published, the spin doctors went to work in a way that would have made Dick Morris and Bill Clinton proud (like the 2 person race between the Soviet Union and the USA, where the USA team wins; Pravda reports Soviet Union came in 2nd and USA came in next to last). I think somebody had a 2 cent positive roi, so they won in all the advertisements. Somebody had the highest win%, so they won. Barry Meadow, much as I admire him, won the spin contest, something like 'the only one to finish in the top 3 in all the various categories.' Everybody wins!

For the record, Jim broke all ties in the data sent to Bayle (Bayle was throwing out ties) and stands by #1 last race speed rating win and in the money % as, although not the best way to use these numbers, a completely objective measure of the RELATIVE value of his speed ratings. No wiggling or squirming here.

I gotta get some work done. Just thought the record should be set straight.

Ron Tiller

[Bruce]

Maybe if we all sit down with a bottle of Jack Daniels, we'll crack this code once and for all.

Anyway, I'll die trying.

[tomcat]

Ron, aren't you suspicious of Bayles actions? All that work that Jim and others did, then for him to walk away?

It really smells.

I think lawsuit or payoff backed him away. After, there is a lot of money in being able to prove who had the best speed numbers.

[Carl]

I think Bayle's "walkaway" from his commitment to complete a study was caused by him seeing a better opportunity. And he took it.

I see events unfolding as follows:

1) For years, people all over the country have been sending Bayle's spot plays for testing in his database. Why not? For $29.95 they thought they were buying a confirmation of their own test results, and it was a bargin.

2) Some of these plays tested plus ROI on Bayle's independent database, some didn't. In either case, Jim now has a record of the tested play in his files.

3) Jim IMO goes humm....."I have a whole bunch of winning spot plays in my db. This independent query business is starting to go south, everybody and their uncle has a db now and can test their own stuff. I wonder what business I should venture into?"

4) Jim decides to open a longshot service selling picks that are based on a series of spot plays .

5) By all accounts I have heard (Barry Meadow who monitored, ESB who whined) his spot plays are VERY GOOD and his longshot business is very successful.

6) Gee, I wonder where the spot plays that his picks are generated from came from? Hehe.

If I am right about the sequence of events above, this guy done makes the many lawyers we have all encountered look done honorable and his dropping of the study he committed to just a small piece of a dishonorable pie......

later,
Carl

[Glen]

Well i will say that what happend regarding the speed fig thing was a real bummer...

BMeadow tracked Bayle for a month and on something like 792 plays over the month he showed an ROI of 1.14...All computer generated plays...

[George]

Sounds like we are similar to lawyers, whereby 99% of db handicappers are giving the rest of us a bad reputation. :)

[Carl]

Interesting point George. And I don't think it really is as simple an issue as I made it appear in my first post.

Db handicappers may or may not own "intellectual property" I guess.

IF someone sent you a "spot play" for testing, and it worked, what would you do? I would think my natural inclination would be to
1) Use it in "real life" first then
2) Improve it, at which point I have probably crossed over the psychological line of "using your play" to "using a play I DEVELOPED".

So my point is, I guess, that we as developers of spot db plays are more vulnerable to being ripped up than most "regular" handicappers. Even if Bayles simply used all the information gathered to make up a series of "SUPER SPOT PLAYS" I personally feel he is/was in violation of trust. And yet, what is the man supposed to do? Die with the information?

And I can also see where he is coming from from some of my own personal experience with db plays. For instance, you sent me a spot play a year or so back for T-J >25%, Par >103. Well, I modified that play slightly over the next month or two and thought of it as "mine" almost immediately thereafter. Granted, you shared it freely fully knowing I might play it, and this is somewhat of a different situation. But where people's ways of being able to support themselves and their families are concerned, what looks like "sleeze" from the outside I am sure looks like "we will do what it takes" from the inside.

And no I did not/and would not have considered sending something I had that worked out for outside testing. Too much risk, to little potential gain IMO.

later,
Carl

[Bruce]

Is it possible when you submitted a spot play to Bayle that you had to sign something that gave him rights to it?


bruce

[hurrikane]

Your last line pretty much says it all Carl.

Because someone sends him a spot play does not give them intellectual right to the information. If he takes the info and runs it across his data and it comes up positive..he may have rights to intellectual information...as it appears in his data. If someone runs it across thier data it becomes thiers...across thier data. There is tremendous grey area in the intellectual field.

For instance. If you were to send me your database..for whatever reason. I could extract your data...put it into a different piece of software and then call it mine. You can do absolutely nothing about it unless you can prove that somewhere, somehow..you created all those little numbers in there...impossible.

I use to have a company that dealt in geographic data internationally..impossible to make the data
proprietary. That is why I charged up the a** for it...because I knew once they got it they would shop it around. Then they figured out how 200 phillipinoes working for 25 cent a day could make a cheap imitation..but that is anouther story for anouther day.

Truth is, this guy probably didn't break any laws..just not real ethical.

[Carl]

Don't know Bruce.

One of the guys at Netcapper's said a month or so back that he submitted a play to him for testing, he'd probably know. I can post there later and ask him if you like.

Jim did/does publish some sort of "study" on the profitability of some common spot plays, he was selling it on site. You don't get the specifics of the plays with that though, just the published result (which I guess would be useful if your were buying a "system", it would be nice to see an "objective" workout first.)

later,
Carl

[tbrown]

I understand that Equibase has prohibited people form selling partimes calulated from thier data. This will impact at least two parties who sell pars.
Now what I object to is that they are not selling the data-they are selling information gained by an analysis of the data. Nowhere in equibase data is their an average time.
Data is not information.
Now if they can tell you that you can't use their data to make decisons, then what happens when tell you that you can't read their PP's and make selections.
Tom

[MikeDee]

I'm guessing but I think that Equibase wants a cut probably would let someone sell information based on their data if they are getting a piece of the action.

[RonTiller]

Anybody who downloads data files from a company with an Equibase contract is required, as part of the terms of purchasing and downloading files, to abide by a usage agreement. The one on the HTR file download site is:

"You may use the information hereunder provided for personal purposes only. You may not re-sell this information or an analysis thereof, either directly or indirectly, nor may you re-distribute this information in any form without written permission from Equibase Company LLC."

The other data providers have similar usage agreements. BRIS own Multicaps because of a violation of BRIS's usage agreement. And it's not just reselling. Many of you are familiar with the free sheets the Stardust Race and Sports Book provides. They have a contract with Equibase to do this.

You can argue till you are blue in the face about who owns data, who has a right to control the economic destiny of the analysis of data somebody else owns, etc., but the fact is that these are the rules of the game and Equibase is in the driver's seat with respect to the interpretation and enforcement of these rules.

Companies like HDW, TSN and ITS have a huge overhead to meet in legally acquiring the rights to resell both Equibase data and an analysis of that data (like trainer stats, pedigree stats, speed ratings, pars, etc.). Whatever else you may think of the legalities or the rightness or the morality of the way data usage is controlled by Equibase, it is certainly NOT in the interests of HDW, TSN or ITS to sell data to a competitor (a person with a home computer making reports, files or pars and selling them on the internet) at a tiny fraction of the cost that they themselves pay to Equibase for this very right.

Shoot, I'm starting to sound like I'm preaching. Sorry.

Ron Tiller
HDW

[tbrown]

Ok, but the bottom line is, they are not offering pars. Like DRF, they have this huge database and do very little with it. If EB were selling pars, I would agree. I would even buy them. But since they aren't, I can't see how someone else doing so can possible take a bite out of thier profits. I gues I would be willing to pay (anyone listening out there?) $100 for a photograph of LasVegas at night, and if they wanted to send me a set of partimes as a gesture of goodwill, at no cost of course, I would be ever aprreciative.

[hurrikane]

I certainly wasn't making a statement on the morality of selling calcs done on thier data. Just I don't know how they can resonable regulate/prove it. Been there..done that.

[Butch]

hurrikane: Your mentioning an atlas company as far as intellectual property goes reminded me of something else. As you probably know, Rand McNally used to put fake cities and other fake topological information on their maps. They did this so that if other map companies or fly-by-nights copied Rand McNally's maps, it could be proven by the existance of the fake cities in the new maps.

A bad side effect of this practice was that people who didn't know better would end up planning their gas stops around these nonexistant cities and would end up running out of gas when the city that was on the map ended up being another stretch of farm land. So Rand McNally started putting their fake cities closer to real ones so this problem wouldn't occur.

One such fake city was North Welton Iowa. Welton is a small town of about 600 located about 25 miles north of Davenport on old Highway 61. Rand McNally inserted "North Welton" about a mile up the road on Highway 61 in its maps. A businessman in Welton posted a billboardish double-sided sign on the road in the area North Welton was supposed to exist at. It read, simply: "North Welton", followed underneath by: "The City that Rand McNally Created." An outhouse-like building stood on the other side of the road. Next to it was a sign which said: "North Welton Municipal Building" and underneath: "Hours of Operation: 10 am to 11 am, July 15th." And true to its word, somebody stood by the outhouse for the one hour of the year. Eventually, the state of Iowa made the man take the signs down.


Eventually, Rand McNally stopped the practice of putting fake cities on its maps.

Butch