Reposting--Friedman fires smoking gun--shoots foot

[TGJB]

Just wanted to make sure this string gets moved up so it doesn\'t get lost.

Okay, I\'m back. There\'s a lot to get to, and I\'ll try to get to all of it, but I want to use this thread for a very specific subject-- Friedman\'s post about the Chilukki race-- because it is of overriding concern in evaluating the figures made by both companies, and impacts several of the other subjects that came up.

Here\'s what Friedman said:

On the day in question (4/28/99) Chilukki won the first race at CD. The track was worked on several times, but specifically between the first and second races. The track was AM muddy, track listed as good for the first, fast for races 2-5, sloppy 6-9 when more rain hit. Friedman says \"accordingly\", they used 3 variants for the card. He then goes on to talk about subsequent events indicating the number they assigned turned out to be correct, which I\'ll get to later.

Now the key stuff, and you have to read this pretty carefully-- I had to read it twice and show it to someone else before I could really believe he said it. Friedman says they came up with the figure by going over the historical data about the relationship between 4 1/2 furlong races and other sprint distances at CD (he says Kee by mistake, but that doesn\'t matter). Now, I could have a field day with that alone, but right now I have other fish to fry-- actually herrings, red ones. In this case, THE RELATIONSHIP BETWEEN THE DISTANCES IS NOT THE ISSUE-- THE RELATIONSHIP BETWEEN THE TRACK SPEEDS IS. The two races could both be at 6 furlongs and it wouldn\'t make any difference.

So, how did they come up with a variant for a race that was not preceded by another race, run over a track that had a different moisture content than the following race, and that was worked on between races?

Possibilities:

1- They did it at the same variant as the second race. This is what I thought they did, and it would have been wrong to do it that way, but Friedman says they did not.

2- They cut the race loose, and did it based on the horses that ran in the race itself. This would be a real news story, a feat worthy of Karnak. There were exactly 2 fillies in the race who had run before, one start each over the downhill 4 1/2 course at Keenland (which Ragozin gave figures to, another future field day). Even if you think those numbers were accurate, they both had run in the mid thirties, and both were assigned jumps of at least 15 points in this race, so they obviously weren\'t used to construct the variant here. So if they did the figures based on the horses that ran in the race they did so using workouts and/or pedigree, which seems unlikely.

3- Which seems to leave only the following-- they used the variant for the second race, but adjusted it with a mechanical correction. And the obvious question is, how did they come up with the correction? Possibility 1 is that it is arbitrary. Possibility 2 is that they used \"careful attention to the OBJECTIVE condition, (and) the track history in SIMILAR situations\" (emphasis added).
Okay, what is the \"objective\" condition? When Friedman said that the listed track designation changed, and they \"accordingly\" changed the variant, is he serious? Is he actually claiming that a good track at CD always has the same variant, and always differs from the fast track variants (which of course are always the same) by an exact amount? And what are \"similar\" circumstances? All times where the track goes from good to fast have the same variant relationship? Really? How about the ones where they worked on the track between races, do those have the same relationship? Do you know all the times they worked on the track between races, at all tracks?
My guess-- at best, this is another example of the Ragozin use of broad averages. They may have taken a bunch of times that tracks went from good to fast (independent of work being done) and averaged them to come up with a rough correction. First of all, the only reason averages are used at all, ever, is because there is variability in the results you are testing-- if the average is a 4 point difference, some are 8, some are no change, etc.. That means the number you are assigning may be assumed to be AT BEST relatively close (in the above example, within 8 points), meaning only if the situation you are looking at actually fits exactly into the situations you measured with your average, and the results you measured were not too variable. And since they did work on the track, all that goes out the window, unless you did a seperate average for all the times they did work, which for starters means you have to know every time it happened.


Friedman\'s other point is that the subsequent figures the fillies ran vindicated the 4/28 figures because all the fillies (except Chilukki, of course) ran back to those numbers in the next 3 starts. First of all, as I\'ve pointed out before, you use earlier figures to make the later ones, so it is to some degree self-fulfilling. But more importantly, PLEASE-- these are 2 year olds, in April. OF COURSE they\'re going forward, rapidly-- if you were to believe Ragozin, the two making their second starts in this race both moved forward 15 points.

As for Awesome Humor winning-- it\'s meaningless. We don\'t leave boxes because the races come up fast-- we do it because there is not enough info to make figures with. It wouldn\'t have mattered if the Chilukki or Awsome Humor races had come up 2 seconds slower.

I urge everyone to read the above and Friedman\'s posts again, carefully. Please keep all comments on this string to the narrow confines of the figure making questions discussed here-- there are other strings for \"all figures are imperfect\", or \"Friedman wins so the figures must be good\".

[Alydar in California]

I think I\'ll be able to get to the rest of this late tonight. For now:

\"Friedman fires smoking gun--shoots foot\"

Whose foot? His own? Yours? Bad title, pal.

[TGJB]

I wasn\'t sure how many words would fit on one line, I was parodying a Daily News headline, and I never heard the expression \"he shot someone else in the foot\".

Yeah, there goes the rest of my free time this week. Of course, there are people who would say it\'s fine to just delete you, or simply ignore what you have to say.

If we\'re going to do this, let\'s try to make it about the big stuff that matters (actual issues), not  line-by-line trench warfare about angels-dancing-on-the-head-of-a-pin minutiae, okay?

[Alydar in California]

    That was my favorite. Was it the Daily News or the Post? How come Zuckerman gets all the great women? I\'ll stick to the big issues. TG vs. Rags methodology. Expect a little bit about The Clash, too.

[Alydar in California]

     A spectre is haunting Lower Manhattan--the spectre of Chilukki\'s 1999 debut figure. All the powers of the SHEETS office have entered into a holy alliance to exorcise this spectre: Ragozin and Friedman, Jake and Hardoon.

 Where is the figure maker who has not hurled the branding reproach of blowing Chilukki\'s debut figure, against the more advanced opposition figure maker?

(I\'m fully expecting Friedman to get this joke.)

OK, all kidding aside, Friedman has just taken a hammer and sickle to your trainer stats, and you want to talk about a 1999 MSW at Churchill Downs. Did Ragozin get the number wrong? If I had to bet, I would bet that he did. I don\'t think he should have made a figure for that race. Did you get Came Home\'s debut number wrong? I don\'t know. Please break out your data, as Friedman did, so that I can have a go at this question. The track was changing speed that day, right? This will be a great place to contrast your approach to Ragozin\'s.

You wrote: \" In this case, THE RELATIONSHIP BETWEEN THE DISTANCES IS NOT THE ISSUE-- THE RELATIONSHIP BETWEEN THE TRACK SPEEDS IS. The two races could both be at 6 furlongs and it wouldn\'t make any difference.\"

  I\'m not sure this is fair. Friedman was working off of a speed chart. These charts are often unreliable at distances shorter than 6F and longer than 1 1/8. Since, as seems plain, he linked (in some manner) the first race to later races, the accuracy of his chart comes into play. Therefore, it is perfectly logical for him to entertain the question of whether his chart was holding up. Please tell me what he did that merits an ALL CAPS rebuke from you.

\"As for Awesome Humor winning-- it\'s meaningless. We don\'t leave boxes because the races come up fast-- we do it because there is not enough info to make figures with. It wouldn\'t have mattered if the Chilukki or Awsome Humor races had come up 2 seconds slower.\"

I\'m not sure I follow your reasoning here. You criticized Friedman for making a number on the Chilukki race. He then points to the Awesome Humor race as an example where his willingness to make a number in a difficult situation produced a good bet. What is the problem? He introduced his Awesome Humor line in a way that made it clear that he wasn\'t presenting AH\'s win as evidence for the accuracy of his Chilukki figure.  


\"Friedman\'s other point is that the subsequent figures the fillies ran vindicated the 4/28 figures because all the fillies (except Chilukki, of course) ran back to those numbers in the next 3 starts. First of all, as I\'ve pointed out before, you use earlier figures to make the later ones, so it is to some degree self-fulfilling.\"

  To what degree would that be, given the manner in which Ragozin makes figures? Throw out the turf races. That leaves us with what, 70 horses Ragozin can use to make the variant when these horses run back? Unlike you, Ragozin doesn\'t split routes and sprints. This means he has a bigger sample with which to work. Did these horses all run back on the same day? Even if they did (and they didn\'t), the numbers aren\'t big enough to support what you\'re suggesting, unless you want to build your tower on your \"to some degree\" linguistic escape hatch. But there is a bigger issue here. Friedman knew damn well how tough it was to make a figure for this race. That\'s why he marked the day for later review. To think Friedman would then allow the shaky figures for this race to poison later races is to think Friedman is an idiot. Is that your position?

Now I want to ask you some specific questions about how you make these figures I respect so much and have defended so often:

1: To what degree do you use speed charts?

2: How often do you update them?

3: Imaginary day. Hasn\'t rained in a week. Three sprints this day, races one, two, and three. The evidence from race one suggests the track is three points fast. The evidence from race two suggests the track is 4.5 points fast. The evidence from race three suggests the track is three points fast. What is your variant for these three races?

Look, Friedman frustrates both of us by being vague when he discusses how he makes figures. For one thing, these historical corrections are impossible to counter to anyone\'s satisfaction. Here\'s something from his book. I believe it\'s important to understanding how he makes figures:

    \"Seriously underconsidered by Beyer and most other analysts is the LIKELY speed of the track today, based on weather--especially precipitation--and on the track superintendent\'s habitual day-to-day changes in grooming the track.\"

    In this sentence, I believe, lies the answer to more than one of these mysteries with which you and I, among others, like to entertain ourselves. Should we suspect skullduggery in Friedman\'s recalcitrance, in his general unwillingness to engage in detailed discussions about figure making? I don\'t think so. A couple of years ago, I asked you a question that would have required an answer of much specificity. Here is your reply: \"I don\'t want people going to school on my variants.\"

[Silver Charm]

Did I read a recent post by Freidman correctly??

Is he now saying that the numbers for the Chilukki race may have been off by TWO or THREE points??

[Alydar in California]

Silver Charm wrote: \"Did I read a recent post by Freidman correctly??\"

Not if he spelled his last name in it.

\"Is he now saying that the numbers for the Chilukki race may have been off by TWO or THREE points??\"

Please tell me you are doing self-parody.

[HP]

Silver,

If you are referring to this

\"I can\'t argue with you on that--if you add two points to the Chilukki race the subsequent development also looks ok.\"

I don\'t think he\'s saying anything about the number they gave being off two or three points. HP

[Silver Charm]

HP,

Thanks for the intelligent explanation.

Alydar,

Go back and study your Economics 201 about \'lagging indicators\'.

[Alydar in California]

    Silver Charm wrote: \"Go back and study your Economics 201 about \'lagging indicators\'.\"

Economics 201? You really should have gone to a school that was more, how should I put it, ambitious.

lagging indicator:
An economic indicator that changes after the overall economy has changed; examples include labor costs, business spending, the unemployment rate, the prime rate, outstanding bank loans, and inventory book value.

If you think the time of a race is a lagging indicator, you have a bit of a problem.

[Silver Charm]

Econ 201 was just one of the courses we were required to take before we went to grad school and before we became licensed in multiple states.

For someone does\'nt know how it goes I thought I might tell you.

[TGJB]

Alydar,

1- The reason I want to talk about the Chilukki race (\"Smoking Gun\" post)is because when it comes to figure making philosophies, hard cases make good examples. That race is the perfect illustration of how the Ragozin reliance on assumptions and averages is wrong. With all the wind and dust that has been stirred up, it is important to note that there still is ABSOLUTELY NO EXPLANATION for how they came up with the variant for that race.

Don\'t have the run-down for the Came Home day on my desk (it\'s in storage, we don\'t store the data in that form in the computer, are about to start), but if I gave a figure (as opposed to a box), there was more data to go on within the race, and/or more surrounding data that could be used without it being ridiculous to do so, than on the Chilukki day. Which is not to say I necessarily got the Came Home figure right-- as I believe I said at the time, it is always possible to blow a figure, since there is an element of judgement involved. Which in turn does not mean it\'s okay to make wild leaps based on assumptions or broad averages.
   Incidentally, my recollection is that Ragozin made that Came Home figure even slower than I did, but I could be wrong.

2- The question of the distance of the Chilukki race was a complete non-issue-- no-one raised it but Friedman. What you say about the reason he brought it up might have some merit if he had also explained how the variant was determined, but he did not-- HE HAS YET TO EXPLAIN HOW THE FIGURE WAS CREATED, DESPITE SPENDING AN ENTIRE POST PRETENDING TO DO SO. That\'s why it rates caps. If he linked the races \"in some manner\", as you say, let him explain why, and how. Ahem-- why not ask him?

3- If Awsome Humor had lost next time, would it have been an indication the figure was wrong? Did she pair up her figure on Ragozin second out? If 2 year olds that are given a slow number first time out jump forward to win second out, is that an indication the first number was wrong? Again, I am offering no opinion on whether either race was fast or not, just that there was not enough information to create an ACCURATE figure. You don\'t need figures to know that an open lengths maiden special win on a major circuit is a pretty good race. If Chilukki\'s first out number was 3 to 6 points slower first time out (or even faster, for that matter) would her second out win (and number on Ragozin) look \"wrong\"?

4- These days, if you throw out the turf races, you are left with about 50 horses. Whether or not the horses come back in the same race (which happens a lot with early season maidens because there aren\'t that nmany races), each horse becomes an information bit you use to make later figures. I said \"to some degree\" because Ragozin does make variants differently than I do, but I would add that I don\'t use early season 2 year old races to make variants. I also made clear that the \"self fulfilling\" aspect was a minor consideration in attacking his argument-- the major reason one can\'t draw conclusions based on 2 year olds \"getting back\" to their  numbers is because the horses are changing so quickly. I will be having another field day with this and related subjects later this week, when we exhaust this one.

\"To think Friedman would then allow the shaky figures for this race to poison later races is to think Friedman is an idiot\". Len has said he is going to be posting the 00 Wood horses. Hold that thought.

5- Speed charts lay out both the points per time interval for a particular distance, and comparitive par times on a \"standard\" (or average) day at the track in question. For the first purpose we use them all the time, for the second we use them as a starting point. If you set the pars in stone you can get in real trouble-- aside from the obvious errors Ragozin makes with sprint/routes, the 7f at Laurel and 1 1/8 at Keenland are two examples.

Depending on how you look at it, we either never update speed charts, or do so every day.

On your imaginary day, \"suggests\" is a key word. But if the evidence seemed pretty solid, I would definitely go 3/4.5/3. A point and a half is nothing, considering timing mechanisms, wind gusts, and track maintenance issues. 3/6/3 is the decision I get paid to make...

...that, and understanding what the correct overriding philosophy should be.

Your last part misses the point. There was no recalcitrance on Friedman\'s part about Chilukki-- he SET OUT to explain how they came up with the number. He only stopped when my post made it clear he was in way over his head.

I was in the Ragozin office for 9 years,  working there and as a client, and I had arguments with Ragozin about this stuff then. Paul worked there for 12 years after I left, so I know about the things they have done since then. (One of which is totally crazy and the real reason they did the Chilukki race the way they did. What I\'ve  been doing is trying to get Friedman to come out and say it, but he has figured out that he\'s dead if he does. It\'s an arbitrary correction). I understand how the Ragozin operation makes figures, and judging by Friedman\'s post, better than he does.

[Alydar in California]

Silver Charm wrote: \"Econ 201 was just one of the courses we were required to take before we went to grad school and before we became licensed in multiple states.\"

You need to ask yourself why you would write something so pathetic. These \"multiple states\" have even more explaining to do.

\"For someone does\'nt know how it goes I thought I might tell you.\"

I bow to your eloquence.

[Alydar in California]

JB,

I can\'t proceed until you clarify a few things:

You wrote: \"Depending on how you look at it, we either never update speed charts, or do so every day.\"

Are you saying you keep a running (constantly updated) average of the figures you give out at various class levels for each distance at each track? If not, please tell me what you are saying.

\" But if the evidence seemed pretty solid, I would definitely go 3/4.5/3.\"

Please define \"solid\" as best you can. If the second race had a seven-horse field, and three horses obviously Xed, and the winner obviously freaked, can the evidence be \"solid\"?

\"I understand how the Ragozin operation makes figures, and judging by Friedman\'s post, better than he does.\"

This sentence is so awkward that I want to know if it has been edited. It looks like one of those Blockbuster movies where they take out all the good parts.

[TGJB]

Alydar,

Speed charts and par levels are two different things. I described how we used the speed charts, which have zero times for each distance based on a standard configuration (and which are adjusted for the average relationships between distances at that track, and used BY US only as a starting point), as well as time-per-point relationships (as in, 2 seconds at 6f = 10 points).

Par levels are made by figuring out the average winning figure for each class of race. Using these only lets you make rough figures, and is only useful when you are first starting your data base-- it\'s another example of a broad average. Once you have a significant data base of individual horses to work with, you are much better off basing your figures on the horses that ran on a specific card than on a broad average of all horses who run in those types of races in general-- obviously, you could be dealing with a strong or weak 10 claimer, etc., and you are using those horses to determine how fast the other horses are.

We use \"mechanical\" variants, as Ragozin used to call them, only to make my job easier-- the first step is to apply a claimer derived variant to the day simply because it usually gets the day within a few points of where it should be, meaning the numbers I have to add and subtract in my head as I do the day are smaller. We do not rely on the par levels in any other way, and I would point out that tying your figures to mechanical variants \"anchors\" your figures-- if racehorses as a group are getting better or worse over a period of time you can\'t know it, making comparisons from one generation to another impossible. You have already determined that 10 claimers are the same forever, etc.

In the example you gave (which is pretty much the most common single way a race comes out), it definitely can be solid if the 2-3-4 horses run numbers very tight to their histories (meaning, pair up last number, run exactly their top, etc.). Think about the alternatives-- if you give the race faster, you are giving the winner an even freakier number, and the 2-3-4 horses faster than you want to give them. If you give the race slower, the winner will look better, but 6 other horses will look worse. This, by the way, is almost exactly the situation that came up when War Emblem ran his first big one last year-- we gave him the number rather than have a whole field of good 3 year olds behind him collapse, and Ragozin went the other way. War Emblem paired the number we gave him EXACTLY in the Illinois Derby.

The sentence about my understanding how Ragozin makes figures better than Friedman does was edited, but I think I got my point across. Speaking of which, you neglected to ask me the most obvious question raised by the post you are responding to.