Repost : Friedman / Changing Track Speeds

[TGJB]

Okay. While I was away Classhandicapper put some questions to Friedman on the Ragozin board, and Len responded. There is quite a bit here that demonstrates the different figure making philosophies of Ragozin and Thoro-Graph, so this is a serious post, and I\'ll avoid sarcasm to the degree possible. Len said:

1-- \"Saying something is \'clear\' or \'obvious\' doesn\'t make it so. All it does is say we should believe your assertions because to be frank you know that you\'re right\".


So right away the anti-sarcasm resolve is tested, since that is exactly what Len does.

Specifically-- Len, about six weeks ago I posted that it appeared you had made a 3 length error in your figures for most of the field in this year\'s Kentucky Derby, and showed in great detail why I thought so, and where you made the error. Your response: no, we got it right, with no detail. And no amount of follow up questioning has been enough to get you to explain what should be pretty straightforward, if you are right.

So I will give you another chance to set the record straight: please tell us the beaten length margins you used between each of the first five finishers in this year\'s Derby, and the ground loss you used for them. If the lengths beaten you used were different than the published ones, please tell us why. This is a simple and straightforward question.



2-- \"Picking on Belmont day to support your general conclusions is terribly misleading. There was a significant (and unusual) change in the variant level from the previous day due to work on the track in preparation for a Belmont Stakes day where there was a forecast of possible heavy rain. There was over an hour between both the 1st and 2nd and the 2nd and 3rd dirt races which is also an unusual situation\".



a) It\'s actually not an unusual situation. It resulted in there being about 75 minutes between dirt races, and every time a track cards two turf races in a row the time between dirt races is at least that long. In a recent 7 racing day stretch at Belmont it happened 4 times-- 6/12, 6/16, 6/18, and 6/20.

But it\'s also not significant. Every single racing day there is at least a 75 minute gap between two dirt races, whether there are other dirt races run between them or not, and whether track maintenance is done between the races or not (which, depending on the track, can be just as significant). The question is simply whether a track can change speed over a 75 minute stretch, or not.

b) Okay. You say the variant changed from the previous day. How do you know that? By knowing what track work was done, and applying a standard trackwork-done correction? Or by looking at how fast the horses ran, on a race by race basis? If the former, how did you arrive at it? If the latter, aren\'t you doing what you always accuse me of, just giving the horses what you want? More on that later.



3-- \"This combination of factors led to a slide in the variant level during the early part of the Belmont day card, although the amount of the slide was considerably less than what some others have apparently concluded. Slides like this happen from time to time, but they are the exception, not the rule\".



a) How do you know that was why the track changed speed? You were there when I made the original presentation on changing track speeds in Vegas (which can now be found in the Archives section of this site), so you know that scientists (like Dr. Pratt of M.I.T), who have studied the physical properties of racetracks, have found that slight changes in moisture content, as well as ordinary garden variety day to day track maintenance, affect resiliency (track speed). On Belmont day, they watered the track after the first 6 races, and not again. Think that might have caused the track to get faster during the early part of the card, then level off? And it\'s not unusual-- this kind of thing happens all the time, at tracks around the country, although not always to this degree. The fact they expected rain probably did exacerbate the situation in this case. But the reason you say a slide is unusual is because you make the ASSUMPTION that it takes a major event for it to happen, and don\'t ALLOW your variants to slide without one-- which the science has shown is wrong.

b) How do you know the track speed changed during the day, and by how much? Once again, if you look at the day race by race, aren\'t you just giving the horses what you think they should run?



4-- \"Changes in the relationship of the routes and sprints at a given track happen from time to time, but again they are the exception rather than the rule except where the variant level and/or the track condition in some cases makes the short/long the usual result (sic)\".



I would probably disagree with this if I could figure out what it meant. I will say that the relationships between distances are not fixed, due to being affected by track maintenance and climactic conditions. At Belmont there are very significant wind current and sunshine issues that depending on wind direction and time of year (where the sun/shade is) can cause the backstretch to be soaking wet while the stretch is bone dry. The percentage of the race being run over each part varies from distance to distance.



5-- \"Making these decisions is sometimes very difficult, as you say, and that\'s why the hard work of insuring accurate and objective raw data and careful review of all aspects of the track\'s history is a better approach than bouncing the variant around to give the horses what the variant maker \'knew\' they must have run\".



a) The simple, basic question is this-- HOW do you make those decisions, if not by using the figure histories of the horses? In that sense everyone who makes speed figures, whether by using par times or by the projection method using entire fields, is giving horses what they \"knew\" they ran. The only other way would be to have a device that physically measures surface resilience.

b) But more importantly-- you CAN\'T just give the horses what you want to. I was hoping Friedman would go down this road in Vegas, but he didn\'t.

The figure relationships between horses in a race are fixed, by the lengths between them, the weight relationships between them, and the ground loss relationships. You can give exactly ONE horse per race \"whatever you want to\". You can\'t pair the winner to his 6, the second horse to his 8 1/4, and the third horse to his 9 1/2, unless those relationships bear it out. If you start screwing with the relationships of horses within a race, or start pairing individual horses at the expense of giving correct numbers to groups of horses, you will get yourself in a lot of trouble real quick-- you will have a data base where nothing makes sense, the opposite of the Thoro-Graph data base, where lots of horses run in a tight range.

I think this is the most common misunderstanding among those who have not made figures

[TGJB]

We are still waiting for a response from Len to my direct question about the beaten lengths in the Derby, which is probably not a surprise to anybody, including those that use his data. It is also kind of sad, if you think about it-- this is the Kentucky Derby we are talking about. If I was using his data I would have enough confidence in him to demand an answer publicly, or I wouldn\'t use it.

But while I was waiting, I had a couple of further thoughts pertaining to Len\'s post.

1-- In your post you say that the change in track speed during the course of the Belmont day card was due to work being done on the track. Now, let\'s for a moment accept your hypotheses that that is why the track changed, and that without something like that happening the track does not change. This begs certain questions:

Do you get complete information from all tracks as to what work is done, and when? In the extremely unlikely event you do (and it is accurate) for all the major tracks (including the ones that your trackman do off TV at simulcast sites or off satellite dishes at home), do you get that info for the smaller tracks where you don\'t have trackmen? Which tracks do you not get the info for?

Because-- if there are ANY tracks you do not get the info for, how do you handle the variants? Do you simply assume no work is done on the track, and therefore wrongly assume it stays the same speed on days when there really is work done, and you don\'t know about it? Or do you not make that assumption, and determine what the track does (gasp!) by looking at how fast the horses ran compared to their figure histories-- in other words, \"giving them what you want to give them\", as you accuse me of doing? Hmmm...

What brought this to mind was the day I just did, 6/25 Belmont, track listed as fast all day, with rain beginning only before the very last race. Did you have the variant changing? When? Do you know what was done to the track, and when? I do-- but even if I didn\'t, I would know how to handle the day, simply by looking at how the horses ran.