Underlying Data

[miff]

Jerry,

I understand your comments.As far as the BC,I already said that what you did, with your BC figs, was in line with what I saw on the track and what I generally expected you would give.I have not read too many comments here which stated anything contrary to what you did on BC day.

Your dirt figs matched Rags pretty much.I mentioned that PH ran a \"pure\" huge fig (no ground loss). That\'s the one fig I had a  minor issue with and you explained why the fig was given.

Mike

[TGJB]

Bit-- to review, you are suggesting that we do a study of variant relationships and watering of the track, ignoring all other variables. Meaning, in the situations studied the variables would, well, vary.

1-- Just for starters, let\'s assume that we did it and it actually gave a result that showed the curve you are talking about.

a) The curve would be based on, in effect, an average of effects of the other variables, which could under individual circumstances have differing effects on the variant. It would therefore be useless in assessing the variant for a SPECIFIC situation where the other variables might be different than the average. Which as a practical matter makes it useless as a tool.

I\'ve raised this point as well here in a different context. In his book Ragozin talks about looking at the history of the effects of track maintenance BETWEEN days in deciding track speed. As a practical matter this is useless-- even if you knew that on AVERAGE the track runs 5 points faster Tuesday than Monday, you are not simply going to do a variant for Monday, and then add 5 for Tuesday. First of all, it\'s an average, and so of no use in specific situations, and second, why not look at the day itself, as you did for Monday?

I suspect what that comment in the book meant-- and it is idiotic on the face of it-- is something much worse, that Ragzoin starts from the assumption that different days-- not just individual races-- are run over a track of the same speed, unless he specifically knows of something that could change it. Like a history of getting faster on Tuesday (and don\'t even go down the road of how he could know that without looking at the figures for the horses that have run on past Tuesdays). This is a terrible assumption to make, especially if you don\'t have DIRECT information about something being done to the track, as opposed to a historical average.


b) One side of your study, the one involving my variants, is based on judgement-- mine. Worse yet, it too is based on an average, of my judgment.

If you assume my judgment is correct-- doing things the way I\'m doing them now-- why do the study? Why not just continue to use my judgment going forward?

Worse yet, if you don\'t think my judgment is correct, how can you use it in the study?


2-- Lets say the idea behind the study is not to use going forward, but to see whether various things done to the track cause it to change \"speed\"-- which is what this entire conversation is really about, whether my approach or Ragozin\'s is correct. Whether it is correct to ASSUME that the track stays the same speed, or changes without severe weather.

If so, there are better ways to test. And some of those tests have already been done.

Rather than send people back to the Expo presentation again, I\'m going to pull out a few things.

a) Former head NYRA track superintendent (now assistant) Jerry Porcelli has tested NYRA tracks for moisture content, and found that they range from 4 to 12%.

b) There was a study done by a group of physicists, with the catchy title of \"Interrelationships Between Moisture Content Of The Track, Dynamic Properties Of The Track, And The Locomotor Forces Exerted By Galloping Horses\". You can find details in the Expo presentation, and the whole study somewhere on the web-- that\'s where I got it, thanks to Mall of this site.

The scientists studied 5 different tracks, using various machines, including a \"Drop Hammer\", which measures energy return (which in our terms is what determines track \"speed\"). The study said:

\"The dynamic properties of the track surface vary with its moisture content, composition and compaction\". And, \"the water content of the track cushion may fluctuate widely\". And, \"Several studies have shown that the composition of the track surface alters the dynamic responses of the soil\".

And specific to what we are talking about, keeping in mind Porcelli\'s 4-12% figures, they found:

\"Changes in moisture content of the track cushion resulted in similar changes in both the percentage of energy returned and the impact resistance of the track. Energy return and impact resistance DECREASED at 8% moisture and progressively INCREASED from 8.5% to 14% moisture\" (emphasis added).

What that means is, if there is 7% moisture in the track (in the middle of Porcelli\'s range), and you add water, it will get slower. But if it is 9% (still in the middle of the \"normal\" range) and you add water, it will get faster. At least the specific tracks (5 of them) that they studied will.

Also, Dr. Pratt of M.I.T., who has studied more racetracks than anyone who ever lived, said in an e-mail to me:

\"Things can deteriorate quickly above 9% depending on the soil composition and fall apart at 6%, again depending on the soil composition. THE WATER TRUCK CAN CHANGE THE % OF MOISTURE CONTENT BY ABOUT 1/2 OF A PERCENT FOR EACH PASS\".

So we know that small variations in moisture caused MEASURABLE differences in energy return (track speed), but there is no cut and dried relationship between ADDING water and tracks getting faster or slower.


3-- Having said all that, there are some interesting REAL studies that can be done, although they won\'t be usable for specific variant making going forward, any more than the studies by the scientists were.

As I said yesterday, there are other studies being done currently on racetracks, and I tried to get involved with them. If they are doing them at tracks where racing is going on, and taking readings between races (moisture content, impact resistance, and of course energy return), I would love to do a blind study and find out if there is direct correlation between track speed (as judged by me) and any specific variables the scientists measure, and/or the results they get when they measure them. In fact, I think I\'ll send Mick Peterson another e-mail.

Which reminds me-- Dr. Peterson, who has probably done studies on more racetracks than anyone but Dr. Pratt, is the one who said about the question of judging track \"speed\":

\"I suspect that your information regarding the performance of horses, even given the other uncontrolled variables, may be the best data available\".

I would like to think he meant specifically Thoro-Graph-- but he most certainly meant the idea of using the past histories of the horses as a guide. And he would laugh at the idea of assuming the track stays the same speed all day.

Once again, I would suggest that anyone who has not already done so check out the Expo presentation in our archives section.

 




 

[bobphilo]

TGJB Wrote:
-------------------------------------------------------
> > b) One side of your study, the one involving my
> variants, is based on judgement-- mine. Worse yet,
> it too is based on an average, of my judgment.
>
> If you assume my judgment is correct-- doing
> things the way I\'m doing them now-- why do the
> study? Why not just continue to use my judgment
> going forward?
>
> Worse yet, if you don\'t think my judgment is
> correct, how can you use it in the study?

Good point, Jerry. There are times when the time and effort to conduct a well-controlled study are justified and times when they are not. The best way to, not only know whether the track conditions have changed enough to break a race loose, but how to quantify it is to look at how much the horses performances have deviated from what would be expected given their past figures relative to the other races for the day. The fact that track maintenance (such as watering) can change the variant has already been well established and just provides additional evidence that breaking the race loose is justified. To do a study to determine the exact extent to which every gallon of water influences this is not necessary and a waste of time. The figures the horses ran do that. Everybody here knows that I have no problem with getting into the excruciating details of research methodology when my training tells me when this is needed to gain relevant information, but I think some of you guys are just working too hard needlessly.

Bob

[marcus]

That sounds pretty good , your interpetation of what happend and subsequently your judgment when assigning values to the many variables for variants etc is the best idea and can be trusted - potential survey averages which measure factors out of context can not be  . Incidently , my ex-wife has worked at MIT for over 20 yrs and she probably could find \"people\" over there who might be interested in getting involved in studies you\'d like to see happen  and I\'ve heard she\'s still talking to me  .  So let me know , I\'ll be glad to check that out ( again ) at some point , no problem ...

[Mall]

I\'m guessing from your post that the \"stuff\" you read was JB\'s interpretation of the research studies, as opposed to the studies themselves. If that\'s true, this is an instance where I can invoke one of my favorite cliches: \"In God we trust. Everyone else we check.\" If you\'re as interested in this subject & have as open a mind as it seems, what you might want to do is read the actual studies & form your own conclusions. As explained & discussed approx 2 yrs ago in a thread entitled \"Moisture Content\", one conclusion I reached when I read the study in question was that because of the relationships between changes in moisture content on energy returned by the track, impact resistance of the track, & locomotor forces exerted by the horses running on the track, the impact of moisture content can vary with how fast the horses run. I\'m pretty sure that one of the specific points the five scientists made was that using their definitions, both \"slower\" & \"faster\" horses both got faster as moisture content increased up to a certain level, but after moisture content reached that level, the speed of the \"faster\" horses levelled off, while the speed of the \"slower\" horses got much worse. This is probably not something which would apply to the BC, but after the specified moisture level has been reached, could definitely come into play & justify a very different variant, to take an obvious & extreme example, even if the moisture content was exactly the same for the feature as it was for a not atypical nightcap consisting of slow NY bred mdns.        

TGJB,

I prefer breaking races out when they don\'t logically fit with the rest of the day as long as we \"know\" there was track maintenance, it was the only 2 turn race of the day, there were other identifiable factors that could account for it - like a strong gusting wind etc....

It\'s breaking out races on the assmption of a track speed change in absence of any evidence that gets into a grey area for me.

My biggest problem is that in those cases you could occasionally be building in adjustments for things other than track speed that might have impacted the final time of several horses in a race, but \"not equally\". That could cause an incorrect interpretation of the result and a lower quality figure.

Also, it\'s extremely difficult to interpret the results of some races. Some races are loaded with fist time starters, surface switches, layoffs, trainer changes, distance changes, complications related to bias etc... that make projecting a figure for an individual race more of an educated guess than an art. Races like that aren\'t so infrequent. I skip multiple races every day because I have idea how well many of horses will run under the conditions.

One could say that in these situations the sample size (the results and the horses prior figures) is much smaller and of lower quality than required to be confident. Even very experienced handicappers can come to very different conclusions about what happened.

It\'s just my opinion or personal preference, but I think the default should be to not break races out absent clear evidence of a track speed change. When the resultant figures would seem nonsensical otherwise, I think they should be broken out and highlighted in some way so I don\'t mortgage the house on a figure that has less evidence for its accuracy than is typical (like a nice neat day where everything fits).

I know you track and follow up on these things, but that\'s after the fact for people betting.

[bobphilo]

Class,

There are several things that can change track speed from race to race and not all of them are always as apparent as track maintenance and gross changes in weather conditions. There are subtle things like changes in ground water levels or combinations of sunlight and wind drying out the track, which could have more than subtle effects on track speed. We can never have all the variables to quantify theses changes in any case. Scietists know that we can presume something to be causal by it\'s effects, even when we don\'t know excatly what the mechanism of causation is.  I think, despite the possibility of error, we have to mainly go with what makes sense with relation to what what the horses are running. That\'s the way varients are calculted. Yes, by breaking a race loose, we are reducing our sample size, but by refusing to acknowledge what the horses performances are telling us because we\'re not sure what\'s causing the change in track speed is to make a geater error.

Bob

[BitPlayer]

TGJB –

I appreciate the time and effort you obviously put into responding to my post.  I know that long repetitive threads regarding figuremaking tend to be disfavored on this board, so I\'ll desist after making a few comments.

As a preface, I\'ll acknowledge that your judgment about what might or might not work with your data is much better informed than mine.  You\'ve seen thousands of sets of your variants.  I\'ve seen one.

I\'ll not respond to your Point 2 about using science to judge the validity of the Ragozin approach, since I\'ve already said I\'m in your camp on that one, and my suggestion assumes you are right.

I will, however, agree with Mall that anyone who relies on the material in Changing Track Speeds without looking at the underlying study is doing themselves a disservice.  The study is more interested in assessing what types of surfaces are likely to reduce injuries than in what makes horses go fast.  Indeed, it points out that energy return can be a bad thing, depending on its timing.  Energy return before the horse starts its next stride just increases the stress on the horse\'s limb without propelling it forward.  That aside, I think it is important to look at the data in the study itself, rather than the authors\' statements about it.  Their curve relating moisture content to energy return doesn\'t fit their data very well, particularly in the range of 6-8% water content.  It looks to me like their use of a polynomial function was misplaced, and that a positive correlation between moisture content begins about 6%, not 8% as their curve suggests.  (See Bobphilo, I did learn something in statistics: always plot your residuals.)

On to TGJB\'s points:

1a)  I think you tend to place too much emphasis on the number of other variables.  As you state, the study lists only three variables affecting energy return:  track composition, track compaction, and moisture content.  (I\'m not sure that\'s right – what about temperature, for example? – but who am I?)  The other things you often refer to such as sun, shadows, humidity, wind, etc. influence moisture content, but moisture content is still only one variable and changes in only one direction (down) absent precipitation or watering.  Further, as you mention in Changing Track Speeds, the track superintendent (Jerry Porcelli in that case) is using his best efforts to keep those other variables as constant as he can.

As for the average (or mean) being \"useless\" in determining a specific variant, I think you are oversimplifying.  I agree that you wouldn\'t blindly assume the mean is applicable to a specific case, especially when your evidence from the horses is strong, but when you are in doubt, the mean is (as a matter of probability) instructive about what is most likely to have occurred.

I won\'t defend Ragozin\'s book.  I haven\'t read it.

1b) \" If you assume my judgment is correct-- doing things the way I\'m doing them now-- why do the study? Why not just continue to use my judgment going forward?   Worse yet, if you don\'t think my judgment is correct, how can you use it in the study?\"

To quote Mall:  \"In God we trust. Everyone else we check.\"  The way I think science works in evaluating a system of measurement is that you apply the system to a case where you are pretty confident about what the measurements should show.  If it shows something else, you start to question the system.  My suggestion would serve as a check on your variants.  If your variants show a racetrack under relatively constant conditions reacting very differently to watering (and subsequent drying), either something is rotten in Denmark (to continue the Hamlet thing) or something other than track speed (pace?) is reflected in your variants.

Beyond serving as a check of your system, the data could, as I suggested above, serve as a source of guidance in making figures for difficult races.  I suspect you are already doing something like this on an informal basis, otherwise the track maintenance data you collect wouldn\'t be of much use to you.  In the BC Juvenile Fillies, you were hesitant to have your variant depart much from the surrounding races, because nothing had been done to the track in the interim.  Similarly, I\'m guessing that you have an idea that watering the track won\'t normally affect the variant more than some amount.

In closing, I don\'t know if the study I\'m suggesting is worth the time or not.  On one hand, data collection is always the hard part, and you\'ve already done that.  On the other hand, more projects like ThoroPattern and figure-based stats probably do more to enhance your product than what I am suggesting.  I just thought SP\'s suggestion was an interesting one and worthy of being discussed more fully than it was.

As always TGJB, thanks for your willingness to discuss these issues and for hosting this board.

BitPlayer

[TGJB]

Bit-- as I understand it, track \"speed\" is directly a function of energy return. Energy return, in turn, is a function of several things-- moisture content, compaction, cushion soil composition, cushion depth, and possibly others, like characteristics of the base of the track, and yet others I can\'t think of.

Moisture content is just one of those things. And

a) how it affects track speed (energy return) may be affected by things like compaction, soil content (sand/clay ratio) etc., and,

b) moisture content itself can be affected by many things. Among them is watering, as you said-- but also wind, sun, temperature, humidity, soil content (sand absorbs faster than clay, and at various times Porcelli would add sand and change the percentage).

What this means is that watering of the track is just one variable that affects just one variable (moisture content) that affects track speed. (This is why I wanted to get involved in the actual studies, where we could correlate our data more directly with actual hard data of more direct variables like moisture content, energy return, etc. themselves).

A point I made in the Expo presentation was this-- even if we assume that two variables are constant-- that the track is being watered before every race, and the amount of water is the same-- the moisture content of the track will only stay the same if the rate of evaporation is a) constant, and b) the same as the rate at which the water is being added.

Example-- lets say that under those circumstances, it\'s overcast and there is no wind for the first 3 races. Well, the track could be getting wetter, because the water is being added, but not evaporating. Then the sun comes out, and the wind picks up-- and the track starts to dry out. So you have it going up through various levels of moisture content, then going down through them. As we saw from the racetrack study, track speed will probably change, with no obvious major weather changes, and no significant track work (the watering is a constant, it\'s the climate that is not). But on top of that, HOW it will change will depend on what the moisture content of the track was when the day began-- and probably other factors as well, like the soil content of the track.

And finally this-- when I read a study that tells me that in some circumstances adding water speeds up a track, and in others slows the same track down (which by the way I have found myself), the last thing I\'m going to do is rely on an average to make a variant decision for me. Looking to see whether I can find something that would account for a change and applying a standard correction for the change are two different things.

All that aside, I would like to commend you and Bob for the high level of discourse. The more of that this site gets, the better.

bob,

I agree with you. However, IMO, there\'s a balancing act required between a potential track speed change (absent clear evidence) and the complications of determining what actually happened in the race. There isn\'t always clear evidence based on prior figures and sometimes races develop in a less than typical fashion confusing the issue further. I would prefer that the figure maker make the best judgement he can about whether to break it out (best guess) or lump it in with a few surrounding races to give a greater sample about what the variant is. No hard fast rule about what to do. But as long as are being honest enough to say some races and some days are tougher than others, I\'ve never seen the downside of a notation in those instances - other than a little extra prgramming work.

[bobphilo]

Class,

I understand your position. I was refering to a situation where, despite lack of \"physical evidence\" like track maintenance, the evidence of the horses performances points to a change in coditions for the race. I agree, it\'s a judgement call and that\'s why figure making is as much art as science.
I personally take for granted that this is likely the case in most of the figures I\'m looking at. As far as designating what is a \"tough call\", I frankly I\'m more confident with a figure that is strongly supported by the horses performance in that race, even if their is no other evidence, like track maintenance, than with a figure where there is other evidence but the horses figures are all over the place. Yet, I dont know how much it would help if Jerry designated which races were tough calls because the figures weren\'t tight fits
It would be even less important for me to know if there was maintenence which might have had no effect on the figure, save what has aready been shown by the horses\' performances.

Bob

[TGJB]

CH-- you are missing the point. There is no basis for \"lumping it in with surrounding races\", because there is no basis for the assumption that the track stays the same speed-- as the science shows.

Furthermore-- I\'m not aware of anyone other than me that is attempting to even GET watering and other track maintenance information. Keep in mind that Ragozin does not regard it as relevant at all-- Friedman has said so many times, most recently regarding giving out those way slow figures (about 5 points, on their scale) for the last 2 dirt races on opening day at Saratoga AFTER THE TRACK WAS SEALED IN THE MIDDLE OF THE CARD.

If you don\'t regard the information as relevant, you won\'t make any attempt to get it. Which in turn means you can\'t even really decide whether it is relevant.
How many times a year do you think there is work done like that affecting track speed-- and that\'s just extreme track work.

[bobphilo]

TGJB Wrote:
-------------------------------------------------------
>
> All that aside, I would like to commend you and
> Bob for the high level of discourse. The more of
> that this site gets, the better.

Thanks Jerry, It is my great pleasure to participate in a forum which features intelligent discourse as opposed to many boards and groups where personal attacks pass for discussion. This is what I'm trying to do with my group:
http://sports.groups.yahoo.com/group/Horses_and_Racing_Forum/

Please excuse the shameless plug. It's just a small group but includes a couple of good handicappers including a South African who is a keen observer of international racing and provides good insights on Euro shippers. Any participation by you guys would be greatly welcome and is invited.

Bob

[bobphilo]

For some reason the name of my group didn\'t come up in my post so, if it\'s ok with you Jerry, I\'ll repeat it.

Horses and Racing Forum. It\'s at Yahoo groups.

Bob

TJGB,

>CH-- you are missing the point. There is no basis for \"lumping it in with surrounding races\", because there is no basis for the assumption that the track stays the same speed-- as the science shows.>

That\'s true, but the tendency is for track speed to cluster within a pretty narrow range except when there\'s an event that could have changed it drastically. (maintenance, weather etc...)

Specifically, I am talking about the \"lesser of two evils\" situation.

In a case where we are working with several first timers, young developing horses, surface switches, trainer changes, distance changes, a non typical race development, or any other combination of similar complications, it might be higher quality to look at the variants for the rest of the day than trying to make one for that really tough race. If all the other races cluster around a certain variant rather tightly (a plain vanilla day), I\'d much rather go with that information than my best guess of a very complicated situation.

That\'s my personal preference. If you don\'t agree, there\'s nothing I can do about that.

99% of everything I have ever annoyed you about is really related to this issue and my beliefs about how race development can impact final time.

At a minimum, I would like to know about days/races like that before the fact and not after further review. I wouldn\'t even bring it up, but I don\'t see the downside. I pester everyone that makes figures for that kind of information. It\'s not specific to you.