There are two times of year when I look at the Ragozin figures and comment on them here-- in the spring with the 3 year olds, and for the Breeders Cup. I have pointed out an awful lot of problems in the past of all kinds, some of them of the variant (figure making decision) kind (SoCal sprint/routes,etc.), some of the ground loss kind-- Touch Of The Blues and several others being given wrong figures based on the ground being wrong in the 02 BC, Senor Swinger wrong in the 03 Wood (an error that resulted in wrong variant decisions being made and patterns being wrong on top 3yo\'s for months).
Now, we take the Derby pretty seriously. It\'s a big field, it\'s an important race, so we have two people do ground, and compare them. This year they came up slightly different on 3 horses, so we had them go back and look again (they had already looked a few times), and we eventually got it all worked out, and I did the race. It was a little tricky to do-- you clearly could not tie it to other races on the card, and an awful lot of horses did not fire, presumably because of the slop, so there wasn\'t much to work with. There were really only a few possible scenarios that needed to be examined, though-- from giving SJ his previous huge top, to giving him 3 points worse than that. Giving him his top would also have given Imperialism a 3 point top and LH a new top as well, and that seemed unlikely, especially under the conditions. At the other end of the spectrum we could have just paired Imp, which would have had 3 horses running 1-2-4 in the Derby while going backward, which also seemed unlikely. So it came down to doing the race either where I did it or giving them 1 point better, and the most likely scenario had SJ and Limehouse pairing their last, Imp going a little forward, and LH going a little back. If it ain\'t right, it\'s damn close, and considering the conditions (and commensurate lack of information) the best that can be done.
Anyway, when Ragozin posted his Derby figures I looked at them with great interest, as usual. At first glance it looked like he was giving them a couple of points slower than I did, which on their scale would mean a point or two faster-- they run 3-4 points slower than we do. I started with the winner, and there were some individual differences due to ground loss-- trackmen have tendencies, some use more of the \"fan\" than others, etc. This creates looseness of 1/2 point or so in everybody\'s figures, but I was looking for horses where the differences came to more than that-- like I said, we had two guys check the race thoroughly, so I was pretty confident ours would hold up.
But as I worked my way down, a funny thing happened when I got to Limehouse. All of a sudden they had given a horse 3 3/4 point worse. Aha! A ground loss error! The next horse, TCE, 3 1/4 point worse-- another one! And as I went down the list, it was true of all the rest of the horses! So we set up a chart, and this is what we found:
First 3 finishers, average deviation TG to Ragozin 2.33 points.
Rest of field, average deviation, 3.875 points.
Now, 1 1/2 points is a lot. In fact, it is 3 lengths at this distance. So there were 3 possibilities that I could see.
1-- Ragozin\'s trackman got the ground wrong for the first 3 horses by the same amount (about 3 paths each, combined 2 turns), getting them all wider than they were. That would be a hell of a coincidence, not likely.
2-- They got the other 15 horses wrong, each by the same 3 paths, all tighter than they should be. Even more unlikely.
3-- One of us got the beaten lengths wrong between the third and fourth horses, affecting all the figures from the fourth horse down.
3 was by far the most likely, so I immediately checked to see if we had it right. Equibase (where we get our data) had it at 2 lengths, we had them check it, they did and confirmed. I looked at the tape myself 3 times-- it looked to me like a fraction of a length more, but certainly not the extra 3 lengths that would account for the difference. We got it right.
But how could Ragozin get it wrong-- he does all that video frame counting stuff, right? We have a conundrum.
Look, guys-- this is THE KENTUCKY DERBY. If they get this one wrong, what does that tell you about their day to day work? Those small \"condition moves\" those guys are talking about all the time? You now know what the margin of error is for those-- plus or minus 1 1/2 points. IN THE BIGGEST RACE OF THE YEAR. And that is assuming that all else is done right-- see my DRF expo presentation on our home page.
Ragozin customers-- there are 2 ways you can react to this. One is to shoot the messenger-- get angry at me for letting you know. The other is to aim your anger and questions where they belong. As I said after the Touch Of The Blues fiasco, it will be instructive to see how the Ragozin office deals with this information.
TGJB-
I had sent an e-mail that you may not have received concerning ground loss (sent it to TGJB@thorograph.com). In it I had asked whether ground loss figures were determined by more than one person. You write here, and I agree wholeheartedly, that given the importance of this race it is important to get the ground loss figures as accurate as possible. So I appreciate the attention to detail. However, were the figures for the Derby prep races given the same consideration? Were ground loss figures determined by separate individuals and agreed upon?
Kingfisher
Same attention to detail yes, same number of people no. The preps-- like most races-- have much smaller fields. Our trackmen-- in many cases the Equibase trackmen, Dave Litfin in New York, etc., have access to as many replays as they need. The major reason for using more than one guy for the Derby is the large number of horses make it more possible for someone to transpose two horses with similar jock colors, and magnifies individual trackman tendencies-- one guy could have a horse 3, another could have him 3-3-4-5 because he uses more of the fan. In a smaller field the differences are less extreme.
Questions like this are better asked here. For private questions, JerryB@thorograph.com, or webmaster@thorograph.com for general stuff.
JB, Belmont has played very outside(VERY DEAD RAIL) for the past two days. It is a distinct advantage to race wide and come wide around the turn. Besides an X for dead rail, do you make any adjustment for the fact that being wide has been a distinct advantage over saving ground on the turn.Normally you would give credit in your # for ground loss, but there seems to be a case not to do so in this instance.
First of all, we have the computer programmed to look for possible dead rail days. When it flags one, I\'m aware of it when I do the day, so I
a) look to see how the rail horses ran in figure terms, and decide whether the rail was in fact dead, and therefore whether to give them the \"x\"
b) if the rail is dead, I\'m not using those horses to make the variant-- I\'m ignoring them and doing it based on the others, so I won\'t be giving extra credit to horses who ran outside. I would point out that this is not true for those who don\'t acknowledge, look for, and therefore take into account, dead rails-- like Ragozin. If you go back in the archives to shortly after the 01 Breeder\'s Cup (Belmont), you will see a real brouhaha on the subject, and if you follow up the posts over the following weeks you will see how it played out.
As I said then, if you don\'t take the dead rails into account, you are creating an average track speed that doesn\'t correctly assess the performances of either the rail horses or the wide ones. It\'s the same concept as using one track speed (in effect an average) when it actually is changing throughout the day.
On an unrelated subject, I want to say that I LOVE CC\'s post on the Rag board in response to Rule By Reason, who is not me, and who is as far as I know a Raggie, posting my comments about the Derby lengths, and asking Len what\'s up. You got it, CC-- Rule By Reason\'s response isn\'t rational. Yours is.
Seems you have this covered well, although I don\'t know how accurately a computer would flag a dead rail.
Giving away trade secrets-- we already have the ground input so that the computer can spit out the \"rundown\", figures for all horses that ran that day based on a mechanical variant, taking into account wind, time, lengths beaten, and ground loss, which I then adjust by creating final variants. We simply have it flag days where 2/3 or more of the horses who are on the rail finish in the bottom of their fields, and mark those horses on the rundown so I can look at them.
seems sound enough logic to me and I consider track bias a very important factor when one truly exist.thanks for the info.