This past Saturday\'s Laurel card was another example of what took place last 9/15 at Belmont, where I correctly said in advance that the inflexible nature of Ragozin\'s figure making approach would force him to do the day completely differently, and wrong. (The Belmont day was the one where he gave Storm Flag Flying a much worse number than in her debut, despite winning a Grade 1 by a block, after breaking her maiden by a length).
There was a lot happening on the Laurel day, but I want to focus on 2 consecutive races, both 7 furlongs, both graded stakes. In the General George, My Cousin Matt won, running 1:22:12. In the next race, Xtra Heat, who can run a little bit, required 1:24:76 to cover the same distance. The difference in figure terms between the two winning times is about 11 1/2 points, and even after adjusting for ground and weight (Xtra Heat carried 125), the colt earns almost 10 points a better figure than a filly who has lots of negative numbers, and was facing a good field, IF YOU DO THE RACES AT THE SAME TRACK SPEED.
Now obviously, this is nuts-- it is extremely unlikely that the colt race went around negative 10, but it\'s also extremely unlikely the filly race went in a plus 10-- not just because of the winners, but because of all the other horses. For whatever reason, the track changed speed A LOT in that half hour (which presumably had something to do with the wet track). So I did the obvious thing, which is to cut the two races loose from each other (Beyer did the same thing). Here are the possible ways the Ragozin office can deal with this:
1- They can tie the two races together. Good luck betting off those figures.
2- They can cut the races loose, in which case they are completely contradicting the reasoning put forth on their site for a whole bunch of other figures over the years.
Since no one but a blind dogmatic would choose 1 it will probably be this one, and I look forward to continuing this discussion when we find out.
Anyway, I used a SEVEN point difference in variants for the two races, which as it happens is very similar to the difference on 9/15 Belmont. Take a look at the sheets for the two Laurel races with the figures the horses ran that day (above), and see if you can find a way to add to the fillies or subtract from the colts a combination of 7 points, and still come up with something workable.
Your seven point difference in variants for the two 9/15 Belmont races turned out to be nearly perfect (as I suspected before the horses ran back). I think this time you might be stretching it a bit too far though. Going 7f over a heavy track, I can easily see My Cousin Matt beating Xtra Heat and the Carson City filly by seven or eight points. If the races were at 6f, I would understand your point, but if you look at the splits, the filly race just collapsed at the end. MCM ran the last 1/8 in an amazing :12.2, while Xtra Heat came home in :14.1. I do not think Xtra Heat will ever run a very fast 7f race over a heavy track. Going 6f on a quick surface, I would guess MCM is only a few points faster than Xtra Heat, but last Saturday, I have a suspicion that MCM ran much faster than Xtra Heat. Can you post the races again?
There is an attachment at the top of the original post.
As an infrequent player -- whose home track in Laurel -- who once had good basic handicapping skills but has become dependent on speed figures, I am especially attuned to the TG v. Rags debate down here.
I was not there on Saturday, which I can attest was a crappy, crappy day down here weatherwise. I actually had a track day on Thursday, when, unfortunately, it was all faves.
Anyway, while I\'m very much interested in how the debate regarding the two stakes turns out, the element I\'m interested in is the following: How would you do the figures if weather cancelations had not placed the two stakes back to back? Let\'s say the feature were followed by a group of typically unpredictable group of $8,500 NW2L or cheap maiden claimers. Or put two groups of equally difficult to predict groups of crappy Lrl horses back to back on a comparably shifting track.
My guess is that relatively large variances between groups of crappier horses are relatively common. As someone who has to bet on crappy horses, I worry that it is sometimes very difficult to do figures on these animals with much accuracy at all. This is important when trying to determine O2X patterns, explosive horses and particularly when a group of several horses coming out of the same race with shaky figures based on that race are running against others who may or may not emerge from easy or difficult-to-predict races.
Anyway, as someone who very much appreciates TGJB\'s ability to crush Derby\'s and Breeders\' Cup days, I\'d love to hear why you\'re better that Rags on the average days.
After all, you did start this string.
My Cousin Matt: 1:09.79, 1:22.12.
Xtra Heat: 1:10.61, 1:24.76.
What follows is from a Beyer speed chart. It\'s NOT the best way to do this, but I have no other way for Laurel, a track about which I know very little:
X Heat\'s 6F fraction, if considered a final time, gets a Beyer of 98. Her 7F final time gets a Beyer of 81.
Relationships between 6F FRACTIONS and 7F final times vary from track to track, but not by all that much, and not necessarily in this direction.
If Xtra Heat\'s 6F fraction is converted to 7F, it becomes 123.40
JB: If you make a quit number at your variant, it is off the chart. The losers are also insanely fast.
how does the rest of the card look when you ignore each of the above races in turn?
as a side note, i saw the replay of both of those races. it was so foggy, it reminded me of that race where the jockey just hung in the stretch until the rest of the field came around the turn and then he broke as if he had run the whole race. that was classic.
mholbert wrote,
\"as a side note, i saw the replay of both of those races. it was so foggy, it reminded me of that race where the jockey just hung in the stretch until the rest of the field came around the turn and then he broke as if he had run the whole race. that was classic.\"
Freidman gave that horse a 6, the same number he gave Chilukki.
First of all, I have a problem addressing anyone as The Kid, since that was my nickname when I was in the Ragozin operation.
However: the direct answer is that it would have made no difference whether the races were consecutive-- if anything it\'s easier to justify splitting them if they are not, since there is more time for the track to change.
On the larger issue, I\'ll say what I said to David Patent a couple of years ago-- horses do crazy things all the time, but groups of horses seldom do. You might want to check out that post(Figure Making Methodology, reposted here 1/31 this year). But yes, when you have a shifting track, and a 4 or 5 horse field of erratic horses where somebody wins by 7, or a field of mostly first time starters, things get tricky, and your chance of getting it right decreases. When it decreases to the point of becoming a guess I leave a box.
mholbert--
The track that day was all over the place, and anyone who uses one variant for the whole day is going to get some really screwed up figures. I used the two stakes as examples because they were well known horses and consecutive races, which makes the situation as clear as possible.
JB: If you use X Heat\'s 6F fraction as her final time, and use the same variant, what figure does she get?
If you use her 6F fraction as her final time, and use the same variant you used for the race before (meaning the My Cousin Matt race), what figure does she get?
There are several potential problems in doing this involving run-ups and differences in pitch of the chute, track maintenance etc., but the biggest (and this race provides a good example) is wind. There was a solid 8 o\'clock wind (if you are looking down on the track from above, wind blowing towards 8 on the clock) which is behind them both on the backstretch and turn, and in their faces in the stretch. You therefore can\'t compare the first 6f to the 6f races on the day, since you are subtracting a furlong into the wind and substituting one with it.
If we ASSUMED no other differences, we could work out a figure for a 6 furlong race ending at the eighth pole, but it would involve reconfiguring the wind program in the computer, which is based on races finishing at the wire, and would be a real pain in the butt. If you want a seat of the pants estimate, I would say there would be about a 3 point difference in the effect of wind on this day for a 6f race starting at the 7f pole and ending at the eighth pole vs. a regular one.
By the way, the 6 and 7f races at Laurel split often, and did on this day, with the 7f races coming up much slower, with the exception of the Gen. George.
Incidentally, I notice that Friedman has deleted a whole string of comments/questions about the Laurel day. Can\'t imagine why.
JB,
My Cousin Matt: 22.50, 46.34, 1:09.79, 1:22.12. Xtra Heat: 22.97, 45.57, 1:10.61, 1:24.76.
First, take a look at the final 1/8 My Cousin Matt ran into this wind.
Second, look at the differences in the times for the second quarter.
Matt was four fifths faster than X Heat at 6F and 13 fifths faster at the wire. Same distance, same wind. Did you look at the 6F fractions when you made the figures for these two races? Do you honestly believe that this track slowed down by seven points in half an hour? Or do you think X Heat ran a solid six furlongs and got tired?
One more on-point question: What are the historical relationships between 6F fractions and 7F final times at Laurel? I know the answer for CA races, but not Maryland races--though I can certainly find out.
Saw the fractions when I watched the races, yes, yes, don\'t know or care and it obviouly would be different for horses running zero\'s than horses running 25\'s. Incidentally, it nets out the same whether you view the track as changing or that it was a \"pace\" race-- you have to do it at a different variant either way, as we do for slow pace races all the time. And no, this does not mean I think we should be giving out quit figures, although you are certainly free to do so.
Should the fillies who came from off the pace be treated as though they were racing over a faster track, one the same speed as for the Gen. George? If so they will get really slow numbers, out of character with their histories. This is the biggest difference conceptually between hot/slow paced races-- when there is a really slow pace, ALL the horses are running slowly.
Kraven, who until recently used Ragozin numbers, had some interesting comments in an e-mail to me about how those slow pace races are screwing up Rags turf numbers, and he may want to get involved in this discussion. Or, he may not.
I thought this was a most interesting comment from Robespierre involving his own board policy:
\"that is an avoidance as much as possible of comparisons with other handicapping products.\"
What is he afraid of????
His product looking bad????
Let me rephrase the title of my previous post.
Handicappers you be the judge. If one horse consistently ducks another horse the connections are telling you something, are they not.
first, thanks for bringing this card up. this is great thought and discussion fodder.
how many variants did you end up using for the lrl 2/22 card?
i don\'t have a track model for lrl, but i loaded the 2/22 data into a generic model just for grins. for the non 3yo races, i had:
-5 (6f), -3 (7f), -9 (9f), 4 (6f), -11 (7f), -10 (9f), 3 (7f), -22 (7f)
these could be off 3-4 points, but the discussion is still the same.
i use fatigue models. (think of combination of pace and figure approaches.) using these, the races appear reasonable. although, even with that, the xtra heat race does require a small leap of faith because the variant on that race has the least variance among the entrants. i would probably just do a sprint/route split -6/-9
Okay. I was trying to get Friedman on the record about the figures before taking this discussion further, but I made the mistake of letting the discussion get sidetracked (again), which I\'m sure Len is very happy about. I\'m going to try very hard not to let that happen in the future.
1- You are correct not to use 3yo races for par levels. But once you have figure histories for the horses in those races, as we do, you definitely can use them as part of your process. In this case, those races came up very interesting, especially since one was the first race on the day, and the other was the only race after the Xtra Heat race (albeit a 2 turn race).
2- At Laurel, aside from the 1 and 2 turn split, the relationship between 6f and 7f is not constant. On the previous day (2/21) the 6-7 split was different than on 2/22, and even more extreme.
3- If you look at the day as flat, even by your method, you can see that the colt race (My Cousin Matt) is as out of line as the filly race. Once you factor in how slow the 7f races were compared to the 6f races, and make a correction, the disparity becomes even more extreme and obvious. If you do the Gen. George with the rest of the day, MCM and the horses behind him would get huge negatives, or you will be giving out awful numbers to everyone else-- unless you do a slide.
4- It rained during the card, and stopped before the two races I discussed here, after which the fog rolled in. The track was VERY slow for the first (3yo\'s, route), then got faster as it got wetter (this is true for the 6f, 7f, and routes independently). The fastest it got was for the MCM race. After that, the water was presumably sinking into the track, but for whatever reason, the track immediately got slower-- not only was the filly race slower, but the track for the final race (a 3yo route) was 5 points slower than for the previous route, the race right before MCM, run when the track was wetter.
i saw the same thing in the last race on the card. i tossed the 3yo races because i saw in an earlier discussion that you do that - just to remain consistent.
the variance from my fatigue models look like this:
1.09 / .38 / 1.05 / 2.83 / 1.46 / 1.17 / 2.15 / 3.64 / -.92 / 5.53 / 6.81
this is one of those cards that make you a fool or a genius.
Your fatigue model sounds interesting. Are you willing to describe it in a little more detail. If not on the board then maybe via email?
Chris
this article actually gives a pretty good introduction:
http://www.chef-de-race.com/articles/fatigue.htm
I am familiar with that model but was interested in how you used it to produce performance figures.
Chris