JB,
Having seen your presentation and listened to the various scientific evidence you gathered from experts, it isn\'t hard to accept the fact that the track changes during the day. Porcelli\'s comments about watching the times and then deciding to water the track are a bit scary, but further re-enforce this.
I have a few questions though. Have there been any studies to try and measure what the difference in running time would be between a track with say 4% moisture and then a track with 10% moisture. Is it significant or extremely marginal? Obviously, your analogy during the presentation about running on beach sand and then running next to the water where the sand is wet is extreme. Do smaller changes matter enough to warrant making changing track variants?
If we accept that the track variant CAN change during the card, my next question would be HOW OFTEN does it happen and do you need to incorporate this into your figures? Is this something you are accounting for EVERY day? Does it happen once a week, once a month? Can you give us a feel for how often you take this into account? It would seem to be this would happen infrequently where the change in moisture, wind, shade, etc would be significant enough to warrant multiple variants? Or is your view that this happens every day and you do multiple variants every time you do a card?
Last question, relating to figures being \"bunched\" and this being the way that every figuremaker does figures. Would you say it is an accurate statement that the Thorograph figures show more pairups and bunched figures than Len\'s RAgozin figures (this was the original point made by the guy I was talking to). George\'s reply explained why the figures are bunched, but it sounded like more of a general statement about all figures, rather than just T-Graph figures.
I understand why this has to be the way figures are calculated, but to be honest, it almost seems like a string of figures can become a self fulfilling prophesy. It almost reminds me of college, where I always felt the single most important papers I would write each class would be the first and second, because once I established with the Prof that I was an \"A\" student, expectations are set and all papers after that were read with that predisposition. Sorry for lousy analogy, but it does remind me of that.
It just seems that one or two mistakes, which we all make, can lead to a long string of mistakes, if you don\'t look at each race independently, instead of such strong reliance on previous races. The horse that immediately comes to my mind is Smarty Jones. Having watched and bet on horse races for 20 years, I will believe the world is flat before I believe that Smarty was the fastest 3 year old ever (I think both T-Graph and Ragozin came up with this).
Jim
Jim-- lots of stuff here, and I\'ll try to get to all of it. But between Red Sox-Yanks, golf, betting 2 tracks and trying to get some work done, it may not be today.
JB
Okay, Jim asked a lot of good questions that go to the heart of the matter when making figures. I\'ll deal with them in sections, partly because this will be time consuming, and because it will be a lot for people to digest. It will make things easier if everyone waits until I finish with this some time tomorrow before asking questions or commenting.
\"Have there been any studies to try and measure what the difference in running time would be between a track with say 4% moisture and then a track with 10% moisture\".
\"Do smaller changes matter enough to warrant making changing track variants\"?
The only published scientific study I\'m aware of directly concerning moisture content is the one I referenced in the Expo presentation, which is available on the web-- Mall put a link to it here originally, so you might be able to get to it with a search on this site, or you can do a larger search using the title (\"Interrelationships Between Moisture Content Of The Track, Dynamic Properties Of The Track And The Locomotor Forces Exerted By Galloping Horses\"). As was noted in my presentation, relatively small changes in moisture content produced scientifically MEASURABLE differences in rebound energy, which in our terms means track speed. Think about it-- if the differences produce a 1% difference in the \"speed\" of the track, and we\'re talking about a mile race going in 1:40 (100 seconds), that\'s a difference of a second. Which in racing (and figure) terms is a lot-- about 4 points.
But there is no way to directly answer your question, because how do you scientifically measure track \"speed\"? That is why one of the scientists sent me the e-mail saying that he thought the most accurate data for measuring this was ours-- if we had a way of correlating moisture content in a blind study with what we determined to be track speed, it would be interesting to see the results. I offered this, did not get a reply-- I\'m going to bring it up again at some point.
The other point to keep in mind is that soil composition plays a big part in this-- some tracks are going to be more responsive to moisture than others. Likewise, a difference between 3 to 5 percent might not have the same effect as that between 5 and 7 percent at a given track, and different parts of the track can have different moisture contents because of shade, drainage, and different track maintenance (at Belmont and Aqueduct, for example, they don\'t use the big water trucks in the chute, according to Porcelli).
So all in all, what we know is that moisture content makes a difference, and that it\'s hard to quantify directly what that difference is. Which leaves us knowing that it is wrong to make the assumption that track speed (to say nothing of the relationship between distances) stays constant, but having no OBJECTIVE way of dealing with it. Meaning, we have to use what that scientist agreed was the best available method-- the past histories of the horses who run over the track-- to determine the \"speed\" of the track.
More tomorrow.
Okay, let\'s get to the next group of questions Jim asked.
\"If we accept that the track variant CAN change during the card, my next question would be HOW OFTEN does it happen and do you need to incorporate this into your figures? Is this something you are accounting for EVERY day? Does it happen once a week, once a month? Can you give us a feel for how often you take this into account? It would seem to be this would happen infrequently where the change in moisture, wind, shade etc. would be significant enough to warrant multiple variants? Or is it your view that this happens every day and you do multiple variants every time you do a card\"?
The short answer to this is, in effect, I do seperate variants for each race, but they often come out the same, give or take-- there is leeway within a point or so either way even if the track is staying basically the same, due to wind gusts, slight effects of pace on the race, slight differences in runups that effect how fast they are moving when they break the beam, etc.
A couple of points--
1-- I never tie the one turn races with the two turn races regardless, because history has shown that relationship is not solid. This is true because of several things that I brought out in the expo presentation, but way before I became aware of WHY it was true, I knew that it WAS true-- you end up making an average variant that doesn\'t treat either sprints or routes fairly, or treats one fairly but not the other. It doesn\'t matter that SOMETIMES they are at the same variant-- it is wrong to ASSUME they are. If they end up that way, fine.
2-- Lots of things can effect the time of a race, and if you are not paying attention to wind and run-ups (I don\'t think the Beyer people are, though I could be wrong, based on a conversation I had with Randy Moss a couple of years ago) you will have to make significant corrections to your figures AS IF you were correcting for changes in track speed, or time errors. The same thing goes for races with an extremely slow pace. In other words, if you are going to ASSUME you have all the information, you better be damn sure you have it-- for a good discussion and a great example of this, see my post \"Figure Making Methodology\", reposted 1/31/03 on this site.
3-- If I\'m going to cut a race loose-- that is, decide that doing it with the surrounding races is wrong, and make a really significant correction, without knowing something happened to the track (or some other event that would affect time), it\'s going to be not just because it looks wrong, but because it is clear what the RIGHT way to treat the race is. If I move it I know what the right variant for the race is by seeing what several horses in the race did, at least to the point of knowing I have it close. I do NOT screw with the relationships within a race to make it come out right. If it looks like it could be wrong but I DON\'T know what is right, I leave a box (no figure)-- this happens mostly with lightly raced horses (like with Chilukki\'s debut effort). And yes, if I give a figure but the decision is really tough, I mark the race and go back to look at it about 6 weeks later-- there are about two a week of these all tracks combined, and I end up changing about 20% of them, usually around 2 points.
Days take many shapes. Sometimes the track stays the same speed all day-- I guess this happens about half the time. Sometimes there is a distinct split variant, but not that often. What you see a lot is the track getting faster or slower as the day goes along, sometimes at a uniform pace, sometimes not. Sometimes it gets faster (or slower) and levels off at some point, or stays the same for a few races, then begins to slide faster or slower. Any permutation is possible, unless you DECREE that it must stay the same speed, and use that as an assumption when making your figures. An assumption that flies in the face of both information and science.
Because, as I said in the expo presentation (still up on this site), moisture content (aside from any other factors) can be changing throughout the day. EVEN IF WEATHER IS A CONSTANT, EVEN IF THE SAME AMOUNT OF WATER IS ADDED BEFORE EVERY RACE, MOISTURE CONTENT WILL STILL ONLY STAY CONSTANT IF THE AMOUNT BEING ADDED EACH RACE IS THE SAME AS THE AMOUNT EVAPORATING BETWEEN EACH OF THE RACES. If it is not, the track will get either wetter or drier as the day goes on. And, of course, weather is not constant-- wind picks up and dies down, temperature changes, the sun comes out and goes behind clouds, humidity changes etc. And if you saw the track maintenance records in the presentation you also know that they don\'t add water at all tracks between all races, let alone the same amount.
A good example of what I\'m talking about took place this year on Belmont day, and is detailed in \"Friedman/Changing Track Speeds\", recently reposted here. They had expected rain which didn\'t come and done a lot of work on the track overnight, then watered the track after the first six races, then stopped. The track got faster as the day went on, then levelled off. Now, I saw this in the figures and would have treated the day that way even if I didn\'t have the track maintenance info-- BUT WHAT HAPPENS IF YOU SIMPLY MAKE THE ASSUMPTION THAT THERE WAS NO WEATHER, SO THE TRACK DIDN\"T CHANGE, as Ragozin says he does in his book? You get it wrong, that\'s what. Friedman indicated in a post that they knew about the work done, so they slid the day-- but what about places where they (or anyone else) doesn\'t get the info?
Make as few assumptions as possible. I\'ll get to the questions about \"bunched\" figures tomorrow.
Okay, on to Jim\'s last set of questions.
\"Last question, relating to figures being \"bunched\" and this being the way that every figure maker does figures. Would you say it is an accurate statement that the Thoro-Graph figures show more pairups and bunched figures than Len Ragozin figures (this was the original point made by the guy I was talking to). George\'s reply explained why the figures are bunched, but it sounded like more of a general statement about all figures, rather than just TG figures.
I understand why this has to be the way figures are calculated, but to be honest, it almost seems like a string of figures can become a self fulfilling prophecy. It almost reminds me of college, where I always felt the most important papers I would write each class would be the first and second, because once I established with the Prof that I was an \"A\" student, expectations are set and all papers after that were read with that predisposition. Sorry for lousy analogy, but it does remind me of that.
It just seems that one or two mistakes, which we all make, can lead to a long string of mistakes, if you don\'t look at each race independently, instead of such strong reliance on previous races. The horse that immediately comes to my mind is Smarty Jones. Having watched and bet on horses for 20 years, I will believe the world is flat before I believe that Smarty was the fastest 3 year old ever (I think both TG and Ragozin came up with this).\"
1-- YOU MEAN THE WORLD ISN\'T FLAT???
2-- I don\'t see Ragozin\'s figures these days except occasionally on big days, so I don\'t have a first hand opinion as to who has more pair-ups. I have been hearing we do for years, and usually from the Ragozin camp as some kind of evidence we are doing something wrong. I suspect that graphics and page layout may have something to do with all this, but I hope (and would certainly think, given the way Ragozin does things) that we do have more pair-ups.
Again, all figure makers make figures by using previous figures of the horses run, one way or another, whether it is by a crude system of claiming pars, or the far more evolved projection method, which increases your number of \"data points\" (thank you Jason Litt) from the few horses who won the claimers to every horse who ran on the card, and all their previous numbers.
Now let\'s say you have a two race card (to make things simple), and the two winners both get the same winning figure. And let\'s say they both ran a 10 in their last race, which was the best for both in their lives. I (and everyone else) will start out by giving them 10\'s, and then look to see what figures that gives those behind them-- remember, those relationships within each race are fixed, by beaten lengths, ground loss, and weight. If when I do it that way the second horse in the first race pairs up as well, it would be extremely strong evidence I\'m on the right track. And so on with looking at the rest of both fields.
In other words, we don\'t say these horses run 10\'s when they run well, so I think I\'ll give them 15\'s, or 5\'s. It may turn out that way depending on the OTHER horses, but for me to assign other figures THE OTHER HORSES WOULD HAVE TO YIELD MORE PAIR-UPS (in a perfect world),OR MORE FIGURES IN THE RANGE THEY USUALLY RUN. Either way, the way you know your data base is tight is then by the number of horses who pair up, or run in a tight range. Every serious figure maker knows this. You try out various scenarios-- for the individual races as well as for the day as a whole-- in an effort to make it happen as much as possible.
If you tie together independent events (races run over tracks with different speeds, as Ragozin does), you will have less numbers that pair up, because you are in effect using an average track speed, and things won\'t be as tight.
If, on the other hand, you DO have a lot of pair-ups, it is not just an indication that you have this day right, BUT THAT YOUR EARLIER FIGURES ARE RIGHT AS WELL.
3-- Yes, mistakes can compound themselves, especially if (as Ragozin says somewhere in his book-- I just took a fast look and couldn\'t find where) you sometimes tie a whole day to one or two \"solid\" horses. (Again tying together unrelated events-- seperate races-- as well).
But there are checks and balances built into the system if you do it right, because the relationships within each race are fixed. You use LOTS of horses, and THEY COME OUT OF DIFFERENT RACES, AND DIFFERENT DAYS. If they moved together from race to race as a block, yes, the problems could compound themselves indefinitely. But if you pair up a horse or horses to bad figures, you will be unable to pair up other horses to their other figures, unless you have happened to make exactly the same size mistake several times, in the same direction.
4-- Ultimately, there is no way to look at an individual race \"independently\", without looking at previous races-- what would you base the figure on? The question that differentiates Ragozin from us (and from other serious figure makers, including Beyer) is, how dependent do you make the figure on other races, run by other horses, over the track at other times?
All done. Comments? Questions?
Earlier, you wrote
\"Also, the Kentucky Cup Card featured some extreme differences in the relationship between our Ellis Park figures and other \"major\" circuits that worked out very well.\"
Can you elaborate on this?
HP
In the KY Cup Juvenile (race 10) Beyer had Storied Cat much faster than the EP horses, and the betting reflected it. We had three EP horses MUCH faster than Storied Cat, and one as fast. Three of them ran 1-2-3, with the winner being second fastest going in, and arguably the most likely winner given the post and spacing. Similar situation existed, though not as clear cut, between Level Playingfield and Heckle/Swift Attraction in the sprint. LP won at 10-1 in a short field, and as Jim said earlier he was second fastest going in, with Cuvee a guessing game as to how he would run.
The \"shipper\" numbers holding up angle is definitely one of the TG strong suits. There are some really interesting track-to-track things that I notice over and over (e.g., Maryland horses running to their Maryland numbers in NY and not getting bet). Thanks.
HP
JB,
I am not at all surprised by figure discrepancies between various figure makers when it comes to turf racing.
1. There are generally fewer turf races per day relative to dirt from which to try to determine the track speed.
2. They change the rails sometimes and thus change the relationships between various distances from day to day.
3. There are many more very slow paces on turf that impact the final time. Not every figure maker bakes the impact of pace into the final figure. Some prefer to view pace and final time as seperate issues.
I am more surprised by discrepancies when it comes to track vs. track figures when we are talking about dirt races.
First, you obviously have to adjust for the different methodologies. Beyer figures do not contain weight and path information, but you could make those adjustments yourself if you wanted to. Then you would be comparing apples to apples instead of apples to oranges. I think most serious Beyer figure users do some adjustments like that anyway.
If you went through that adjustment process, I would be surprised if the discrepancies were huge for any of the major tracks for dirt racing. I know for certain the Beyer group has access to all the figures they supply the Racing Form via computer and that they cross check all their figures when horses move from track to track to ensure that the figures at various tracks are in sync.
I am not saying that the Beyer figures are "equal" to TG for dirt racing, but I think on an apples to apples comparison they are unlikely to be bad for the major circuits. The turf figures are an entirely different matter.
I think the TG turf figures are much more representative of the horses abilities.
Post Edited (09-21-04 10:53)
Speaking of shippers, how do you ensure consistency between your Euro and US turf figures? I remember there being some discussion after the Arlington Million that the Euros seemed to outrun their figures.
CH--
I agree that the Beyer figures are decent at representing what they aim to (meaning they don\'t take ground and weight into account), and they don\'t do some of the dogmatic, unscientific things Ragozin does. I was sitting 3 feet from Andy at the expo when he announced his track-to-track computer program to the world, and it sounded so good that we\'re going to do our own, as soon as we get done with a couple of other projects. That notwithstanding, they have some problems (as does Ragozin), and the relationship between the EP figures and Saratoga is one. At first glance it is also reflected in the figures assigned for the winners on KY Cup day-- I haven\'t done the day yet, but I\'m guessing I\'ll be giving out better figures. You make the figures based on the earlier figures.
The funny thing about turf figures is, although on the surface they appear to be tough to make (not as many races), and some are almost impossible to make (lightly raced horses running on days there are no other grass races, or weather makes it wrong to tie the race to others), on balance they are much easier to make, and easier to get right, if
1-- You go by the horses. Grass horses are unbelievably consistent, running in very tight ranges. And before anyone says I just give them what I want, remember that the relationships between horses in a race we are looking at are fixed-- I can\'t pair the winner to his 7 and the second horse to his 9 and the third to his 10 1/2 unless the winner beats the second horse by 2, and the third by another 1 1/2 points.
2-- You use weight and ground. Because they run fewer grass races, and grass horses do run in tight ranges, getting the figures really accurate makes all the difference when you use them to do variants later. In simple terms, if horse A gets a rail trip around 2 turns and horse B is 3 wide both turns, the difference is about 2 points, which will not show up on Beyer (it would be 6-7 of his points). But it makes an enormous difference both in hanficapping these closely matched races and in making figures using the figures. Likewise weight.
3-- You don\'t do anything else that can screw things up. By this I mean, take the time of the race too literally, as crazy as that sounds. Pace has a huge effect on final time in grass races because on average it is considerably slower than in dirt races, despite grass courses being on average much faster (especially nowadays, with all the sand in the tracks)-- and sometimes the pace is REALLY slow. If you take the time too literally, or try to use some formula to adjust for slow paces (as Friedman indicated on the Ragozin site they did) you can seriously mess things up, and make it harder going forward to use those figures to make other figures (Jim\'s point from a couple of days ago). Trust your figures, go by the horses, and don\'t screw with the relationships within the races, and you\'ll be fine after you get your data base in shape, which takes time.
By the way, this is what TimeForm and other serious figure makers do. When you deal with short race meets, hilly courses over about distances, and races not only with slow paces but no fractional times, there is no other way to do it.
We TRY to ensure consistency of the European figures by looking at what happens when they run against North American horses, here and in Dubai. But it\'s tricky, because a lot of them are first Lasix here. It\'s a never ending struggle...
Interesting comments about making turf figures. (I never made turf figures for myself even when I was heavily into making both pace and final time figures for dirt)
If Ragozin is using a formula for the slow paced turf races, IMO, those figures would have to be highly suspect at best.
I\'ll let you know if I see anything funny in the relationship between EP and Saratoga for Beyer.
JB,
Figures for these European turf horses are a real challenge. I know I have only a few months experience with the T-Graph figures, but it seems that on both Beyer and T-Graph, the European horses that have come over have \"outrun\" their figures.
For years early on in the Breeder\'s Cup, I made some of my few winning bets on that day betting against European \"superstars\" like Dancing Brave and taking our best turf horses like Theatrical and Manila. However, it really looks like in the past 3-5 years, there is a drastic superiority in the European turf horses. Magistretti is really a \"second stringer\" from Europe and he trounced some of our best turf horses.
So, I guess I have two questions on this topic.
The first is more general, any ideas by anybody out there as to WHY the shift in superiority to the Europeans? Are they getting better bred horses? Is it the lack of focus of American trainers on grass racing, relative to turf racing? If so, why did it take until now to show up, we used to do well against Europe\'s best in past years.
The second is to Jerry. Not sure if you agree with my point about Europeans outrunning their figures, as I know I am putting a lot of emphasis on only 3 months of data, but if it is true, what do you attribute it to? Do you get good information on the European races with which to make figures? Is it the difference in style where they really lope along in Europe for most of the race and then sprint home, thereby skewing the final times a bit. I know you don\'t really believe that pace impacts figures much, but it has to impact final time if the pace is ridiculously slow, which happens a lot in Europe.
Thoughts?
Jim
I would point out that it hasn\'t been just 3 months, but very few races within that time. A lot more European based horses should be running here over the next 2 months.
As I said, one of the problems with analyzing this stuff is that some horses could be jumping up first lasix. If over a period of time it looks like we haven\'t given them enough credit (and that is far from clear-- in the past couple of years and in Dubai the stuff held up very well) it probably means our whole scale for Europe needs to be moved a little. I will add that we are now working on a full English data base wich should be ready for the beginning of their next racing season, and will be the most accurate, most comprehensive, most user-friendly data base for those races ever devised.
Truly accurate figures for Europe would be very welcomed. If you feel they are holding up well, maybe I need to revist this. I don\'t use them yet.
I prefer to use \"class\" when evaluating the Europeans vs. the USA - partly because IMO the complexities of making turf figures in general are even more extreme in Europe.
In general, I find that equal grade stakes races in Europe (from France and Great Britain) are better than those in the US.
I have had a great deal of success (for a limited sample though) by classing the horses, looking at the purses for races (even from other European countries), and only casually following the scene there.
Three things seem apparent.
1. If there are several European shippers in the same race, the public tends to focus on the one that looks best on paper and bets that one. However, the 2nd or 3rd best on paper often go off at a very big price even though they are only sightly worse. IMO, there\'s value in that.
2. More than I would expect run huge races first time out and then run quite poorly 2nd time in the US. (bounce?)
3. There are some huge bargains in Breeder\'s Cup races.
Anyone interested can get free \'DRF style\' PP\'s here for UK cards in PDF format...
https://secure.raceform.co.uk/
Also if you want to get a feel for UK internal split timings you can have a look at some here...
http://www.newmarketracecourses.co.uk/results/sectional.jsp
Added later...
I should add that Raceform have been around since Jesus was a boy. And that with regard to the sectional timings you need to remember the penultimate furlong is downhill and the final furlong uphill. Flat otherwise with minor undulations.
Post Edited (09-21-04 19:25)