Good question but probably not the answer you wanted
[ Follow Ups ] [ Post Followup ] [ Thorograph -- Ask The Experts ] [ FAQ ]
Posted by Derby1592 on April 23, 2001 at 13:39:10:
In Reply to: Re: Note - It's much easier to read if you look at the Post Followup version posted by WRT on April 23, 2001 at 07:34:57:
: : : Last week I posted some unofficial, unauthorized guidelines for using Thoro-Graph sheets to handicap the Derby. In that post, I mentioned a computer model that I have recently developed that uses those principles to project the Derby figures for each horse. Below are the projections for the 2001 Derby (based on who I thought will be entered). Take all this with a grain of salt (make that a table spoon or two). I built this model on a limited data set and it is not that hard to develop a model the fits past data fairly well. It is much harder to develop a model that can predict future results. Since I just developed the model, I have no idea how well it will perform as a predictive tool but I thought I would share it with you all anyway. Who knows, I may be on to something here. Regardless, it has been fun putting it together and I have learned a lot in the process.
: : : Note that this does not take into account ground loss so, if you want to use these projections to help you pick the winner this year (you do so at your own risk), then you will need to make any adjustments accordingly since the fastest horse does not always win the race (e.g., check out Victory Gallop in 1998).
: : : I put the horses into 3 groups. The first being the real contenders since, if you have any confidence in the model (and I do not blame you at all if you do not), then if a projected figure is not within 2 of the highest projected figure, then the horse is very unlikely to win or even fill out the exacta. The horses in the second group have an outside chance to get a piece but are very unlikely to win. The horses in the third group really have no chance (I am sure I will hear it from the Balto Star fans...). Again, this is all based on the model. Not my personal opinion. In fact, sometimes I cannot even quite figure out how the model came up with a projected figure.
: : : So, if you are looking for the most likely winner, then the model says you have to go with the favorite this year (I can already hear the groans and chuckles...). Personally, I hate betting the favorite, particularly, in the Derby, but Point Given earned the highest projected figure that the model has ever spit out (by far). He looks like a legitimate favorite. Note that the model has only projected the favorite as the low fig horse twice in the last 6 years. Fu-Peg won the Derby and Timber Country ran a troubled 3rd in the Derby but went on to win the Preakness. So you have to respect Point Given.
: : : OK, you can stop making cracks about how anyone could come up with Point Given in this race. I have to admit that I am a little disappointed as well. But you cannot argue with the numbers and the computer could care less about the odds. However, the model also indicates that Point Given does not look unbeatable and that you could get some real value on Jamaican Rum and Dollar Bill. These two both look like they have a legitimate shot at winning or, at least, filling out the exacta. In the last 6 years, only 8 horses have had a projected fig of 4 or less and of those eight, 4 won, 2 ran second, 1 ran 3rd and 1 ran 6th. So these two are definitely very live longshots. Note, however, that this is by far the fastest field in the last 6 years, so these 2 could run a very fast race and still get beat. You may also get decent value on Millenium Wind and Congaree although they will probably go off as the 2nd and 3rd choices.
: : : So, depending on how you like to bet, you could hammer the favorite, key him in the exotics or pick one of the other contenders at a price.
: : : Perhaps a wiser move would be to handicap the race yourself and go with whatever horse you like. If you do, think twice though about tossing out any of the horses that the model identified as contenders.
: : : Good luck to all.
: : : Chris
: : :
: : : Horse Projected Fig
: : : Point Given 2
: : : Millenium Wind 2.5
: : : Jamaican Rum 3.5
: : : Congaree 3.75
: : : Dollar Bill 4
: : : Fifty Stars 4.75
: : : St. Damien 5
: : : A P Valentine 5.25
: : : Invisible Ink 6
: : : Monarchos 6.25
: : : Songandaprayer 6.25
: : : Express Tour 6.25
: : : Percy Hope 7
: : : Balto Star 8.25
: : : Thunder Blitz 9
: : : Keats 9.5
: : :
: : : I also provided the top 4 model picks for the last 6 years so you can see how it fits past data. I listed the year, the horse, the actual finish in the Derby and the project fig along with the actual fig earned by the horse in the Derby.
: : :
: : : Year Top 4 Projected Fin Proj Actual
: : : 95 Timber Country 3 3.5 5
: : : 95 Thunder Gulch 1 5.25 3.5
: : : 95 Tejano Run 2 5.25 5
: : : 95 Dazzling Falls 12 5.5 10
: : : 96 Grindstone 1 4 5
: : : 96 Louis Quatorze 16 4.75 13.5
: : : 96 Unbridled Song 5 6 5
: : : 96 Cavonnier 2 6 4.5
: : : 97 Silver Charm 1 4 2.75
: : : 97 Captain Bodgett 2 5.25 4
: : : 97 Jack Flash 7 5.5 7
: : : 97 Phantom on Tour 6 5.75 7
: : : 98 Real Quiet 1 3.75 3.75
: : : 98 Victory Gallop 2 3.75 2.25
: : : 98 Parade Ground 6 4 8.75
: : : 98 Halory Hunter 4 4.75 4.5
: : : 99 Menifee 2 3.75 3.75
: : : 99 Prime Timber 4 4.25 4.75
: : : 99 Lemon Drop Kid 9 4.5 5
: : : 99 First American 16 5.25 11.25
: : : 99 Charismatic 1 6 3.5 (FYI: not top 4 projection)
: : : 2000 Fusaichi Pegasus 1 2.75 2.75
: : : 2000 Anees 13 4.25 13.5
: : : 2000 Impeachment 3 4.25 5.5
: : : 2000 Aptitude 2 6 3
: I find this very interesting and was wondering if you have developed any future projection methods for say sprints, routes and turf races. In otherwords, a means of projecting or forcasting the likely performance of each horse in its next race. If so, would you share your ideas?
: Thanks in advance for your comments,
Sorry, the only model I have developed is specifically for the unique conditions of the Derby. Having some apparent success (too early to really tell) I have toyed with the idea of developing some more general models. I am sure it could be done. A rule-based expert model that was continuously tuned with real data using Thoro-Graph would probably be vastly superior to anything similar. Some intensive interviews with Jerry Brown combined with a good software developer and lots of data is probably all that is needed.
The big obstacle is the lack of data that can be directly input into software. I had to enter all the Derby info in by hand, which is tedious, time-consuming, and error prone. There is no way that I would do that on a regular basis for a normal race card. Because of this, I imagine that there are very few models out there even though the potential exists. The Thoro-Graph guys would have to produce a software data file similar to what BRIS does before any sheet-based software would be very useful.
If all that could happen, the possibilities are endless. If Thoro-Graph also produced pace numbers, then they could even incorporate expected ground loss in to the projections and who knows what other valuable handicapping factors pace numbers would provide.
It is a good question that raises even more good questions.
Post a Followup