Friday, October 12, 2007

Breeders Cup Profiles

One of my favorite pass times is working on projects that involve statistics and numbers. The Thoroughbred Championship Ranking System is a point system that attempts to rate the accomplishments of horses and condense it all to a single number. Accomplishments are an important component of decided championships. Here at Kennedy’s Corridor we host the TCR Awards where the ranking results and votes are combined to hopefully give us the most deserving champion.

Another project that I dearly love is one that I call the Profile Of A Winner. In essence its an attempt to identify and combine meaningful statistical factors to come up with a clearer picture of who is and who isn’t a good candidate for success. Much of the inspiration for the project came from reading Progressive Handicapping Inc’s Crushing The Cup. Although the system itself is quite different the intent is basically the same, and that is to give you a clearer picture of how well suited each contender is from a statistical perspective. It is not meant to be an automatic selector that removes the need for discretion. A players own thoughts and ideas must be applied but I think you’ll find this system does give you a different perspective on horses who you may not have given much consideration to or sometimes paints a negative picture of heavy favorites.

Profiling horses for specific races is not always effective and the main reason is because in any given race nearly all of the horses are entered with different goals. Some are coming back from a layoff. Some are at the end of their form cycle. Some are being raised in class, others dropped in class. A specific race may have been the target of a particular horse while another might be using it as a prep. With so many different types of intentions from the connections it is impossible to build a successful profile. However the Breeders Cup does not have those same variables. All horses who enter it are pointed specifically for it. No one uses it as a prep and it is almost never an after thought. Everyone comes up to the race in the manner that they feel gives them the best chance for success. Consequently we are able to identify and isolate those patterns and combine them into a profile.

None of the factors included are jinx or myth. There is no quantifiable reason why a certain horse would not be able to win a Breeders Cup race simply because he won a certain “jinxed” prep race, or because his name starts with a certain letter. Every single factor must be aimed at identifying six key ingredients to winning: Speed, Fitness, Current Form, Class, Experience and Suitability to the conditions.

The factors for each Breeders Cup race are unique to that race. As you might expect one requires different qualifications from a possible Breeders Cup Turf winner than one would a Juvenile winner. However some of the factors do overlap. For instance every race Profile demands that the qualifier finish in the money (win, place or show) in their last start before the Breeders Cup. 153 of the 169 (or 90%) Breeders Cup race winners met that criteria. The only exemption from that factor is runners exiting the Prix de l’Arc de Triomphe. Given its championship status it is treated different from other races. It is not a prep race but rather a target and sharp races can be readily obscured because of the depth and quality. Again this is not to say that a horse who was out of the money in their last race cannot win. Miesque’s Approval just did in the 2007 Mile, but rather the stat draws attention to the trend so you can make a more informed decision about who is likely to win. If you know that 90% of the races are won by a horse who did run in the money in their last than you may decide to stay with the trend or go against it, but most likely you would only go against it for a very good reason.

The systems focus is not on eliminating horses who can’t win. The focus is in trying to quantify the relative chances of each entrant. Every BC starter is screened through the Profile and given a score based on how well they fit the Profile. I’ve entitled that the 20-20 score. I named it 20-20 because at the time I was working on the Derby Profile which has 20 statistical factors and 20-20 is a description of perfect vision. Perfect vision more or less describes the intent of the project, it does not select winners, it gives you an unclouded look at the field so you can select winners. Each entrant is given a 20-20 score and graded based on that score.

For the purposes of the project I have compiled the 20-20 scores of each Breeders Cup entrant and proposed that a win wager equal to their profile score be placed on each horse. So in essence you’re betting every horse (unless they have a negative profile score) but in different amounts based on their statistical chance of victory. You can view the 20-20 scores of each BC entrant from 1996 to the present by visiting this link.

I realize that no one in their right mind would wager like that, but the purpose of the exercise is to show you the value of the system even when no discretion is applied. If one can make money by blindly following the system then you ought to make even more by exercising some sound judgement when you’re actually at the windows.

If you visit the spreadsheet you can see that betting on every horse at the levels determined by the profile you net you a 27.92% profit. Meanwhile betting the same amount on every single horse would have yielded a –2.28% loss. So clearly there is value in what the profile produces. Another great thing about it is that it’s not dependant on single boxcar winners for profitability. The equity curve is actually quite smooth with only 2 losing years out of 11.

Another way to use the system might be to only look at the horses who had a perfect profile score for their race. Over the last 11 years there were 186 horses with a perfect score in 78 races. So on average you’ll get 2-3 horses who are “perfect”. These 186 had a cumulative BC record of 186-56-24-22, so 1 in 3 was a winner and 56 of the 78 races were won by a horse with a perfect profile score. That’s 71.79%! Betting on all of these horses would have yielded a 131.24% profit with no losing years.

I will perhaps post a complete profile for one of the BC races that shows all the statistical factors included and the reasons why they are included later this week.

Before we get too carried away with visions of wild profitability you should know that this is the first year of live testing using this system. Past results are not always indicative of future performance because it is possible to manipulate data to make it say what you want. I have not done this consciously but I would not advocate blindly following this project. From the outset the idea has always been to create something that is useful in refining your handicapping not eliminating the need for judgement entirely.

When I started the blog in October of 2006 I posted a primitive working of this system for the Breeders Cup that focused only on the “perfect” qualifiers. I have since made many revisions so I’m looking forward to seeing how it does this year. I posted the Kentucky Derby version of this project back in May and like the BC system it was the first year of live testing. The system did not perform very well as Street Sense was not given a high profile score. So there, you have your fair warning!

No comments: