[BJennet]

Originally Posted by J2EEDeveloper

Ted,
I appreciate your lengthy reply, and I will certainly take your advice seriously. To avoid any misconceptions, I am not a confirmed loser attempting to validate my losses by declaring everything impossible to win with.

On the contrary, I have had a healthy profit every year for the past eight years, using past performances, hand methods, and simplistic calculations. Whether or not it is possible to win is not the issue; almost anyone can turn a profit if they are willing to do the work of pre-race analysis, race observation, and post-race recordkeeping. My problem with that approach is that I am limited to a very small number of bettable races each day, and that the amount of time required is substantial.

You are correct that I did not ask the advice of others on your forum. That should not have been necessary. The software is fairly simple, the video seminars are well prepared (if a bit lengthy, considering the material covered), and I think I have studied each issue of the Follow Up at least three or four times. In addition to virtually everything else available on the Sartin Methodology--new, old, or in-between.

Similarly, I did not ask for advice on the forum because I have never found "sample races" to be much more than filler material. The lessons learned, if any, apply only to that race, or an exact replication of that race, neither of which is particularly useful in analyzing future races. An extensive dissection of the thought processes used to select a winner in a specific race is intellectually interesting, but does not help much in analyzing future races, other than pointing out areas that might be worth noting.

The only value in knowing the thought processes of another analyst is contingent on the consistent positive return of that analyst. That, in turn, implies consistency in the decision-making process. It does no good whatsoever to receive advice from someone who selects A in a given race for one reason, then B in another race for another reason, and so on, and the re-works the same race a month later only to select different cointeders, and a different finsih scenario.

If the analyst is consistent, he or she can work a race, then work the same race a month later and reach the same conclusion. If the analyst is consistent, he or she can be modeled, and in short order algorithms can be created that outperform that analyst. With all due respect to all the seat-of-the-pants handicappers in the world, structured decision-making processes will outperform unstructured processes every single time. This is horse racing, not rocket science.

If the analyst is inconsistent, but still wins, he or she is making selections intuitively, and the software readouts are simply accessories. That is, the intuitive analyst makes a superficial examination of the available data, makes a selection, and then searches the readouts for confirmation that his or her selection is a "good" one.

Finally, my question about whether or not your software could create models was not made lightly. I think race models are an invaluable aid in race analysis and wagering. HTR makes pretty models that seem a lot better than they actually are, but are not very useful unless seriously tweaked to cleanse the data. I was unfortunate enough to argue in behalf of short-term "Brohamer" models in a graduate class in descriptive statistics. The professor explained in excruciating detail exactly what the conceptual (as well as practical) flaws were in that approach, and directed me to a very readable explanation in an article by Tversky and Kahnemann called The Law of Small Numbers.

Best Regards,

J2EED,

After reading both your e-mail and the later post, it's still not entirely clear what you expect from RDSS. As Ted says, you may be happier with database software, but you expressed unhappiness with HTR, which is used by at least a few pro handicappers, and is highly regarded.

If you're looking for black-box software, there are none commercially available, although there are some successful proprietary programs of which William Benter's logit model is the most famous and undoubtedly the most successful. All developers of the latter describe having invested years of time and labor in the development of their programs. It took Benter five years and hundreds of thousands of dollars to get his into the black. I think it's easy to understand why these programs they won't be available commercially.

In addition, a number of people with graduate-level training in statistics have described, on the Paceadvantage website, their many failures to come up with a successful model, and the grueling nature of their investigations, despite their knowledge. The dangers of backfitting and multicollinearity are two frequently mentioned issues.

From the tone of your questions, I'd say you somewhat underestimate these difficulties, since only a very few people have succeeded. And if black-box-like performance is what you seek, you'll probably have to develop your own.

Re the Sartin methodology, it sounds as though you always had difficulties with it, so one wonders why you've returned. However if you give it a chance, and follow the instructions, the program works as claimed. In my experience, the top 3 choices win 80% of the races, but as many have noted, the low-hanging fruit of the early years is gone. This is why, as Ted points out, Doc Sartin adopted Dick Mitchell's method of betting only on overlays, a process that he describes in detail in the last dozen or so issues of the Follow Up. In my experience, the wager value of the Sartin pace figures continues to hold up, and although you probably wouldn't believe my own results, I think that an ROI of 15% is within the range of what a moderately successful user can expect.

Apparently Ted is working on a software package that will provide an additional degree of automation for this process, which may be closer to what you're looking for.

Whatever you decide, I wish you the best of luck.

Cheers,

B Jennet