Quote:
|
Originally Posted by pandy
Well, these handicappers I've mentioned who believed in speed figures did search for ways to improve the performance of the figures in specific types of races, because they interpreted the speed figure differently. For instance, in a maiden race, I often project that a horse is ready to run a much better figure than it's run before.
The thing is, I was speaking purely of picking winners. This one guy I mentioned, who made a small fortune alone just by selling his picks because he was so good, no sophisticated computer program or any modern handicapping development I've ever seen could pick more winners than him.
|
It is the interpretation part that is difficult--not generating the numbers. It is possible (as in being done every day) to code algorithms that reverse-engineer the strategies used by "very good" handicappers to determine how "believable" a number is. If it is "believable" it can be used as part (not all, but part) of the final decision process. That final decision process can also be (as in being done every day) automated almost entirely.
Consider how many on this forum have dabbled in "picking the proper pace line(s)" or "finding the most representative race" for predicting performance. And how many have fiddled around with those pacelines enough to realize that--even when carefully selected--they can "predict" differently when the best recent is used, the average of best two of the last five is used, best recent at this distance is used, and on and on and on.
I wrote a training app (for a blackjack team) to automate the pattern recognition part of "finding predictive pace lines." It was a lot of work, but a good learning experience. It made crystal clear that the number affixed to a horse's race--
unless viewed in the context of its performance in that race--was "not very uiseful."
