Projected Improvement: Pace/Speed

[Cesario!]

Hi all.

I've come into a bit of money -- enough to start a bankroll -- so I plan to be a horseplayer professionaly for the summer. As part of this, I'm researching software again. Anyhow, a very specific question:

Can anyone recommend a program (if one exists) that attempts to project improvement in horses, running in "class-step" races, i.e MSW, N1X, N2X, GS. I'm generally thinking younger horses, but improvement (although more form than talent-based) is a concept applicable to older horses as wel. My biggest complaint with most software (and most approaches) is that they are too focused on what has happened in the past and not on what theorectically could happen in the future and the probabilities of different scenarios happening. I think speculative handicappers do this intuitively, but I wouldn't mind automating the process some to be able to hit more tracks. I believe the key to good mutuals (at least what I've found over my handicapping life) is seeing improvement when it's not explicitly apparent from the past data.

If anyone reads the baseball prospectus, it's similar to their project stats and their probality calculations for improve, decline, etc.

[xfile]

Sounds like you want something with a vast amount of AI. Most programs out there are mostly mathematical and/or scientific. IMO you should invest in a custom made program made for your exact specs. Find a good programmer with horse racing knowledge. My guy is tied up on my project probably for the rest of this year on a big AI thing for me. I'm sure a few guys on PA here are programmers with enough skill to write you a custom deal. If you knew the path a horse takes with a pattern for improvement you could artificially inject that criteria into a program. Well anyway good luck with your bankroll and your idea. :cool:

[traynor]

Cesario! wrote: <My biggest complaint with most software (and most approaches) is that they are too focused on what has happened in the past and not on what theorectically could happen in the future and the probabilities of different scenarios happening.>

I tend to agree, except that studying the past realistically can enable you to forecast with increased precision; that is the basis of statistical analysis. The weakness of most software is that they fail to follow even basic statistical procedures in favor of "results" that are essentially unrepeatable. Most importantly, correcting for outliers. Any application that fails to normalize the dataset used in calculating "averages" is not worth betting on.

As for theoretical implications of possible future events, any study implicitly agrees that, a) events are not completely random, and, b) past tendencies can be used to predict possible future events. Take that first big step of culling the outliers, layering samples that are truly representative, and there are dozens of applications that do a credible job of predicting. Most go by the acronyms of ERP or DSS; enterprise resource planning and decision support systems.

If you still want to hire a programmer, find some graduate student who knows absolutely nothing about racing, and give him or her a set of specifications for what you want. Handicappers in general (as well as most statisticians and number crunchers) are hopelessly pedantic and unimaginative when it comes to designing innovative software.
Good Luck

[46zilzal]

really significant PROJECTED improvement only occurs in babies, equipment changes, and surface changes

[Cesario!]

I tend to agree, except that studying the past realistically can enable you to forecast with increased precision; that is the basis of statistical analysis.

I agree with that statement. I would only add that most methods seem to view the future as limited by the past. I think experience would suggest otherwise in terms of improving horses.