Class Handicapping

[philcski]

Originally Posted by the_fat_man

Rather than fighting change and trying to get a reaction out of me, you might want to consider a suggestion:

Watching replays intently at KEE is pretty much a waste of valuable time. This is because with TRAKUS you get a more precise sense of how the race was run. The idea, is to GENERALIZE. I'm not so much concerned as to what happened in a particular race to a particular horse as much as I'm interested in how a particular angle/situation generalizes. Once I'm at this point, i.e., have formulated a method of identifying 'mismatches', I can then play multiple tracks. The idea behind winning today is to find these prime plays, and as many of them as you possibly can. It allows you to not force plays.

If I have an automated pace/speed system in place, I can easily spot mismatches at multiple tracks. Why in the world would I want to spend hours watching races when I can see from a chart who ran against the grain and who didn't?

As for how a track is playing, the charts instantly show this. I don't buy into all the modeling of tracks BS; if this was of any value, then all the Sartinistas would be counting their winnings rather than spending even more money on yet another program.

The idea, broadly, is to model in a way that allows you to play ANY track, without preparation, without specific modeling, without figures, etc. If I'm spending more than 5 minutes to handicap a race, then I'm wasting valuable time.

Obviously that would be the truth, the light, an ideal situation- but I've in fact tried modeling and backtesting it and it doesn't work. It identifies too many false positives, especially on synthetic. For example, from today's 7th at KEE, Bigshot would have been identified by a computer model as an absolute gold PRIME play because of his closing performance against the strong speed bias on the 8th in the Alcomatch race- which a human analysis would conclude as skeptical at best because it fell so far outside of the rest of the day. Likewise, being able to toss his performance and reanalyzing without that he was an easy toss at 5/2.

[the_fat_man]

Originally Posted by philcski

For example, from today's 7th at KEE, Bigshot would have been identified by a computer model as an absolute gold PRIME play because of his closing performance against the strong speed bias on the 8th in the Alcomatch race- which a human analysis would conclude as skeptical at best because it fell so far outside of the rest of the day. Likewise, being able to toss his performance and reanalyzing without that he was an easy toss at 5/2.

Bigshots' race, per the non Trakus charts, is not nearly against the grain enough to warrant a look; certainly not at 5:2. Moreover, he's never run a race in his history where he's run against the grain (or done anything of note in terms of his setups). This is key: I want horses that run well against the setup not those that plug along when they don't get one and run only when things go their way. A chart history, while flawed (as all data presently is) goes a long way towards pointing them out. (Another way would be to watch all their races and keep copious notes.)

And, it's become evident to me that even the strongest bias is not immune to pace/setup. Even if there were a speed bias on 4/08, race 6 was not run as such.

[philcski]

Originally Posted by the_fat_man

Bigshots' race, per the non Trakus charts, is not nearly against the grain enough to warrant a look; certainly not at 5:2. Moreover, he's never run a race in his history where he's run against the grain (or done anything of note in terms of his setups). This is key: I want horses that run well against the setup not those that plug along when they don't get one and run only when things go their way. A chart history, while flawed (as all data presently is) goes a long way towards pointing them out. (Another way would be to watch all their races and keep copious notes.)

And, it's become evident to me that even the strongest bias is not immune to pace/setup. Even if there were a speed bias on 4/08, race 6 was not run as such.

We agree on this. While the pace was moderately fast for the level, it still shouldn't have fallen apart like it did.

However, I don't know how you can say this horse wouldn't be identified on a computer-based system as a stickout; he's exactly what you're trying to find, closing from well back on a day where very little passing occurred. I can understand not wanting to get involved at 5/2 but then you are right back to square one- needing extensive human analysis to finalize your wagering decisions. There is no substitute for hard work in this game- be it making quality speed figures, watching races, creating customized charts, or building a database of troubled trips. It is a lot closer to poker than blackjack in that artificial intelligence can only do part of the work instead of all of it.