Rick,

The model uses TG figs and looks at patterns. It also looks for horses that most closely meet the profile of a Derby winner (the TG seminar covers this very well) with a fast (but not too fast) race late in the 2yo year followed by steady development at 3. It also looks at running style (front-runners are up against it in the Derby) post position(accounts for expected ground loss) and breeding as well as some other factors such as number, spacing and distance of 3yo preps. The factors are statistically weighted based on analysis of past Derbies and then the Derby is "simulated" thousands of times to see how often each horse would win (referred to as Monte Carlo modeling). According to the computer, Perfect Drift would win 29% of the time (and lose 71% of the time). So at his 15-1 morning line, the computer says he is a good value play.

As you said, the computer is only as good as the programmer and the data. The key source of data is the TG figures (so you can blame TG if the data is bad) and I am probably not capable of offering an unbiased opinion of the programmer.

I Hope that helps and good luck in the Derby.

"Computers are fantastic: in a few minutes they can make a mistake so great that it would take many men many months to equal it." Unknown.

Chris