model research

[ronnie h]

can some of you explain how you use the modler myself i break it down to see what factors are working for the type of race at the current meet .so far i have had good succes with it at keeneland

[Neil]

I use it for three things:
1. see what sort of races are playable (e.g., can you beat the m/l?)
2. see what factors, if any, make it playable
3. see if any factors turn up regularly as longshot hints

As for Keeneland, I have 3 race meetings in my database rather than just the current meeting.

Neil

[chuck]

Neil,

Do you find that the same factors hold up meet to meet?

[Neil]

Chuck,
Not exactly. But I do find that the same types of races are "playable" (in the sense you can get a price).

I actually have two years of Kee in my database, and I find that it is pretty futile looking for w-w winners on the turf over 8.0f (males, anyway). For fillies at 9f, only 1 out of 19 winners had a p-scan variant greater than -2.6. Perhaps even more surprising, since Kee has the reputation as a speed favoring track, for male alw runners at 8.5f on the dirt, only 1 F* winner. However, check this out: last spring, every single 8.5 f alw (F&M) winner had a P-scan 1 rating! 6 of 6. This year has not been so predictable, but 3 of 6 were P-scan 1 and another p-scan 2. The other two had a velocity of 1 and 2... perhaps supporting the idea that speed is holding just a little better this Spring. Some years are just wetter than others (even though all these stats are from a fast track).

So I think there is value in keeping multiple years in the database.

[Neil]

I had another look at those F&M stk/alw 8.5f dirt races. Only one P-scan 1 filly has finished otm in these 8.5f dirt races in the two spring meets for which I have records (Siren Song, unraced since Jan). Two finished 2nd this year (Two Dot Slew and Golden Ballet, both as favorites). About as predictive a single factor as I've found so far.

Neil