Modeling

[chuck]

I'd like some opinions on this.

Is it better to use short term or long term models? By short term I mean like the last few weeks. Long term the whole meet or prior meets. I did some work and compared 2 whole meets for 2 consecutive years and devoloped spot plays only to find some held up and others did not. It gets back to the same question how much data is enough?

Anybody using track/surf/dist specific models successfully?

[hurrikane]

I use the modeler to develope most every spot play I have...with exception to the trainer plays.

not just track/surf/dis but also type -2/3/4 and older- and SEX (especially with the jockies)!

(sorry, I always seem to overemphasis that one.):D

[chuck]

hurrikane,

Do you short term trends or do you use as much data as you can get?

[hurrikane]

Chuck,
Initially I use long term but will cut back to 10-25 races to maybe catch a trend(little trick Glen uses)...worked real well at CD with jockeys and at FG with MLO....you just have to be ready for the trend to stop.

with spots..when I find something useful I'll run the data set without the last month of data and then check it with just the last month to see what would have happened had I bet it for 30 days and then I start slow and sometimes even just track it if I dont' feel comfortable.

There are probably some more seasoned players up here that can offer other advise...most of what I got I got from them...and of course paying dues.

[Glen]

Regarding the Modeler, I really enjoy looking at the Records. It enables you to look at all horses who have won and what their characteristics are. Both long and short term.

Regarding how much info is enuf? Dunno...But don't think there is a right answer...Say we have 10 years of data. Now does that mean what was working 8 9 or 10 years ago will work today? With the public perhaps changing their betting trends over the years, I think not.