Quote:
|
Originally Posted by Vortech
Thanks for all your feedback Aussie!
When you have a database going back to 12 years and approximately 60 filters there requires a lot of time and patience with your testing.
In relation to variables do you try to initially test against a particular horse applied against a particular race.
For example certain variables will remove certain races. Call them Variable A
Then you have certain variables that will remove certain horses. Call them Variable B.
Examples of variable A include - Min. LS winners, Day of the week, Track conditions, Prizemoney of Race, Class. All of which can remove certain races from a program
Then variable B will include your typical - Barrier, Weight etc.......
Should one be starting to test with one Variable A aganist Variable B?
Where does one start when they have so much information!!!
|
yes i know what you mean.
my method is to test for a particular horse statistic first.
for example last week I was looking at how the top 4 in average prize money fared:
the results were Strike Rate: 21%,15,12,10.
the top ranked having a -14% POT on TATTSBET odds.
place % 1..4 was 20%,15,12,10: rank 1 -15% POT.
I had broken the DB into 2 and it was consistent over both data bases
whereas another ranking 1..4 was : 16,13,12,11 but POT
-7%.
that last one would be where i start:
Thus my data base is set up so that it captures information for a particular starter and it at the same time collects information for the races that it is in.
as per the above I then look across the filters and see if there is anything that sticks out as bad or good - that can be a horse statistic such as place % or it can be a race characteristic such as average race prize money , average race rating etc, number of last start winners etc.
and so on it goes.
cheers
aussie