I'm still importing all the data, but I agree that this year has an element of flukiness to it because there was no racing for awhile and it changed the way horses were prepared for the year.
I'm not sure this is correct, but when I want to test a potential insight I usually test it against my entire database. I started it in late 2014. Then I'll
test it against the last 2 years, the last year, and the last 6 months.
I'm trying to make sure I have very large samples and also do my best to make sure that things that may have worked in the past are still working and still have some value. But I can envision some scenarios where I might want to leave 2020 out of the testing just to see the results.
__________________
"Unlearning is the highest form of learning"
|