Genetic vs Random
5 replies
Threshold
8 years ago #113644
Does genetic make strategies that random gen cannot?
Threshold
8 years ago #130088
Thanks I wanted to be sure. I’m not a programmer but this has been my thought for 1+ year but I never actually asked to be 100% sure. Personally all my best strategies have actually come from random generation and perhaps that is because all my filters are designed to already produce near-finalized strategies, but I use genetic on optimization, where I will run it several times.
I believe this process actually saves even more time. Generally only need to run gen for a couple days/500,000 tests to get some nice candidates, then its on to retest, then optimization if retests showed promise. (or perhaps a slight de-optimization by rounding off the parameters).
mikeyc
8 years ago #130097
test
tnickel
8 years ago #130110
I think genetic strategy generation is better. But if finds more simmilar strategies.
thomas
https://monitortool.jimdofree.com/
Threshold
8 years ago #130111
Yes, its best run a few times with several populations. I tend to find when using it on optimization it will produce nearly identical optimizations, then when I run it again, it has new ones. It has positive and negative side. Once it finds a “good DNA” it will tend to focus around that. Brute force keeps generating anything and a variety.
geektrader
8 years ago #130158
I use both and let it stop if not fitness improvement after 3 generations so the next random generation can begin. I have some strategies that just came out good from the genetic evolution (you can see this by the strategy number). Some of the random candidates wouldn´t have made it through my filter if not improved by genetic evolution – yet those are still very simple strategies, which is great. So using both seems the way to go for me.
Viewing 5 replies - 1 through 5 (of 5 total)