raphael at Mon, 18 Nov 2024 20:51:53 UTC No. 16481234

trade spx and qqq

raphael at Mon, 18 Nov 2024 20:52:47 UTC No. 16481236

>>16481234
not spx but s&p futures and nas100 i havent backtested it but use all learning rates and simulate brute force every value between .01 -> 3 then extract whatever has an infinite sharpe

raphael at Mon, 18 Nov 2024 23:40:04 UTC No. 16481440

i guess none of you care

Anonymous at Tue, 19 Nov 2024 02:35:23 UTC No. 16481613

your post is interesting but explain in detail what you mean by q learning with regression lines.
q learning i know but where do the regression lines come from? also, do they train a different model for each asset or what.
how much data do they need to get useful forecasts?
what witchcraft
etc

Raphael at Tue, 19 Nov 2024 03:01:54 UTC No. 16481631

>>16481613
I’m smart as shit nigga don’t test me

Anonymous at Tue, 19 Nov 2024 04:28:10 UTC No. 16481717

>>16481631
do you want to share your findings or not

raphael at Tue, 19 Nov 2024 04:32:19 UTC No. 16481724

>>16481717
shut up nigger i gave sauce

Anonymous at Tue, 19 Nov 2024 07:23:02 UTC No. 16481845

I've made regressor I fed with TA and it could trade crypto in 2019, but now it doesn't get it.

Anonymous at Tue, 19 Nov 2024 07:28:15 UTC No. 16481847

>>16481845
I had problems with reward model for Q learning, I've made gym environment, but there are multiple points, or totally narrow point for entering in trade... It fit to no trading at all, and I couldn't make it into trading, so I just used NN later that have output of three (BUY,SELL, DO NOTHING)

Anonymous at Tue, 19 Nov 2024 07:34:37 UTC No. 16481854

If you had any way I don't yet know of vectorizing trade data for feeding into machine, please go on and explain.