tf_deep_rl_trader
tf_deep_rl_trader copied to clipboard
How can I improve it ?
I can duplicate your result of episode 100 (1000000 -> 1586872.177567203 ), but I try to episode 200,300,500... , the result became worse , it make a loss . My question is 1.How improve it , more samples , feature , episode , window_size or other? 2.How can I check the algorithmic is correct and bug free ?
if you watch the result closely the result, it shows that their was no order opened. ./info/ppo_1586872.177567203_LS_0_0.info 0 buy and 0 sell. If you try ppo, you may need lots of change to make it work. I had better results when i tried dqn. Cause at least i could open trades in a test set. Also you should put your own dataset on the test and train directory. Of course this is not bug free, but its a very good baseline if you are serious about auto trading strategy.
@greg2paris could you help me get this set up? like a very basic step by step, ive been running into problems every step trying to get this to run