deep-RL-trading
deep-RL-trading copied to clipboard
Replay Memory
Your replay memory only stores samples one by one. Since you're also training RNNs, you should use sequntial training samples. This paper by Hausknecht and Stone explains what I mean.