Hi,
i am using StockPortfolioEnv to simulate portfolio optimization via DRL, with PPO and A2C agents from SB3. My main issue (or doubt?) is that including technical indicators does little to no difference in the algorithms performances. Firstly, i thought that it was related to scale, since I didn't implement any of that. But even after implementing Robust Scaler to both train and trade (and inverting the transformation when calculating the reward, after the action was taken), including or not T.I. doesn't seems to make any difference.
Is this considered to be normal?
Hi, i am using StockPortfolioEnv to simulate portfolio optimization via DRL, with PPO and A2C agents from SB3. My main issue (or doubt?) is that including technical indicators does little to no difference in the algorithms performances. Firstly, i thought that it was related to scale, since I didn't implement any of that. But even after implementing Robust Scaler to both train and trade (and inverting the transformation when calculating the reward, after the action was taken), including or not T.I. doesn't seems to make any difference. Is this considered to be normal?