This project is for portfolio optimisation with RL (Of course can be generalised into other methods - just add your strategy in trainer and execute in main)

The repo is structured as follows:

1. data file represents how data is processed. is one wants to add different ways of analysing and processing data, please write a function there.

2. env file includes basic env for portfolio optimisation with RL. If ones use their own data, please make sure close price for each datum is appended at least. Feature change can be adapted by changing self.observation_space.

3. main file includes the entry for different strategies. By strategies, we mean a specific way of processing data + a env design

4. model the model the agent learns

5. results the return and cumulative returns for the agent in the test set

6. trainer This part is comprised of two parts. In principle, RLalgo should not be modified unless ones wants to add more algorithms. config can be modified as wishes.

7. utils includes all necessities supporting the agent to run.

To run

Please directly run the strategy in the main file