virtualenv venv
source venv/bin/activate
pip install -r requirements.txt
tf_rrl.py
contains the code for running the recurrent reinforcement learning algorithm using a deep neural network, an LSTM, or a simple RNN. You can also pick the optimizer of your choice (SGD or Adam), set hyper-parameters s.a. transaction costs, learning rate, number of epochs, etc.
tradingrrl.py
contains the code for running the layered system i.e. the RRL algorithm using a simple RNN trained using gradient descent with the risk management and the dynamic optimization layers.
utils
contains multiple files mostly to clean and download the dataset.