An example of Reinforcement Trading using OpenAI Baseline.
This is made by Adrian Portabales based on Mr. Peter Henry and inspired by Tito Ingargiola, hackthemarket.
There are some improvements in code:
4 States:
The bot only operates with one open position at the same time.
At the end of every episode, a test is run. The size of train and test set in train_split
variable
Now you can add more features (like indicators, momentums etc) with feature_engineering module.
https://github.com/AdrianP-/gym_trading/blob/master/Baseline%20DQN%20Gym_Trading%20Tutorial.ipynb
OpenAI Baseline: https://github.com/openai/baselines
Gym_trading by Peter Henry: https://github.com/Henry-bee/gym_trading/
Also, good thougths in Trading and Q learning: https://github.com/savourylie/Stock-Price-Forecaster
Thanks to all!
Requirements: -Pandas -Matplotlib -Numpy -gym -TA-Lib -baselines
This framework was built in Python 3.5.2
Twitter: @porta4k
Mail: [email protected]