This job is mine personal implementation of an existing paper. For a detailed description and references to the original work check the report.pdf file.
- Note 1: Since I am not an expert of financial markets, the environment definition is just a sketch. To train the model for a real use the environment must be redefined. I am open to any suggestion on how to implement a real simulation of a financial market.
- Note 2: This job is just an implementation, with some modifications, of an existing work. The model's parameters are not tuned, they just replicate the paper setting. In my opinion, as stated and motivated in the report.pdf, those parameters are not valid for a well trained Deep Reinforcement Learning agent.
- Note 3: DRL techniques available in the model: Deep Reinforcement Learning, Double Deep Reinforcement Learning, Dueling Deep Reinforcement Learning, Dueling Double Deep Reinforcement Learning.
- Add command line arguments to run custom training/test
- Implement real market simulation