TraderNet-CRv2 - Combining Deep Reinforcement Learning with Technical Analysis and Trend Monitoring on Cryptocurrency Markets
TraderNet-CRv2 - Combining Deep Reinforcement Learning with Technical Analysis and Trend Monitoring on Cryptocurrency Markets
This system architecture is an extended version of the original TraderNet-CR architecture, which is described by this paper: https://link.springer.com/chapter/10.1007/978-3-031-08333-4_25. In this work, we combine Proximal Policy Optimization algorithm (PPO), which is a DRL learning algorithm, with 2 rule-based safety mechanisms: N-Consecutive & Smurfing. Our experiments on 5 popular cryptocurrencies show very promising results.
Technical analysis has been applied on market data in order to train TraderNet. The following popular technical indicators have been used:
To run and evaluate our agent, You need to download the following libraries/packages:
Download Python 3.6 or higher and the libraries that are described on requirements using pip
installer (e.g. pip install numpy)
. Then:
download_datasets.py
to download the datasets from CoinAPI platform (https://www.coinapi.io/).train_tradernet.ipynb
to train TraderNet module.train_smurf.ipynb
to train Smurf module.integrated.ipynb
to evaluate the Integrated agent.This AI is not a commercial product and is intended for research purposes only.