This repository contains step by step process training RL for trading cryptocurrencies. Please note that most of this project is currently being migrated from private repo to here. There are three main notebooks:
- Data – Generating data for testing via binance download or sythethic data.
- Train – Related to model training with varying training method
- Test – Validate model accuracy
To use these notebooks, you will need:
- Python 3.10 (or later)
- Jupyter Notebook
- Stablebaselines
- Tensorflow
Additionally, make sure to clone the Binance public data repository into this folder:
git clone https://github.com/binance/binance-public-data.gitTo find a trading setup using reinforcement learning
Book 1 - Data
- Migrate sinewave script from private repo
- Generate complete data traing set: sinewave, sideways, real data
Book 2 - Train
- Complete training script for each training data
- Display tensorboard graphs on notebook (fitness level)
Book 3 - Test
- Create evaluate script