Workflows:
Project homepage: lcsrodriguez.github.io/qf/ml
This academic project follows the below outline:
- Personal skill development on crypto-currencies trading and algorithmic trading techniques
- State-of-the-art of most common trading strategies on BTC
- Implementation of a custom trading strategy on BTC
- Backtest of the given trading strategy (on at least 200 trades)
- Use of transfer learning to extend our strategy on other crypto-currencies
For ML/DL, we have used XGBoost and LSTM models to forecast future market movements (binary classification).
Academic project with financial datasets based on G-Research Crypto Forecasting Kaggle competition.
.
├── README.md
├── assets
│ ├── README.md
│ ├── csv
│ └── parquet
├── data
│ ├── asset_details.csv
│ ├── example_test.csv
│ ├── supplemental_train.csv
│ └── train.csv
├── main.ipynb
├── notebooks
│ ├── README.md
│ ├── bitcoin_trading.ipynb
│ ├── eda_visualization.ipynb
│ ├── ethereum_trading.ipynb
│ ├── models
│ └── processing.ipynb
├── out
│ ├── README.md
│ ├── backtests
│ └── models
├── requirements.txt
└── src
├── Backtesting.py
├── Strategies.py
└── Utils.py
To reproduce on local machine the file architecture, please run:
tree -L 2 -I 'site|*__|img'
- Clone the repository
git clone git@github.com:lcsrodriguez/cryptotrading.git
cd cryptotrading/- Download the pre-requirements modules
pip3 -r requirements.txt- Execute the Jupyter environment
jupyter-notebook .
- Arian NAJAFY ABRANDABADY - Lucas RODRIGUEZ - Bastien TRIDON
- Academic works (March - May 2023)