Skip to content

Latest commit

 

History

History
87 lines (65 loc) · 2.75 KB

README.md

File metadata and controls

87 lines (65 loc) · 2.75 KB

ML/DL Crypto Trading strategies

    

GitHub release (latest by date)  python version | 3.10+   Code style: black

Workflows:

Project homepage: lcsrodriguez.github.io/qf/ml

Overview

This academic project follows the below outline:

  1. Personal skill development on crypto-currencies trading and algorithmic trading techniques
  2. State-of-the-art of most common trading strategies on BTC
  3. Implementation of a custom trading strategy on BTC
  4. Backtest of the given trading strategy (on at least 200 trades)
  5. 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.

Architecture

.
├── 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'

Getting started

  1. Clone the repository
git clone git@github.com:lcsrodriguez/cryptotrading.git
cd cryptotrading/
  1. Download the pre-requirements modules
pip3 -r requirements.txt
  1. Execute the Jupyter environment
jupyter-notebook .

License

See LICENSE file

  • Arian NAJAFY ABRANDABADY - Lucas RODRIGUEZ - Bastien TRIDON
  • Academic works (March - May 2023)