Crypto Trader Classifier with Logistic Regression Overview

Welcome to the repository for our project on classifying cryptocurrency traders using logistic regression. This tool aims to categorize traders into three groups: Good Trader, Average Trader, and Bad Trader, based on their trading performance metrics.

Features Used

• Base Cumulative Return: Total return over time.
• Portfolio Return: Percentage return on the entire portfolio.
• Daily Sharpe Ratio: Risk-adjusted performance measure.
• Number of Trades: Total trades executed.
• Unique Tokens Traded: Diversity of assets traded.

Methodology Data Collection: Gathered using Flipside Crypto's SQL terminal, focusing on traders with volumes over $10 million. Analysis and Modeling: Utilized Python for:

• Statistical analysis
• Feature engineering
• Model training with logistic regression

Key Findings Correlation Analysis: Identified Portfolio Return (0.2718) and Base Cumulative Return (0.2387) as the most influential features. Model Performance: Achieved an accuracy of 97.87% in trader classification.

Project Structure

• /data: Contains CSV files with raw and processed trading data.
• /src: Source code for data analysis, model training, and testing.
• /api: Flask application for model predictions.
• Dockerfile: For containerizing the application.
• requirements.txt: Python dependencies.

How to Use Clone the Repository:

• bash
• git clone https://github.com/apostleoffinance/web3_ml.git
• cd web3_ml

Setup Environment:

• bash
• pip install -r requirements.txt

Run the Model:

Locally:

• bash
• python src/train_model.py

With Docker:

• bash
• docker build -t classifier-test .
• docker run -p 8000:8000 classifier-test

API Usage: After running the Flask server, you can make POST requests to /predict with trader data in JSON format to get classifications.

Applications

• Copy-Trading: Mimic strategies of high-performing traders.
• Risk Management: Avoid strategies of underperforming traders.
• Portfolio Optimization: Use data-driven insights for better asset allocation.

Technologies

• Machine Learning: Python, scikit-learn
• API Development: Flask
• Containerization: Docker
• Version Control: GitHub

Further Reading

Check out the Flipside Crypto Dashboard for the SQL queries used in this project: Flipside Dashboard

Contribution

Feel free to fork this project, make your contributions, and submit pull requests. Any improvements and ideas are welcome!

License

This project is open-sourced under the MIT License (LICENSE).

Thank you for visiting this repo, and happy trading analysis!