202412-12-Dynamic-High-Frequency-Trading-Model

Overview

This project implements a dynamic high-frequency trading model using GRU (Gated Recurrent Unit) for predictions and leverages Apache Airflow for workflow orchestration. The project outputs include portfolio values, portfolio positions, an interactive portfolio dashboard, and a trained model file.

Project Structure

202412-12-Dynamic-High-Frequency-Trading-Model/
│
├── dags/                          
│   ├── project_dag.py                 # Airflow DAG script
│
├── notebooks/                     
│   ├── project_gru.ipynb              # GRU model generation notebook
│
├── output/                        
│   ├── data/                      
│   │   ├── portfolio_values.csv       # Example portfolio values data
│   │   ├── portfolio_positions.csv    # Example portfolio positions data
│   │
│   ├── html/                      
│   │   ├── portfolio_dashboard.html   # Example portfolio dashboard visualization
│   │
│   ├── model/                    
│       ├── gru.pth                    # Trained GRU model file
│                 
├── README.md                      
└── requirements.txt               

Setup

1. Clone the repository:

   git clone https://github.com/Sapphirine/202412-12-Dynamic-High-Frequency-Trading-Model.git
   cd 202412-12-Dynamic-High-Frequency-Trading-Model

2. Install dependencies:

   pip install -r requirements.txt

3. Execute the project components:
   • Run the Jupyter Notebook: notebooks/project_gru.ipynb for GRU model generation.
   • Use the Airflow DAG: dags/project_dag.py to automate workflows.

Outputs

• Portfolio Values: Example data stored in output/data/portfolio_values.csv.
• Portfolio Positions: Example details stored in output/data/portfolio_positions.csv.
• Dashboard: Example visualization in output/html/portfolio_dashboard.html.
  • Live Demo: Portfolio Dashboard
• Trained Model: The trained GRU model is saved as output/models/gru.pth.

Video Overview

For a detailed walkthrough of the project, watch the video: Project Video Overview

Dependencies

The project requires the following Python libraries (install only these, as others are part of Python's standard library):

• airflow
• yfinance
• pandas
• numpy
• torch
• pandas_market_calendars
• cvxpy
• pendulum
• plotly
• scipy
• sklearn
• matplotlib

To install all dependencies, use:

pip install -r requirements.txt

Contributors

• Honghao Huang (hh3042)
• Yining Gan (yg2960)

Notes

• Ensure Apache Airflow is installed and properly configured.
• The portfolio_dashboard.html file provides an example of portfolio insights.
• The gru.pth file contains the saved GRU model for reuse.