Project Summary

Overview

This project aims to build a robust and professional algorithmic trading bot using NautilusTrader and Interactive Brokers (IB). The setup showcases modern software engineering best practices, emphasizing maintainability, cost-efficiency, and scalability. The bot will be deployed in two different Google Cloud Platform (GCP) setups for cost and execution comparisons:

1. Serverless Container Execution (Cloud Run): Scheduled daily execution using Docker containers for rapid startup, optimal cost efficiency, and minimal runtime.
2. Continuous VM Deployment (Compute Engine): A virtual machine running continuously (24/7) to assess the trade-off between execution flexibility and ongoing costs.

Project Structure

nautilus_trading_bot/
├── README.md                   # Project overview and setup instructions
├── .gitignore                  # Git ignore rules
├── pyproject.toml              # Dependency management (uv)
├── uv.lock                     # Locked dependencies
├── Dockerfile                  # Docker container definition
├── .github/
│   └── workflows/
│       └── ci_cd.yml           # GitHub Actions Continuous Integration
├── scripts/
│   └── startup.sh              # VM startup automation script
└── src/
    ├── main.py                 # Entry point for trading bot
    ├── config/                 # Environment-specific configs
    ├── data_ingestion/         # Market data retrieval and preprocessing
    ├── strategies/             # Algorithmic trading strategies
    ├── execution/              # Order placement and risk management
    ├── backtesting/            # Historical simulation and evaluation
    └── utils/                  # Common reusable components

Technologies & Services

• NautilusTrader: Core trading platform
• Interactive Brokers (IB): Broker API integration
• Google Cloud Platform:
  • Compute Engine (VM Instance)
  • Cloud Run (Serverless Docker)
  • Cloud Scheduler (Task Scheduling)
• Docker: Containerized environment
• GitHub Actions: Continuous Integration & Deployment (CI/CD)
• uv (Astral): Python dependency management

Deployment Strategies

• Cloud Run (Serverless):

  • Minimal costs; only pay for container runtime.
  • Daily scheduled task execution.

• Compute Engine (VM):

  • 24/7 runtime; greater responsiveness but ongoing cost implications.
  • Automated VM management via startup/shutdown scripts.

Future Enhancements

• Integrate MLFlow for experiment tracking.
• Expand to multiple concurrent bots for diversified strategies.
• Explore infrastructure automation options.

This approach highlights cost-effective resource utilization, best DevOps practices, and provides a foundation for scalable and sophisticated trading strategies.