muou55555 / back-tester

pet-project for quant trading self-training

Back Tester

Status

This is a personal project and it's currently under development as part of my self-study efforts. The project is continually evolving as I learn and incorporate more features and improvements. Feel free to explore, learn from it, or provide suggestions!

---

The project has been paused as I have shifted my focus into C++ now. I might re-implement this via C++.

About

Pet-project for quant trading self-training.

Key Features

• Event-Driven Architecture: The backtester operates on an event-driven system that accurately simulates the order of market events, providing a strong foundation for concurrent processing.
• Asynchronous Programming: The backtester uses asyncio for writing single-threaded concurrent code, which allows for efficient multiplexing of I/O access over various resources.
• Data Handling with Polars: Leveraging the power of Polars, a lightning-fast DataFrame library implemented in Rust, for efficient and effective data handling.
• Flexible and Powerful: The backtester is designed to be adaptable, allowing for the testing of a wide variety of trading strategies, from simple moving average cross-overs to complex machine learning-based approaches.

Main Components

• Data Feed: The data feed component is responsible for providing the backtester with market data. This can be historical data for backtesting purposes or live data for paper trading.
• Strategy: This component is where you define your trading strategy. It processes market data and generates trading signals.
• Execution Handler: The execution handler simulates order execution. It takes trading signals from the strategy component and simulates the execution of the corresponding orders.
• Performance Tracking: The backtester includes functionality for tracking the performance of a strategy over the course of a backtest. This includes calculating various performance metrics and generating performance charts.
• Risk Management : involving setting limits on the total exposure to any one asset, ensuring that the portfolio is sufficiently diversified, rebalancing the portfolio periodically, etc.

Future Development

• Polars as data handller
• Redis as historical data & store frequently used data (e.g. like the latest prices or indicators calculated over a rolling window)
• Restful API to fetch real-time or historical data from an external source

Strucutre

Inspiration

This project was inspired by FunCoder. I'm grateful for the ideas and knowledge gained from his work which served as a foundation.

Disclaimer

Please note that this project is primarily for my personal learning and exploration. As such, it may not be regularly maintained or updated. I welcome any constructive feedback or suggestions for improvement!