A self-hosted, zero-operations quantitative workbench for A-share stock selection, monitoring, and backtesting. It targets individual retail investors and quantitative enthusiasts, offering a customizable platform that integrates seamlessly with various data sources, including TickFlow, Tushare, and user-defined data. The primary benefit is empowering users with a powerful, personalized quantitative analysis tool without the burden of infrastructure management.

How It Works

The system employs a FastAPI backend leveraging Polars for vectorized data computation and DuckDB for efficient querying, with data stored in Parquet format. The frontend is built with React and Vite. Key differentiators include a UI-driven custom signal configuration system, AI-powered strategy generation from natural language prompts (via OpenAI-compatible APIs), and a robust backtesting engine based on vectorbt that incorporates realistic trading constraints like T+1 settlement, fees, slippage, and stop-losses. It supports multiple data sources, allowing users to integrate TickFlow, Tushare, or their own proprietary data into a unified query interface.

Quick Start & Requirements

• Prerequisites: Python ≥ 3.11, Node ≥ 20, uv (latest), pnpm 9.
• Installation:
  • Development Mode: Copy .env.example to .env and configure API keys. Run ./dev.sh (macOS/Linux) or .\dev.ps1 (Windows).
  • Docker: Copy .env.example to .env. Run docker compose up --build. Access at http://localhost:3018.
• Dependencies: Docker requires specific CPU features (AVX2, FMA); a legacy-cpu option is available for older hardware. Backtesting requires optional backtest extras.
• Links: TickFlow data: tickflow.org.

Highlighted Details

• Self-hosted, zero-ops quantitative analysis platform.
• Extensible data integration: TickFlow (free tier available), Tushare, custom data sources.
• AI-driven strategy generation from natural language descriptions.
• UI-based configuration for custom trading signals without coding.
• Vectorized indicator calculations using Polars.
• Realistic backtesting engine (vectorbt) with T+1, fees, slippage, and stop-loss support.
• Real-time monitoring center with customizable rules and SSE push notifications.
• Individual stock analysis page with 9 types of key price levels and AI-driven multi-dimensional analysis.

Maintenance & Community

The project is developed by tickflow-stock-panel contributors. While it mentions being linked and recognized by the "LINUX DO 社区," specific community channels (like Discord or Slack) or detailed roadmap links beyond the README's internal structure are not explicitly provided.

Licensing & Compatibility

The project is released under the MIT License. Users must also adhere to the terms of service for the TickFlow data service, which is a core dependency. The MIT license generally permits commercial use and linking, but specific data usage restrictions may apply based on the TickFlow subscription tier.

Limitations & Caveats

The project explicitly states it is not a full-featured stock software like Tongdaxun/Tongdaxun and does not include AI stock recommendations or price limit predictions. The individual stock analysis feature is currently in Beta. Docker deployment may face issues on older CPUs lacking AVX2/FMA instruction sets without enabling the legacy-cpu runtime option. The free tier of TickFlow data has limitations on the number of instruments and data access.