TSP is a self-hosted quantitative trading workbench for Chinese A-stock screening, monitoring, and backtesting.
The tool addresses fragmentation in quantitative workflows where stock selection, backtesting, and monitoring typically require separate systems with misaligned data conventions. TSP unifies these functions around a single enriched data format that flows from screening through backtesting to monitoring and review. It routes data requests across six dataset categories to whichever source has the required capability, allowing users to switch data providers without rewriting logic or changing calculation conventions. The system executes strategies at minute-level granularity and monitors for anomalies across three market phases: pre-market, intraday, and post-market, with real-time alerts via desktop notifications, voice announcements, and Feishu integration.
Developers building personal quantitative research systems should consider this tool if they want to avoid the operational overhead of cloud platforms while keeping data local. The project suits individual traders and quant researchers who need to iterate on factor definitions and strategy logic without vendor lock-in. It includes twenty-five built-in screening strategies, a DSL for custom factor definition, backtesting with transaction costs and slippage modeling, and factor attribution analysis. The tool explicitly positions itself as a research platform rather than a stock recommendation or price prediction system, and it does not attempt to replicate commercial terminals like Tongdaxin or Flush Finance.
Development is maintained as a personal open-source project with plugin-based data source integration. The maintainer is reachable via email for project issues and welcomes community engagement through GitHub stars.
