xtdata

What This Skill Does

The xtdata skill provides a comprehensive interface for the XtQuant library, enabling AI agents to access professional-grade market data directly via a local miniQMT client. It bridges the gap between raw trading infrastructure and intelligent analysis by offering methods to subscribe to real-time market data, retrieve historical K-line data, access Level2 tick data, and fetch fundamental financial metrics. It is the primary data backbone for quantitative analysis tasks performed by OpenClaw agents.

Installation

To use this skill, ensure you have the xtquant Python package installed in your local environment. Run the following command in your terminal: pip install xtquant After installation, ensure your miniQMT application is running, as xtdata requires an active TCP connection to this client to function. In your agent configuration, use clawhub install openclaw/skills/skills/coderwpf/xtdata to integrate the skill into your workflow.

Use Cases

This skill is highly versatile for financial agents. Common use cases include:

1. Quantitative Research: Fetching long-term historical data for backtesting trading strategies.
2. Real-time Monitoring: Subscribing to tick data to trigger instant alerts or automated execution when specific price levels are breached.
3. Portfolio Analysis: Retrieving financial reports and fundamental data to assess the valuation and health of specific stocks or sectors.
4. Data Aggregation: Building customized datasets by downloading large blocks of historical K-lines for offline analysis and machine learning model training.

Example Prompts

1. "Download 1-minute historical data for 600519.SH from January 1st, 2024, to today, and save it for analysis."
2. "Subscribe to real-time tick data for 000001.SZ and alert me if the price changes by more than 2%."
3. "Get the latest financial report data for 600000.SH to help me calculate its current P/E ratio."

Tips & Limitations

• Two-Step Process: Always remember that you must call download_history_data before attempting to get the data. Data resides in the local cache; failing to download first will result in empty returns.
• Rate Limiting: The system is sensitive to excessive concurrent requests. When analyzing large datasets, prioritize batch queries rather than looping through individual stock codes to optimize network performance.
• Dependency: The skill relies entirely on an active local miniQMT process. If the client is closed or the network connection is interrupted, the agent will lose data connectivity. Ensure miniQMT is launched with proper credentials and market permissions (especially for Level2 data).