A Natural-Language Workflow for EDB-Based Quantitative Research
Summary
This guide outlines a workflow that combines TQ EDB market and indicator data services with Coze as a conversational research assistant. After adding the EDB skill, a researcher can describe a task and request data retrieval, idea checks, simple backtests, plots, reports, or reusable Python scripts. It recommends specifying the instrument or indicator, sampling period, date range, required fields, research objective, assumptions, and desired outputs so the work can be reproduced.
Examples include splitting futures returns into overnight and intraday components, charting and examining a metal spread with rolling statistics, and generating a moving-average strategy backtest. The guide also distinguishes free data access from token-required access: daily history and a limited recent span of minute data are available without a token, while extended minute history and indicator data require professional access. These examples are prompts and workflow guidance, not evidence that any resulting analysis is valid. Researchers still need to check data definitions, permissions, execution assumptions, and backtest methodology.
Key ideas
- A conversational research workflow can connect natural-language requests to historical data retrieval and Python analysis.
- Useful prompts specify instruments, sampling frequency, date range, fields, research goal, and requested outputs.
- Example tasks include futures return decomposition, spread analysis, and moving-average backtesting.
- Access to extended minute history and EDB indicator data requires a token.
- Generated code, charts, and conclusions need independent validation and explicit assumptions.
Tags
This summary was written by Stratmill's research agent from the original; it is not a copy of the source.