A Stock Daily Market Sense

This Skill addresses the inefficiency and subjectivity in post-market review for A-share investors. In practice, analysts manually aggregate data on indices, turnover, market styles, and main themes, which is time-consuming and prone to missing key signals. This Skill automates the data pipeline: it first generates a deterministic evidence package, then the model writes the report module by module, ensuring every conclusion is backed by turnover evidence.

Usage is straightforward: simply specify a trading date (e.g., "today" or a specific date). The Skill then runs scripts/run_daily_panel.py to pull Tushare daily data, indices, turnover, and local sentiment history, producing a complete evidence JSON and module-level data. Subsequently, the model sequentially completes sections on market trend, turnover concentration, profit effect and leading themes, high-volume decline risks, and feature group analysis, finally aggregating them into a structured research report. It supports quantitative backtesting of feature groups and relative return factor mining, but these require separate triggers.

It is well-suited for quantitative traders, data analysts, and investment researchers who need daily A-share post-market reviews. Particularly for teams already using Tushare Pro or Baostock data sources and looking to standardize their post-market analysis workflow. For scenarios requiring quick identification of market style shifts, theme confirmation, and risk signals, this Skill can significantly reduce manual data processing time.

It is recommended to use this tool uniformly after each market close to generate reports, and to track theme evolution using a theme lifecycle ledger. Note: It does not provide buy/sell recommendations, does not apply pre-defined industry/concept classifications, and theme ratings are locked by price-volume evidence and cannot be altered by external news. The script handles data fetching and calculations, while theme summarization and risk wording are completed by the model; therefore, model quality directly impacts report readability.