A-Stocks Market Analysis and Investment Advisory

The Specific Problem Being Solved

For engineers or data analysts tracking the A-stock market, manual analysis of individual stocks presents several challenges:
- Data Fragmentation: Information on market trends, financials, news, and sentiment is scattered across various sources (exchange announcements, financial websites, forums), making manual collection and organization time-consuming and inefficient.
- Low Analytical Efficiency: Extracting meaningful insights (like trend judgment or fundamental evaluation) from raw data requires specialized knowledge and substantial computation.
- Information Overload: It is difficult to quickly cross-validate technical indicators with fundamental signals, or to integrate public news with market sentiment.

This skill aims to automate the entire process from data acquisition to structured analysis via natural language interaction, freeing analysts from data wrangling and allowing them to focus on decision-making.

How the Skill Works

Core Capabilities and Data Coverage

The skill retrieves data through a series of /stocks/... API endpoints, covering the following dimensions:
- Market Data: Fetches daily, weekly, and monthly lines, calculates limit up/down and adjustment factors, and queries valuation and trading activity metrics via daily-basic.
- Fundamental Data: Accesses interfaces like fina-indicator, income, and cashflow to obtain financial metrics, three major statements, dividend records (dividend), and shareholder structure (top10-holders).
- Market Environment: Analyzes the performance of major indices and sectors through index-daily, index-member, and index-weight.
- Textual Information: Leverages the LLM's search capabilities to retrieve irm-qa-sh/sz (board secretary Q&A), research-report (research report metadata), and conducts web searches to gather relevant financial news and forum discussions.

Key Processing Steps

1. Intent Parsing and Routing: The skill is automatically triggered when it detects a stock code (e.g., 600000 or 600000.SH), name, or related analytical terms. The user must provide a clear query target (code/name) and time range.
2. Layered Data Acquisition and Analysis: By default, it executes a multi-step analysis as sub-agents:
   - Data Layer: Calls APIs to fetch as much raw trading and financial data as possible.
   - Market Layer: Analyzes trends and volatility from daily/weekly/monthly data.
   - Fundamentals Layer: Verifies company quality and performance through fina-indicator and similar interfaces.
   - Environment Layer: Cross-validates using index data (index-*), corporate interaction information (irm-qa-*), and research reports (research-report).
   - News & Information Layer: Searches and summarizes the latest news and forum information related to the target.
   - Note: If data is insufficient at any step, the analysis conclusion will explicitly state that "the confidence of the current conclusion is limited."
3. LLM-Integrated Analysis and Report Generation: The collected structured data and prompt information are fed into the LLM for comprehensive analysis. The output is a structured report containing:
   - Data Overview: The queried target, time range, and core metrics.
   - Market Scan: A brief report on the sector's recent market performance.
   - Fundamental Analysis: An assessment of financial health.
   - Related Information & Market Sentiment: A summary of news and forum information, plus the LLM's interpretation of market trading sentiment.
   - Observational Insights & Consistency Risk: Key findings and whether data points contradict each other.
   - Suggested Next Queries: Provides 1-3 interfaces or questions that can be further verified.

Applicable Boundaries and Important Notes

• Data Limitations: It only provides queries on publicly available historical data and does not guarantee real-time accuracy. All data originates from third-party APIs.
• Not Investment Advice: The skill strictly adheres to risk control rules. It does not provide any form of trading execution, return promises, or "guaranteed rise" conclusions. Any output involving investment advice is accompanied by risk disclaimers and a statement that it does not constitute investment advice.
• Privacy and Security: The skill is designed not to echo, solicit, store, or output the user's TAX_API_KEY or any personal sensitive information.
• Query Failure Handling: When API errors occur (e.g., 401/403 authentication issues, 400 parameter errors, 502 service anomalies), specific remediation suggestions are provided (e.g., checking the ts_code format, confirming key status).