A Share Stock Analysis

A-Share Stock Analysis

Route A-share market questions through a staged workflow: identify the task, validate data, analyze evidence, then produce decision support with explicit risk controls.

First Rule

Do not load the whole suite by default. Read only the workflow and reference files needed for the current stage. If live market, news, policy, exchange rule, index constituent, or corporate event data is needed, verify it with current sources before relying on it.

If the user asks for value factors, quality factors, financial health, profit trend, cash-flow direction, 1-month main-fund flow, volume ratio, turnover, order imbalance, industry growth, policy support, or volatility risks, load references/five-factor-stock-selection.md with the relevant workflow.

For every individual-company analysis or material holding review, use industry-panorama-cognition when available; otherwise load references/industry-panorama-template.md. Include an 行业全景认知 section unless the user explicitly asks for a very short answer.

Safety Boundary

This skill produces decision support, not guaranteed investment advice. Always separate:

• evidence: observed data and cited sources
• inference: what the evidence may imply
• uncertainty: missing data, conflicting signals, timing risk
• suggestion: watch, hold, add, reduce, avoid, or conditional action
• invalidation: what would prove the thesis wrong

Never present a buy or sell suggestion without risk controls.

Router

If the request spans multiple stages, run them in this order:

1. Data discovery
2. Data validation
3. Market hotspot or market analysis
4. Five-factor review when the request mentions value/quality, basic financial base, capital behavior, growth/policy, or risk avoidance
5. Stock, portfolio, or signal analysis with industry-panorama cognition for each material company
6. Recommendation

Input Contract

Ask for missing inputs only when the answer would otherwise be unsafe or empty. Use reasonable defaults for non-critical details and state them.

Minimum useful inputs:

• analysis date and market session
• market scope or stock tickers
• available files, sources, or APIs
• horizon: intraday, swing, medium-term, long-term
• for holdings: ticker, name, shares or weight, cost basis if available

Output Defaults

Use the user's language. For Chinese users, default to Simplified Chinese. Prefer concise tables plus short interpretation.

Every final analysis should include:

• data coverage and freshness
• core conclusion
• industry panorama cognition for individual stocks or material holdings
• five-factor evidence summary when applicable
• evidence table
• risks and contradictory signals
• next checks
• conditional suggestion with invalidation line

Bundled Tools

• scripts/validate_holdings_csv.py: validate holdings CSV columns and basic numeric fields.
• scripts/validate_market_dataset.py: validate common market dataset fields, dates, duplicates, and missing values.
• scripts/validate_trade_signals.py: validate A-share trade-signal CSV files, duplicate signal IDs, lot sizes, directions, prices, and timestamps.