A-Share Fundamental Indicator Analysis

Problem

Fundamental analysis for A-shares often breaks down when metrics are computed with inconsistent periods: ROE, operating cash flow, and PEG can mix single-quarter, cumulative, and TTM values, or accidentally use announcement dates from the future when backfilling trading days. This skill turns that calculation into a reproducible script that covers 90 indicators across 9 categories, including profitability, growth, efficiency, earnings quality, safety, governance, valuation, shareholders, and size.

How it works

The skill runs scripts/run_fundamental_analysis.py through Bash and selects an indicator set with a category argument such as profitability, growth, efficiency, valuation, or all. Its core logic handles several engineering details:

• Quarterly fundamentals: uses consolidated statements and matches REPORTING_PERIOD for YoY, QoQ, TTM, and single-quarter calculations.
• Safe arithmetic: division uses safe_div to return NaN when the denominator is 0 or NaN; change indicators use _safe_diff to require a 75–110 day gap between adjacent reporting periods.
• Daily indicators: valuation, shareholder, and size metrics are generated from trading calendars, closing prices, equity structure, dividends, holder counts, and top-holder data, with point-in-time filling to avoid lookahead bias.
• Data conventions: total shares are converted from AmazingData’s 10,000-share unit to market value; only consolidated statements with STATEMENT_TYPE='1' are used, and the latest record is selected for duplicate reporting periods.

Scope and limits

This is best for generating structured indicator series from financial and market data, not for narrative event study or qualitative analysis. It assumes a configured Python environment, AmazingData dependencies, and authentication variables. Some metrics require dividend, holder-count, and top-holder APIs, and missing fields may produce NaN. AmazingData enforces single active login per account, so concurrent runs may conflict.