Financial Report Teardown

What problem it solves

When reading listed-company reports, the usual failure mode is mixing press summaries, consensus forecasts, and disclosed figures into one narrative. Gross margin, FCF, share count, and segment revenue can look like facts even when they are paraphrased or estimated. 财报拆解 targets quarterly, semiannual, annual, and preliminary earnings reports, requiring primary data from official IR PDFs, exchange filings, or SEC documents, then producing a fixed 12-section teardown note that is easier to compare across periods and companies.

How it works

• Scope check: lock the company, period, report type, and currency first; for fiscal years that differ from calendar years, mark the FY label and calendar range to avoid treating an old annual report as the latest filing.
• Primary data extraction: pull two-period segment revenue, income statement items, and all three cash flow sections, especially financing activity signals such as buybacks, dividends, debt issuance, and preferred stock.
• Self-calculated validation: use scripts/calc_teardown_metrics.py to compute FCF, gross margin, segment rollups, net income / EPS share-count checks, and YoY growth instead of relying on mental math; self-calculated values must include formulas, while disclosed values remain separately labeled.
• One-off item handling: adjust polluted metrics for large impairments, sale gains, or mark-to-market gains; for A-shares prefer non-GAAP net profit attributable to shareholders, while using adjusted or Non-GAAP figures for HK and US filings, with a multi-basis comparison table.
• Data gaps: if exact numbers are unavailable, mark them as N/A or source not covered rather than filling with estimates, forecasts, or UGC content.

Limits and cautions

The output is an objective analysis, not buy/sell advice, target prices, or position sizing. It is useful for engineers, researchers, and investment research workflows that need a structured working paper, but it should not replace compliance-approved investment opinions. When using A-share, HK, or US data, be careful about accounting standards, currency, fiscal calendars, and disclosure conventions; if API sources are configured, use at least two sources for cross-validation and keep API keys in environment variables only.