Every report on this site is generated from a company's own SEC filing. This page states where each number comes from, how the written analysis is produced, and how the fundamentals score is calculated — so you can judge the output rather than take it on trust.
A large language model reads the filing's management discussion and risk-factor sections and extracts the points that moved the results. It is given the filing text only — it does not browse, and it is not asked for an opinion on the stock.
Every extracted point carries the filing section and page it came from, so any claim can be checked against the primary source. If a point has no citation, treat it as unverified and let us know.
The 1-10 score is a fixed rule set over the filing's own figures plus the number of high-severity risks found in it. There is no model judgement in it. Each report shows the exact rules that fired under "How is this scored?". Starting from 5:
The score describes the quarter's fundamentals as filed. It is not a price target, not a recommendation, and says nothing about valuation or what the market has already priced in.
A filing's GAAP diluted EPS and the adjusted figure analysts forecast are different measures, and for companies with large one-off items they differ materially. Where our analyst comparison uses the adjusted basis, the report labels it and states the GAAP figure alongside it. On annual reports we do not compare the two at all, because the filing covers twelve months while the consensus figure covers a single quarter.
Chinese reports are machine-translated from the English analysis and checked automatically: every monetary figure and percentage in the translation is compared by value against the source, and a translation that changes a number is rejected rather than published. Amounts are written in the 亿/万 convention Chinese financial media uses. Older reports have not been translated and show the English analysis under a notice.
Filings are parsed automatically, so a company using unusual XBRL tags can produce a missing or mislabelled line. AI-extracted analysis can misread emphasis even when it cites correctly. Nothing here is investment advice, and the SEC filing is always the authority — we link it on every report.
Found something wrong? We want to know: support@earningslens.ai