About the data

Financial statements

Every figure comes from the company's own filings with the US Securities and Exchange Commission, through EDGAR's XBRL company-facts API. Nothing is typed in by hand and nothing is estimated.

US GAAP has thousands of tags and filers do not agree on which to use, so a row labelled "Revenue" is the first concept a company actually reports from an ordered list of candidates. For most companies that is revenue from contracts with customers; for a bank it may be interest and dividend income, which is the closest published total and not the same idea. Where it matters, compare against the filing itself - every company page links to it.

A year restated by a later filing shows the restated figure. Both versions are kept, so the earlier one can be recovered.

Prices

End-of-day closes from Polygon on its free tier. They are unofficial, the most recent close is a day behind, and they occasionally have gaps. They are good enough to size a company and rank a valuation, and they are not good enough to trade on. Market cap uses the share count from the latest 10-K cover page, not the weighted-average diluted count used for earnings per share.

Derived numbers

Margins, returns and multiples are computed from the figures above - around ninety figures in all, and the screener lets you choose which of them to put on the table and filter by. Five conventions worth knowing:

A figure is also blank where the filing never tagged what it needs. A company that reported no debt concept has no debt/equity here rather than a tidy zero: "reported no debt" and "did not report debt" are different claims, and only one of them belongs at the safe end of a leverage sort.

Sectors and industries

The sector on each company page is our own grouping of the SIC code SEC assigns. It is not GICS, which is licensed and cannot be republished. SIC was last meaningfully revised in 1987 and it shows - a great deal of the modern market sits under one code for computer services. SEC's own SIC description is shown alongside ours on every company page so the authoritative label is never hidden.

The NAICS code beside it is derived, not reported. NAICS is the classification that replaced SIC for US statistical agencies in 1997 and is revised every five years, but SEC publishes no NAICS code for a filer and nobody else publishes one freely - so ours is the SIC code carried through the Census Bureau's own concordances to the 2022 vintage. Where a 1987 industry became several NAICS industries and nothing in SEC's data says which, the code shown is the level above them: a four-digit group rather than a six-digit industry, and shorter means less precise rather than wrong. Every company with the same SIC code gets the same NAICS code, because that is all SEC gives us about it.

Annual report text

The business description, risk factors and management's discussion are cut out of each company's latest 10-K. There is no structural marker in a filing that says where an item begins, so this is pattern matching over prose written by thousands of different law firms, and it is sometimes wrong. A section is only shown when it passes length and shape checks; one that does not is kept but hidden rather than displayed as if it were right.

What this is not

Not investment advice, not a recommendation, and not a real-time source. It is a way to read public filings faster.