Writing
Four kinds of piece, four different jobs: what the data shows, how the accounting is decided, how the machinery works, and what changed in the basis.
Insights
Findings from our own data — companies on one basis, and what moves when the treatment changes.
Nothing published yet.
Accounting & method
Treatments, standards, and how a basis decides. The reasoning, with citations.
Nothing published yet.
Engineering
How the machinery works — extraction, the calculation graph, determinism.
How we measure extraction recall — and why most pipelines can't
Filings contain a built-in answer key: numbers that are both XBRL-tagged and written out in the prose. Rendering every tagged fact into the string forms it would take in a sentence, then searching the text for them, gives a ground-truth set to score the extractor against. On the reference filing the sweep found 545 of 545.
The part of XBRL every vendor throws away
Alongside the numbers, filings ship a calculation linkbase — the company's own declaration of which line rolls into which subtotal, with signs. It makes adjustments propagate correctly, it ties out arithmetic automatically, and it exposes comparability gaps invisible in flattened data: Danaher's filed graph computes a gross-profit subtotal and Thermo Fisher's does not.
Releases
Basis changelog: what changed, why it is or isn't cited, and which companies moved.
Nothing published yet.