How the output is made
This page is the honest description of the tool: what it reads, what it does with it, what it refuses to do, and how it fails.
Data we read
SEC filings — annual and quarterly reports, and current reports — and company announcements. Historical price and volume series for US-listed common stock. Industry classification. We do not read data behind a paywall and we do not use non-public information.
What the model does
Change detection: it compares a new document against the previous document of the same kind and marks what is new, what was removed and what was reworded. Anomaly flags: it marks a price or volume move against that series' own trailing range and average. Digest generation: it summarises the changed passages in plain language.
Every output carries the date it was produced and a link to the source it came from. We publish no accuracy figure for these outputs, because we do not have one we could stand behind.
What it does not do
- We do not publish target prices.
- We do not give ratings.
- We do not make buy or sell calls.
- We do not offer personalised advice or portfolio construction.
- We do not forecast price direction.
- We do not manage money.
Validation
Outputs are drawn and reviewed by hand on a sampling basis. Every output links to its source, so you can check the reading directly against the document. If you find an error, email contact@lanohub.net with the date of the output: corrections are applied to the archived entry, which keeps its original date and gains a correction note rather than being quietly replaced.
Known limitations
- A change of accounting basis or segment definition can make two periods look comparable when they are not.
- One-off items can move a margin line without saying anything about the underlying business.
- Suspended securities and thinly traded names produce price and volume series that carry little information.
- Third-party data is sometimes late, revised, or wrong.
- Text extraction fails on unusual filing layouts, and a digest is never a substitute for the filing itself.
No performance claims
We publish no return figures and no comparison against any index or benchmark. We have no results to report from these outputs, and we make no claims about how any output would have turned out.