Editorial and corrections: how NadirAlpha checks reports

What “checked” means at NadirAlpha

Every report is built from the company's original documents (SEC, EDINET, investor relations). A pipeline reads the filings, extracts the figures with their location and checks them automatically against the original tables, against the company's own history and against plausibility rules. For accepted names the core figures are frozen as reference values: every later run is compared against that reference, and every deviation must be explained and released before it reaches the page. The date of the last acceptance is shown in the report header (“checked”).

What is automated and what people do

Automated: extraction of the figures, source link for every figure, consistency checks, generation of the texts with AI assistance, translation. People: acceptance of the reference values per name, spot checks of figures and texts, the decision on every rule change in the pipeline, handling of reader reports. There is no editorial sign-off of every single sentence. That is why every figure carries its source and every page carries the AI disclosure.

Disclosure

The reports are AI-assisted and labelled as such (Art. 50 EU AI Act). They contain no recommendations, no price targets and no ratings. Advertising on the site has no influence on the selection or content of the reports.

Reporting errors, corrections

Please use the contact page and name the ticker and the chapter. Since every figure carries its source, a report can be verified quickly. Confirmed errors are corrected with the next update of the name; errors in the reference values lead to a new acceptance and an updated date in the report header. Reviewing a report usually takes a few working days.

Responsibility

The NadirAlpha editorial team is responsible. Provider details are in the legal notice. How the reports are produced in detail is described in the methodology.