DEMO / 09 · Demo ready
SEC business evidence
Cached SEC EDGAR company-facts ingestion with source-cited ratio analysis and review flags, never a fraud verdict.
What decision is supported
Which reported financial facts for a public company deserve a closer read by an analyst, and which filing each number came from.
Who uses it
A business analyst preparing a due-diligence note who needs every ratio traceable to an accession number, not a fraud investigator.
What data enters
Company-facts JSON from the SEC EDGAR public API, fetched by the user with the provider's terms displayed first and cached locally. Tests and the demo use an authored synthetic facts file shaped like the API response.
What is computed
Standardised concepts across taxonomy variants, period alignment, a set of ratios with the filing citation attached to each input, year-over-year deltas, and rule-based review flags with the rule text shown.
What action is suggested
A list of flagged items with the reason and the source filing for each; the tool does not classify a company and does not produce a fraud score.
What evidence supports it
No metric is reported. The demo output is a flow of facts to ratios to flags on synthetic data, labelled demo.
No metric is claimed for this project. Anything computed by the repository on its synthetic fixture carries the label demo.
What fails or is uncertain
- Concept mapping across taxonomies is rule-based and incomplete; unmapped concepts are listed, not silently dropped.
- A flag means a rule fired, nothing more. The rules are documented and can be wrong for a given business model.
- Live fetching is never exercised in tests or CI; only the cache reader is.