About insiderdelta

The SEC publishes a tremendous volume of filings every business day: Form 4 insider transactions, 10-K annual reports, 8-K material event disclosures, Schedule 13D activist stake declarations, 13F institutional holdings. The data is technically public, but the raw APIs and bulk archives are not directly usable for most questions. insiderdelta ingests, normalizes, and surfaces this material.

The site is best understood not as a discount Bloomberg terminal but as a focused research surface for specific signal categories: insider conviction (Form 4 patterns), corporate-language change (10-K year-over-year diffs), activist intent (13D filings), institutional positioning (13F holdings), and the underlying financials that frame all of them.

• XBRL normalization, not redistribution. We do not republish a third-party aggregator’s database. We ingest directly from SEC EDGAR and resolve the XBRL tag inconsistencies (tag drift, dimensional collapse, multiple revenue concepts) that make raw SEC data unusable. The Validation & accuracy page documents the techniques.
• Form 4 as a first-class citizen. Most financial-data products treat insider trades as a peripheral data type. We treat them as the core signal - cluster buying, repeat insider participation, and signed realized returns are computed and surfaced explicitly.
• Item-level 8-K parsing. 8-Ks are sliced by item number (1.01 M&A, 2.02 earnings, 5.02 officer changes, etc.) with referenced exhibits (e.g. earnings press releases) fetched and rendered inline.
• Continuous validation. We maintain a canary-ticker suite that checks our pipeline against hand-verified values from 10-K cover pages and Yahoo Finance snapshots. Discrepancies surface in our CI before they reach users.

insiderdelta is built independently. We are not affiliated with the SEC, with any data vendor, with any issuer surfaced here, or with any registered investment advisor. We are not, ourselves, a registered investment advisor.

See Methodology for how 10-K language signals are computed, and Validation & accuracy for sources, tolerances, and disclaimers.