The Filign methodology
Filign is a transparent, rule-based pipeline — not a black-box predictor. Each step below contributes to a structured Filign Signal Score that describes how a company's disclosed risk profile shifted between filings.
- 01
Pull latest and previous SEC filing
Retrieve the most recent 10-Q or 10-K and the prior comparable filing directly from SEC EDGAR, preserving full document structure.
- 02
Extract comparable sections
Segment each filing into standardized sections (Risk Factors, MD&A, Liquidity, Legal Proceedings, Controls) so like is compared with like.
- 03
Detect added, removed & intensified risk language
Diff each section and classify changes — newly added risk language, removed assertions, and intensified phrasing weigh into a disclosure-risk subscore.
- 04
Compare XBRL financial statement data
Parse structured XBRL tags to compute period-over-period changes in revenue, margins, cash, debt, and share count.
- 05
Calculate rule-based Disclosure Drift Score
Combine disclosure drift and financial reality checks into a transparent 0–100 signal with source evidence and limitations.
- 06
Generate a filing intelligence memo from structured data
The research memo summarizes only the structured findings into plain language. It never introduces opinions, predictions, or recommendations.
- 07
Show disclosure drift over time
Compare recent filing pairs so users can see whether disclosure pressure is improving, worsening, or staying stable over time.
Why rule-based first
The core score is deterministic and auditable: given the same two filings, Filign produces the same components every time. The AI layer only translates those structured findings into readable language. This keeps the platform an analytical research tool, not a speculative signal generator.