SEC financial-statement health (agent-curated)
Monthly enrichment panel scoring financial-statement strength for ~58 SEC-registered US and Canadian companies. Each run fetches the keyless SEC EDGAR XBRL companyfacts JSON per company (ticker->CIK resolved from the SEC's own company_tickers_exchange.json) and extracts annual-report facts (10-K/20-F/40-F) with a documented tag-priority list (the winning tag per metric is recorded in fact_tags_used; consolidated figures only; amendments resolve to the latest-filed value; fiscal years follow each company's own filings via anchor-shift alignment). Per (company, fiscal year): the nine Piotroski (2000) criterion flags (f1_roa_positive, f2_cfo_positive, f3_roa_improved, f4_accrual_quality, f5_leverage_down, f6_liquidity_up, f7_no_dilution, f8_margin_up, f9_turnover_up), profitability/leverage/efficiency component subscores, f_score 0-9, health_tier (strong 7-9 / neutral 4-6 / weak 0-3), a deterministic health_rank within the fiscal year, and the underlying ratios (ROA, leverage with leverage_basis, current ratio, gross margin, asset turnover, shares) so the score is auditable. Only fully-scorable companies ship (missing facts are documented exclusions, never zero-filled); a run with fewer than 20 scorable companies fails loudly. Who joins this: an equity analyst or portfolio team joins on ticker (or cik) + fiscal_year to screen holdings and watchlists for financial-statement strength (primary key (as_of, fiscal_year, company_name); join keys ticker, cik, company_name, fiscal_year, country_code ISO alpha-3). Caveats: mechanical transforms of reported filings, not investment advice; Piotroski was designed for non-financials, so fintech/payment balance sheets read differently; restatements (10-K/A) change history by design. No filing text is stored.
- sec
- financial-statements
- fundamentals
- monthly