US Congress Stock Trades Scraper – STOCK Act Disclosures

Get every stock trade a sitting US House member disclosed under the STOCK Act, one clean JSON row per transaction. Built for retail-investing dashboards, journalists, and researchers and AI agents who want to track congressional trading without downloading and parsing PDFs by hand. Source: the Clerk of the House's own financial-disclosures site — the STOCK Act requires this data to be published, and it is US government work, not copyrighted.

• One row per transaction – member, asset, ticker, transaction type (purchase/sale/exchange), dates, and the legally required amount band, parsed straight out of the Clerk's Periodic Transaction Report (PTR) PDFs with pypdf, no OCR.
• Sitting members only, verified per filing – the Clerk's index also lists candidate filings (private citizens running for office); this Actor never fetches or emits those. Every filing is additionally checked for Status: Member inside the PDF text itself before its rows are emitted, and a filing that fails that check is skipped and counted, not guessed at.
• Defensive parsing – a transaction row the PDF layout makes genuinely unparseable is never crashed on or invented; it is skipped and reported as an uncharged error row so you can see exactly what was dropped and why. Measured on the September 2026 filings: about 3-4 % of transaction rows hit this, plus roughly one filing in eight that the House published as a scan with no extractable text at all (reported as no_text). You are never charged for either.

Sample row

One purchase from a September 2026 filing (maxFilings: 50, default input):

{
  "member_name": "Josh Gottheimer",
  "member_prefix": "Hon.",
  "state_district": "NJ05",
  "doc_id": "20035455",
  "filing_date": "2026-09-14",
  "owner_code": "JT",
  "asset_name": "Applied Materials, Inc. - Common Stock",
  "ticker": "AMAT",
  "asset_type_code": "ST",
  "transaction_type": "P",
  "transaction_type_raw": "P",
  "transaction_date": "2026-08-06",
  "notification_date": "2026-09-14",
  "amount_min": 1001,
  "amount_max": 15000,
  "amount_range_raw": "$1,001 - $15,000",
  "cap_gains_over_200": null,
  "pdf_url": "https://disclosures-clerk.house.gov/public_disc/ptr-pdfs/2026/20035455.pdf",
  "source_url": "https://disclosures-clerk.house.gov/public_disc/financial-pdfs/2026FD.zip",
  "scraped_at": "2026-09-16T11:54:40+00:00"
}

owner_code is SP (spouse), DC (dependent child), JT (joint) or null (the member). Family members are never identified by name, even when one appears in the PDF's free text — this Actor only ever emits the owner code the House form itself uses.

Filters

Input	Default	What it does
year	2026	The Clerk's financial-disclosures index year to scan. A year with no published index becomes one clean error row; the run still succeeds.
sinceFilingDate	–	Only filings submitted on or after this date (YYYY-MM-DD).
members	[]	Only filings whose member surname contains one of these (case-insensitive), e.g. ["Pelosi"]. Empty = every sitting member.
maxFilings	50	Stop after fetching this many filings (PDFs). Each filing is one PDF fetch and can produce several transaction rows — keep this low to keep runs cheap.
transactionTypes	[]	Keep only P (purchase), S (sale) and/or E (exchange). Empty = every type.

Pricing

Pay per event, no subscription: $2 per 1,000 transaction rows ($0.002 each). A default run (50 filings) typically produces 300–400 transaction rows, so around $0.60–$0.80. Set a maximum charge on the run and the Actor stops cleanly when it is reached, charging only for rows actually saved.

What this Actor does and does not cover

• Sitting House members' Periodic Transaction Reports (PTRs) only. Candidate filings (private citizens running for office, FilingType C in the Clerk's own index) are deliberately excluded — this Actor never fetches their PDFs.
• The House only. The Senate discloses separately at efdsearch.senate.gov, whose search interface blocks automated access; this Actor does not cover it.
• Owner codes, never names. Spouse and dependent-child holdings appear only as SP/DC (or JT for joint) — this Actor never emits a family member's name, even where one appears in the PDF's free-text fields, and it never emits addresses or birth dates.
• The legally required amount bands, not exact figures. The STOCK Act requires members to disclose a range (e.g. $1,001 - $15,000), not an exact dollar amount, and that is exactly what amount_min/amount_max/amount_range_raw reflect. A handful of spouse/dependent-child holdings above the top closed band are disclosed as an open-ended "Over $X" — amount_max is null for those rows rather than a guessed ceiling.
• This is public-record data, not investment advice. It shows what was legally disclosed, not what to do with it.

Details

• Source: the Clerk's own disclosures-clerk.house.gov — the annual <year>FD.zip filing index (tab-separated) and each filing's ptr-pdfs/<year>/<docId>.pdf. No login, no API key, no robots.txt restriction, US government work (not copyrighted).
• Reliability: retries with exponential backoff on 429/5xx/network errors (5 tries), a request paused at least 0.5s apart, and one bad filing never stops the run — a PDF that 404s or has no extractable text becomes an uncharged error row (error, errorMessage) and the rest continues. A run fails only when it produced zero rows and hit real request errors; a filter that simply matches nothing is a successful, empty run.
• Run stats: the STATS record in the run's key-value store holds filings fetched, filings skipped (Status ≠ Member), transaction rows that failed to parse, and request/error counts by category.
• Output: export as JSON, CSV or Excel, fetch through the Apify API, or schedule runs and pipe rows into Google Sheets, Make, Zapier, n8n, or your own code. Eligible for agentic use via Apify's MCP server — an AI agent can pull a member's recent trades mid-task and pay only per row it reads.

Not affiliated with the U.S. House of Representatives or the Clerk of the House. Data comes from the Clerk's own public financial-disclosures site.