Streetbeat Unquant

Unquant guide · Updated 2026-08-03

US Congress transaction disclosures for AI agents

Unquant normalizes US House and Senate financial transaction disclosures for agentic research. It preserves the distinction between when a transaction occurred and when the disclosure became public, which is essential for avoiding look-ahead bias.

What the records contain

Individual records include the member, chamber, transaction type, transaction date, disclosure date, issuer or symbol where available, and the reported amount range. History begins in 2012 for the currently available dataset.

The aggregate dataset groups disclosed activity by ticker, including purchase and sale counts, member breadth, disclosed amount bounds, and median filing lag. It summarizes filings; it does not reconstruct portfolios or prove a coordinated view.

Disclosure lag and amount ranges

A transaction can be disclosed as many as 45 days after it happened. Historical analysis must use the disclosure date when asking what was knowable at a point in time. Reported amounts are ranges, not exact trade sizes; any midpoint is only an estimate.

A congressional disclosure does not establish current holdings, intent, consensus, or expected return. It is a public filing record and should be described that way.

Questions this data can support

  • List recent disclosed transactions for a symbol with both relevant dates.
  • Measure historical filing lag without treating the filing as a live signal.
  • Compare purchase and sale disclosure counts across a bounded period.
  • Join disclosures to daily market data using an explicit, bias-aware date rule.

Public contracts

Datasets used in this guide