Methodology and data sources

What the numbers on this site are made of, and where they stop being reliable.

Where the trades come from

Members of Congress must report stock transactions within 45 days under the STOCK Act. This site combines three public sources, never mixing the same period twice:

• House, 2014 to 2025: periodic transaction report PDFs published by the House Clerk, read automatically. Scanned paper filings have no text and are not included; that is roughly 10 to 60 percent of filings per year, higher in early years.
• Senate, 2014 to August 2026: filings from the Senate eFD system, as collected in the MIT-licensed kadoa-org congress-trading-monitor dataset. It covers about 60 current and recent senators.
• 2026 onward: the CapitolStockWatch open data feed (CC BY 4.0), updated every two hours.

Latest filing in the dataset: 2026-11-05. Amounts are disclosed as ranges, not exact values. Ticker symbols are missing for about a third of the lines in the feed and the dataset also contains bonds, options and funds; pages list only stock and fund tickers.

How a copy-trade backtest works

• The simulated investor acts on the filing date, never on the date the politician traded, because nobody could have known earlier.
• A copy trade executes at the market open on the first trading day that is at least the chosen delay after filing and strictly after the filing date.
• No leverage, no shorting; a sale only sells shares that were bought by copying that same politician.
• Splits and dividends are applied from the price history; deposits are removed from return calculations.
• The benchmark is the S&P 500 (SPY) with the same deposits on the same days.

What can go wrong

• Disclosed amounts are wide ranges, so sizes are estimates.
• Companies that were delisted or renamed may lack prices; their copy trades are rejected and reported.
• Past results say nothing certain about the future, and this is not investment advice.