What Congress Actually Traded in Crypto (2026): Every Disclosure, Parsed

Last Updated: August 2026

Disclaimer: This article is for informational purposes only and is not financial advice. It reports public disclosures required by the STOCK Act. Nothing here is a trading signal — see the section on disclosure lag for why that is mathematically the case.

Quick answer: Across 322 Periodic Transaction Reports filed in 2026, I parsed 1,815 individual transactions. Exactly 21 of them — 1.2% — touched crypto in any form. Not one was a direct holding of Bitcoin or any other coin; every single one was an equity proxy (exchanges, miners, ETFs, treasury-strategy companies). The disclosed value totals roughly $518,010 at band midpoints, and it is heavily concentrated: Gilbert Cisneros alone accounts for 15 of the 21 trades.

Why nobody is handing you this data anymore

Congress has to disclose any trade over $1,000 within 45 days. The filings are public, free, and posted by the Clerk of the House. What is not provided is anything machine-readable: the index tells you who filed and when, and the actual trades live inside individual PDFs.

The free JSON mirrors that used to parse those PDFs — house-stock-watcher and senate-stock-watcher — now return 403. The data is still public; the convenience layer is gone. So I rebuilt it, and the method is documented at the bottom so you can run it yourself.

The 21 crypto-linked trades

Every row links to the original PDF so you can check my parsing against the filing itself. Download the full dataset as CSV — free, no email required.

What the numbers actually say

It is rare. 21 crypto-linked trades out of 1,815 is 1.2%. Congressional trading is overwhelmingly ordinary equities and funds. Any headline implying a stampede into crypto is not supported by the filings.

It is concentrated — but measure it two ways. By trade count, Gilbert Cisneros accounts for 15 of 21 trades, 71% of the entire crypto footprint of the House. Strip that one member out and just 6 trades remain across everyone else.

By money, the ranking flips. Sheri Biggs discloses roughly $175,000 across 1 trade — the largest crypto exposure in the dataset — while Gilbert Cisneros's 15 trades total about $144,508. One member trades small and often; another took a single large position. "Most active" and "most exposed" are different questions, and most trackers only answer the first.

It is not directional. 9 buys against 12 sells. There is no consensus to front-run here, in either direction.

Nobody is buying coins. Every position is reachable through an ordinary brokerage account:

• COIN (7 trades) — Coinbase (exchange)
• HOOD (7 trades) — Robinhood (crypto brokerage)
• SQ (3 trades) — Block/Square (BTC on balance sheet)
• IREN (2 trades) — IREN (miner)
• IBIT (1 trade) — iShares Bitcoin ETF
• MSTR (1 trade) — MicroStrategy (BTC treasury)

The lag is why this cannot be a signal

This is the part most 'follow Congress' products skip. Measured across these filings, the median gap between making a trade and disclosing it was 24 days (range 2–43 days). 0 of 21 were disclosed more than 45 days after the trade.

By the time you can read that a member bought something, the price has had 24 days to move. You are not copying a trade; you are reading history. Any tool selling you congressional filings as a real-time edge is selling you a 24-day-old newspaper.

Limits of this dataset — read before quoting it

• Amounts are bands, not numbers. Filings report ranges like "$1,001 - $15,000". Every dollar total here uses band midpoints and is therefore an estimate, not a measurement.
• House only. Senate filings live in a separate system and are not included.
• Auto-parsed. These figures come from machine-reading PDFs. I have linked every source document precisely so errors are checkable rather than trusted.
• Disclosure is not the same as benefit. Many rows are spouse or joint accounts, and several filings note professionally managed accounts where the member may not direct individual trades.
• Absence of evidence. A trade under $1,000, or one made in a qualified blind trust, does not appear here at all.

Reproduce it yourself

```

python scripts/politician_crypto.py 2026

```

The script pulls the Clerk's year index, keeps the filings marked type 'P' (Periodic Transaction Reports — the actual trades, as opposed to annual holdings statements), downloads each PDF, extracts the text with PyMuPDF, and parses the transaction rows. PDFs are cached, so a second run costs nothing.

Source: Clerk of the U.S. House of Representatives, Financial Disclosure Reports, 2026.