House crypto-linked transactions: the 1 August 2026 scan

Last Updated: 2026-10-26

The retained extraction contains 21 crypto-linked rows among 1,815 parsed transactions from House Periodic Transaction Reports indexed under filing year 2026. The data snapshot is dated 1 August 2026. This is a finite House scan, not every congressional disclosure or a full-year census. Transaction dates in these selected rows run from 19 December 2025 to 16 June 2026: filing year and trade year are different fields.

The selected rows concern exchange and brokerage shares, mining and treasury-strategy companies, and a Bitcoin ETF. No direct coin transaction is identified in this selected set. The filter and extraction can miss records; this is not evidence that no member held or traded coins elsewhere.

Inspect the original records

Each row links to its source PDF. Download the retained 21-row CSV, without an email requirement. These are parsed records, not individually certified transactions. The CSV includes filing and notification dates that are omitted from the compact table.

Scroll sideways to see every column. Keyboard: focus the table region and use the arrow keys.

Count, turnover and exposure are different

The selected share is 21 / 1,815 = 1.16%, approximately 1.2%, of parsed transactions in this extraction. Gilbert Cisneros accounts for 15 of 21 rows, about 71.4%. That describes the extracted set, not the entire House's crypto footprint.

The sum of reported amount-band midpoints is $518,010.50. This adds purchases and sales, including repeated transactions in the same instruments. It is an estimated gross transaction total, not current portfolio exposure, net investment or profit. Midpoints are conveniences, not estimates derived from an observed distribution inside each band.

Sheri Biggs's one IBIT purchase has a $100,001–$250,000 band, midpoint $175,000.50; the 15 Cisneros rows sum to $144,507.50 in midpoints. That makes Biggs's selected transaction total larger in this arithmetic, not proof of a larger current holding or personal investment decision. The Biggs filing, spot-checked for this revision, labels the owner as spouse and the account as professionally managed.

The rows contain 9 buys and 12 sells. Counts ignore transaction sizes, earlier holdings, other investments and the reasons for trading. They cannot establish a bullish or bearish consensus. The ticker counts are COIN 7, HOOD 7, SQ 3, IREN 2, IBIT 1 and MSTR 1; ticker labels are preserved from the extraction, not silently normalized to today's names.

Filing lag is measurable; predictive value is not established

Subtracting each transaction date from its filed date gives a median of 24 days, with a 2–43 day range across all 21 selected rows. The separate notification-date field gives different intervals; it must not be substituted without changing the definition. The observed range is descriptive, not a legal-compliance finding.

A filing is therefore not a contemporaneous trade alert in this set. The median lag alone does not mathematically prove that the information has no predictive value. That would require a separate time-aligned test using actual public availability, later prices, costs, an explicit benchmark and out-of-sample data. No such result is established here, and this page does not recommend copying the transactions.

What this extraction cannot settle

• Coverage: House reports only; Senate records, unparsed documents, later filings and amendments are not certified as covered. A zero match is not proof of no activity.
• Classification: the parser uses ticker and text rules for crypto-linked exposure. A diversified company's stock is not equivalent to holding Bitcoin.
• Ownership: self, spouse and joint labels describe the filing. They do not establish who directed a trade or benefited from inside information.
• Amount: reported bands do not reveal exact cash amounts, quantities, execution prices, current holdings or returns.
• Extraction: machine-read PDF rows can be misaligned, duplicated or omitted. Links make inspection possible; they do not make the parser infallible. Only the specified Biggs example was freshly spot-checked in this revision, not every PDF.

Reproduce the arithmetic before collecting new data

The repository retains content/data/politician_crypto.json, the downloadable CSV and scripts/politician_crypto.py. For this revision, the row counts, amount-band sums, ticker/action counts and filed-date lags were recomputed from the retained JSON; no new scan was run and no trade records were changed.

The collector command is python scripts/politician_crypto.py 2026. It fetches the Clerk's index and PDFs, uses PyMuPDF, and writes data and a draft. Running it later can produce a different set; save the extraction date, code revision, attempted/successful/failed documents and any scan limit. A generated draft must be reviewed before publication. Cached PDFs can reduce repeated downloads, but do not establish complete coverage or zero resource cost.

Primary record source: Clerk of the U.S. House, Financial Disclosure Reports. This article reports an extraction and its limits, not legal advice, a compliance audit or a trading signal. AITR may earn commissions from commercial links elsewhere; the filing and download links here are research references.