Official portrait: unitedstates/images (public domain; source GPO Member Guide)

Rep. Devin Nunes (R-CA)

House · Republican (most recent term) · CA-22 · in Congress 2003–2022 · Bioguide ID N000181

Party history: R 2003–2022. Source: term data in unitedstates/congress-legislators (CC0); party shown is as recorded per term there.

Sensitivity band for the average (all ranges at floor vs. all at ceiling, same holdings both years): +6.6% to +7.5%. This band is narrow because it moves both years together; the true uncertainty is larger — see the per-year “strict envelope” column and the limitations.

Estimated year-over-year change

Bars: midpoint estimate; whisker: same-position floor/ceiling band. Axis clipped to −100%…+150%. Years with insufficient traceable holdings are omitted.

Filings used

Each annual-report year uses one filing (most recently filed version among originals/amendments whose Schedule A row count is ≥ 40% of the largest). Electronic (text) PDFs are parsed. Scanned/image-only filings are OCR-read and used only if they pass the explicit validation gates described in the project README (value/income coverage, two-engine agreement, exact bracket snapping, page completeness, year-to-year position continuity); such years are labelled OCR-derived below. Scanned filings that fail validation are listed as “insufficient data – scanned”.

Largest movers 2015→2016

Largest movers 2016→2017

Largest movers 2018→2019

Method (summary)

1. Schedule A asset rows are read from each annual report; the dollar range for each row is converted to its midpoint. Open-ended top ranges (“Over $50,000,000”) are excluded from the traced set and counted as a coverage loss.
2. Positions are matched year to year by owner code (self/SP/DC/JT) + normalised asset name. A position that appears only in year t is included only if Schedule B shows a purchase of it in year t (opening value 0); a position that disappears is included only if Schedule B shows a sale (closing value 0). Otherwise it is excluded as unexplained.
3. Estimated gain = (closing − opening + reported income − purchases + sale proceeds) ÷ opening, all at midpoints, for the traced positions. “Capital gains” income is not added again (it is already in value change and sale proceeds).
4. A year counts only if traced opening value ≥ $50,000 and traced coverage ≥ 60% of the opening midpoint. A member is ranked only with ≥ 3 such years (exactly 2 = “limited data”, shown but unranked). Average = geometric mean of (1+yearly estimate).

Limitations

• Ranges are wide (up to a factor of 5 or more, and up to $50M+ open-ended); a holding can move between ranges without any real change in value, and stay in a range while moving materially. Yearly estimates are therefore noisy; the “strict envelope” shows the mathematical worst/best case and is typically wider than ±100%.
• Purchases and sales are only disclosed as ranges and only for transactions over $1,000; partial sales, dividend reinvestment, spousal/dependent accounts and retirement-plan funds are reported inconsistently. Excepted/blind trusts, personal residences and certain pensions are not reported at all.
• Filers sometimes change how they list assets (e.g. listing a brokerage account vs. its individual holdings), which can make positions untraceable across years; these are excluded rather than guessed.
• Asset values in an annual report are as of Dec 31 of the reporting year; the report year t−1→t covers that calendar year. Late or amended filings are used when filed; extensions can delay them.
• The text of these reports is machine-parsed from PDFs; parse errors are possible. Every number links back to the cited filing. This is not investment advice.
• Use of these filings is restricted by 5 U.S.C. § 105(c) (no unlawful, commercial or credit-rating purpose, and none for solicitation of money).

Generated 2026-10-01 by an automated pipeline. Data: Office of the Clerk, U.S. House of Representatives (disclosures-clerk.house.gov); member roster: unitedstates/congress-legislators (CC0).