Methodology
How the lab works
How we chose seventy traders in three groups, how each paper portfolio trades, and where the numbers can mislead you.
Every rule on this page was frozen before its portfolio went live, on November 6 or November 9, 2026. Results before a portfolio's live date are a backtest. Results after it come from the same rules, recorded one session at a time.
Choosing the seventy
Funds, independent investors and members of Congress leave different public records, so we score them separately. Institutions file 13F holdings or run listed ETFs, individuals publish their methods on YouTube or X, and House members report their trades. Each group has its own Jev questions, weights, checks and seats. We listed 156 candidates and confirmed each identity against a primary source, such as a fund's website, an SEC filing, the House Clerk or the person's own channel. Anyone we couldn't confirm was dropped.
Audience
On November 6, 2026 we recorded each individual's YouTube subscribers and X followers from their public pages and added them together. Reach turns that total into a score from 0 to 1 on a log scale: 100,000 people scores 0 and 10 million or more scores 1. Only individuals use reach. Funds and members of Congress are followed through filings and news coverage, so Jev's mindshare score stands in for popularity.
What we asked Jev
Jev is a scoring model from TypeSafe. For each candidate we sent the same short profile: role, markets, published method, how positions are disclosed and the observed audience, plus the measured book for funds and the trade record for members of Congress. Jev returns an expected score on each scale and a confidence. The wording below is exactly what the model received.
Loading…
How the scores combine
Each answer is divided by the top of its scale, so every input runs from 0 to 1. Every group weights popularity 60% and simulability 40%.
composite = 0.60 × popularity + 0.40 × simulability
Institutions popularity = mindshare
simulability = 0.50 × mirror fidelity + 0.25 × holdings disclosure
+ 0.25 × US equity focus
Individuals popularity = 0.50 × reach + 0.50 × mindshare
simulability = 0.45 × codifiability + 0.30 × method transparency
+ 0.25 × US equity focus
Politicians popularity = mindshare
simulability = 0.70 × record usability + 0.30 × US equity focus
Checks and seats
Every candidate needed a confirmed identity. Each group also had its own checks:
- Institutions: 13F-HR filings or a listed ETF, a holdings-disclosure score of at least 2.5, a mirror-fidelity score of at least 1, and a 13F filed within the past 200 days.
- Individuals: a codifiability score of at least 1 and a public YouTube or X channel.
- Politicians: a sitting House member with at least 10 stock or ETF trades naming a ticker since January 2025, at least five positions still open after replaying those trades, a stock trade in the past year, and a record-usability score of at least 1.
The highest composite scores that passed fill each group's seats: 15 funds with a mirrored book of $10 billion or more, 15 smaller funds, 30 individuals and 10 members of Congress. If one fund size can't fill its seats, the other size takes them.
Why some famous names are missing
Popularity alone doesn't get anyone in. Jim Cramer, Keith Gill, Chamath Palihapitiya and Graham Stephan scored high on mindshare but publish no rules specific enough to code. George Soros, Ken Griffin and Paul Tudor Jones file 13Fs, but a long-only copy misses most of what their firms do, and Michael Burry's firm no longer files a current one. Gary Black, David Ryan and Byron Donalds passed every check and ranked just below the cut in their groups.
| # | Trader | Audience | Popularity | Simulability | Composite | Path | Outcome |
|---|---|---|---|---|---|---|---|
| Loading… | |||||||
How portfolios trade
- Every portfolio starts with $1,000,000 in cash on January 2, 2026.
- Decisions use each session's closing prices. Every trade fills at the next session's open.
- Each buy and each sell pays 5 basis points (0.05%) of its value.
- Prices are adjusted for dividends and splits, so dividends are treated as reinvested.
- Cash earns nothing. There is no leverage, shorting or options.
- Fractional shares are allowed, so each portfolio can match its target weights exactly.
- The benchmark buys SPY on the first session under the same rules and holds it.
Fund mirrors
When a fund's real positions are public, we copy them instead of guessing. All thirty institution portfolios work this way and are graded A.
Listed funds
Three managers run a listed ETF: Cathie Wood (ARKK), Tom Lee (GRNY) and Dan Ives (IVES). Their portfolio buys the fund with the full $1M and holds it. The fund's price already reflects every trade the manager makes, after fees.
13F filings
Large US managers must report their long US stock positions to the SEC within 45 days of each quarter's end. For 27 managers we read every 13F-HR filing and:
- keep long common stock only. We drop option rows and bond rows, short positions never appear in a 13F, and we also drop convertible notes, preferred shares, warrants and rights that some filers list as shares;
- hold the 100 largest positions by reported value, weighted by value. Seven managers run very broad books (Bill Nygren, Ken Fisher, Joel Greenblatt, Ron Baron, Steve Cohen, Mario Gabelli and Tom Gayner), so we hold their 300 largest;
- rebalance at the next open after the filing becomes public, so the portfolio only uses what investors could see that day.
Filings identify securities by CUSIP, not ticker. We match each CUSIP to a US listing with OpenFIGI, then by company name against the SEC's ticker list, then by search. A match that fails our name checks stays unmapped rather than guessed. Positions with no US listing or no price history, and cases where two share classes resolve to one ticker, are left out and the rest are scaled up. Each profile shows how much of the filed value we cover.
The size line uses the value of the stock book we mirror: $10 billion or more is a large fund, anything smaller is a boutique. For a listed ETF we use the fund's net assets.
Distilled rules
Most independent investors don't publish positions, but many publish their method in videos, books, courses and posts. For all thirty individuals we wrote the method as code that runs on daily US stock and ETF prices. Each profile lists the rules in plain language and links the sources we used. Eleven are graded B and nineteen C.
- Grade B: the person published concrete rules and Jev scored them as closely reproducible (codifiability of 2 or more).
- Grade C: a proxy. It follows the published style, but we had to choose thresholds, lookbacks or position sizes ourselves.
Code can't capture judgment, news, intraday timing or options. A rules portfolio shows how a published method would have done, not what the person actually traded.
Congressional trade books
Under the STOCK Act, House members report stock trades over $1,000 in periodic transaction reports, due no more than 45 days after the trade. We rebuild a book for each of the ten members from those reports. These portfolios are graded B: the trades are the member's own, but the sizes and timing are our estimates.
- We read every report the member filed with the House Clerk since January 1, 2024. Scanned paper reports are listed but can't be parsed.
- We keep stock and ETF trades that name a ticker. Options count only once exercised into stock; bonds, funds without a ticker and private assets are left out.
- Each purchase is sized at the midpoint of its reported value range and converted to shares at that day's close. A full sale closes the position, and a partial sale subtracts the midpoint without going below zero.
- Positions are weighted by estimated value. The portfolio rebalances at the next open after the first session on or after each report's filing date.
Reports cover the member's own accounts, a spouse's, joint accounts and dependent children's, and each profile shows the split. Positions bought before 2024 and never traded since don't appear, so a member's real holdings can differ a lot from the rebuilt book.
Reference ETFs
NANC and GOP are listed ETFs that invest in stocks disclosed by Democratic and Republican members of Congress. We hold each with the full $1M as a reference next to the rebuilt books. They aren't scored or ranked.
Why the House only
The Senate publishes its reports through efdsearch.senate.gov, which refuses automated requests. We don't work around that, so senators aren't included.
How we use the reports
These reports are public. Federal law (5 U.S.C. 13107) bars using them for a commercial purpose, except by news and communications media for dissemination to the general public. We publish our analysis openly for the public, and we don't sell the data or use it to raise money.
Metrics
total return = final equity ÷ $1,000,000 − 1 vs SPY = total return − SPY total return max drawdown = largest fall from a previous equity peak Sharpe = mean(r − r_BIL) ÷ stdev(r − r_BIL) × √252
r is the portfolio's daily return and r_BIL is the daily return of BIL, a 1–3 month Treasury bill ETF that stands in for the risk-free rate. Sharpe is annualized over 252 sessions.
Data sources
- Prices
- Daily open and close prices from Yahoo Finance, adjusted for dividends and splits. It's free and unofficial. Symbols it fails to return are left out of the build.
- Holdings
- 13F-HR filings from SEC EDGAR; ETF prices for fund mirrors.
- Trades
- Periodic transaction reports and the yearly filing index from the House Clerk's financial disclosure site. Party, district and committees come from the Clerk's member data.
- Identifiers
- OpenFIGI and the SEC company ticker list, checked by hand where they disagree.
- Scores
- Jev (jev-1.14.0) from TypeSafe, recorded on November 7, 2026.
- Audience
- Public YouTube channel pages and X profiles, recorded on November 6, 2026.
- Universe
- Current S&P 500 constituents for stock rules and rankings.
- Photos
- Official House portraits from the Clerk, which are US government works in the public domain, and freely licensed images from Wikimedia Commons, used only when the file belongs to the verified person's own Wikipedia article. Everyone else is shown with initials.
Limitations
- We picked these seventy in November 2026, knowing who is famous now. People become famous partly because their calls worked, so the January–November backtest flatters the group. The period after each portfolio's live date is the fairer test.
- 13F mirrors trade weeks after the manager did and miss anything bought and sold within a quarter, as well as shorts, options and foreign holdings.
- Congressional books trail the member by weeks, estimate sizes from value ranges, and can't see positions bought before 2024 that were never traded. They can look more concentrated than the member's real holdings.
- Stock rules pick from today's S&P 500, which excludes companies that dropped out during the year. That makes the backtest look better than it would have in real time.
- Rules portfolios are our reading of a published method. The people named didn't write, review or approve them.
- Results ignore taxes, market impact and slippage beyond the flat 5 basis points.
- Ten months is short. One bad session, like QQQ's 6.5% fall on November 5, 2026, can reorder the leaderboard.
Revisions
- November 7, 2026. The lab grew from 30 traders to 70 in three groups, and a third method now rebuilds House members' books from their disclosed trades. The 43 new portfolios and the NANC and GOP reference ETFs go live on November 9. Grade A now also covers ETFs that track a group, and grade B covers rebuilt disclosed trades. Costs, execution and the rules of the 27 portfolios that stayed are unchanged. Paul Tudor Jones left because a long-only copy of his 13F misses most of his macro trading. Gary Black and Ross Gerber ranked 16th and 20th for 15 boutique seats.
- The new filers report TotalEnergies, James Hardie and Liberty Global under two CUSIPs each. Under the existing rule, a ticker reported under more than one CUSIP is left out rather than guessed. David Einhorn's mirror now covers 96–98% of each filing instead of 98–100%, which moves his backtest by under 1%. Joel Greenblatt's moves by less than 0.1%.
Disclosures
Unpriced is a research and media project. We don't take outside money, manage accounts, sell signals or give investment advice, and nothing here is a recommendation to buy or sell anything.
Every portfolio is hypothetical. No real money is invested and past or simulated results don't predict future returns.
The people named here aren't affiliated with Unpriced and haven't endorsed the lab. Names, funds and marks belong to their owners and appear only to identify whose public strategy or record is being studied.
AI does much of the work. Jev scored the candidates, AI coding assistants helped write the strategy code and summaries, and an AI model drafts the short notes in the Daily Archive. The rules, data and scores are published so anyone can check them.
Photo credits
House members' photos are their official portraits from the House Clerk. Other photos come from Wikimedia Commons under the licenses below. Everyone without one of these is shown with initials.
- Loading…