Latticework

About

Track what 77 super-investors own and trade. Refreshed daily as new 13F filings land at the SEC. Free, no signup, no ads.

What this is

Every institutional manager with more than $100M in qualifying US equities must file a Form 13F-HR with the SEC within 45 days of each quarter-end. That filing lists every long equity position they hold. This site reads those filings for the cohort and surfaces:

Universe

77 institutional managers who file SEC Form 13F-HR — a curated list of concentrated stock-pickers, activists, and quality compounders chosen for the signal value of their public disclosures. 23 of those are kept in the dataset but excluded from the cohort-level quarterly brief— they run diversified mutual-fund books, long/short hedge funds, or quant strategies that don't reflect stock-picking conviction (Greenblatt's Gotham, Polen, Oakmark, Dodge & Cox, Pzena, Tweedy Browne, Yacktman, Causeway, Oaktree, Maverick, Viking, Markel float, etc.). They still appear in the manager directory and still get their own per-investor brief; they're just dropped from the cohort-wide synthesis where their diversification or quant book would drown out the high-conviction signals.

How was this cohort chosen?

Fair question. The list is curated by hand from publicly available 13F filers — a focused set of concentrated value investors, activists, and quality compounders whose books carry signal because they bet big on a small number of names. Examples include Buffett, Burry, Klarman, Pabrai, Ackman, Tepper, Greenblatt, and Druckenmiller; we exclude diversified mutual funds, long/short hedge funds with heavy non-equity sleeves, and quant strategies that don't reflect stock-picking conviction.

There is real selection bias in this kind of universe: the managers were chosen becausethey're interesting, and many became interesting because of historical performance. The site does notclaim “clone these managers and beat the market.” In fact the backtest table below shows the opposite: most underperform when cloned with the 45-day disclosure lag. Treat the data as research input, not a recipe.

Cohort Conviction Score (CCS) methodology

The CCS is Latticework's signature 0-100 signal per stock. It synthesizes everything the super-investor cohort is doing on a single ticker into one number. Higher = the cohort, weighted by who actually picks well, is collectively pointing at it.

Five components, fully transparent, summed to 100:

Tiers: 70+ “High conviction”, 50–69 “Strong”, 30–49 “Moderate”, 15–29 “Limited”, < 15 “Minimal”.

What CCS is not: a buy signal. A high CCS means the cohort collectively signals interest. The cohort can be wrong; the cohort can be early; the cohort can be late. Use CCS as a research filter, not a recipe. The full math is in pipeline/cohort_conviction_score.py — anyone can verify, fork, or argue with the weighting.

Backtest results & vs SPY methodology

63 of 76 backtested super-investors trail SPY when their 13F is cloned with the realistic 45-day disclosure lag. The reason isn't bad stock-picking; it's what 13Fs don't show (shorts, CDS, bonds, foreign listings, cash).

How the “vs SPY” backtest is computed

Trade date. Each filing trades on its actual filing_date at the SEC, not the period-of-report. So a Q1 position disclosed May 15 hits the clone on May 15, not March 31. No look-ahead. Median lag across the database is ~38 days; the regulatory deadline is 45.

Holdings.Long US equity only — whatever the 13F discloses. Options are excluded (rights/warrants too). Cash, bonds, foreign listings, shorts, CDS, and any other instrument the SEC doesn't require on Form 13F-HR are not in the clone.

Weights. At each rebalance, position weights are renormalized to 100% of the disclosed equity book. This removes cash drag— if Buffett holds 30% cash, the clone doesn't. It tests the stock-picking signal in isolation, not the full asset-allocation decision. Defensible but worth knowing; managers with structural cash sleeves (Buffett, Klarman) will look weaker than they really are because the clone is always 100% long stocks.

Rebalance. Only on each filing date (typically quarterly). Between filings, share counts stay constant and the book is marked daily.

SPY benchmark. Total return — dividends reinvested ( Adj Close from yfinance). Clone stock prices are also total return. The comparison is apples-to-apples.

Window. Varies per investor: starts at their first 13F filing in our database, ends at the latest mark. Range across the cohort: 4.3 to 12.1 years. A 4-year sample is much noisier than a 12-year one — when comparing managers, look at similar window lengths (the Years column is there for that reason).

Survivorship bias.The cohort is curated TODAY of well-known super-investors. Managers who blew up or quietly underperformed and left the spotlight aren't in our universe. Cohort-wide excess vs SPY likely overstates a naive “follow super-investors” strategy because you can't pick the survivors in advance.

InvestorCloneSPYExcessYears
Michael Burry27.0%16.1%+10.9%10.2y
Duan Yongping25.8%16.0%+9.8%7.1y
Robert Vinall23.5%16.4%+7.1%7.3y
Clifford Sosin19.2%14.8%+4.4%8.3y
Pat Dorsey19.1%15.3%+3.7%9.3y
Stanley Druckenmiller17.3%13.9%+3.3%12.0y
AltaRock Partners17.0%13.7%+3.3%11.2y
Valley Forge Capital Management18.3%15.1%+3.2%9.3y
Bryan Lawrence17.8%15.2%+2.6%9.0y
David Tepper17.8%15.6%+2.3%10.0y
Viking Global Investors14.5%14.0%+0.4%12.0y
Chris Hohn14.6%14.3%+0.3%10.7y
Triple Frond Partners14.2%13.9%+0.2%12.0y
Greenhaven Associates13.9%14.0%-0.1%12.0y
John Armitage13.7%13.9%-0.2%12.0y
Polen Capital Management13.5%13.9%-0.5%12.0y
William Von Mueffling13.5%14.0%-0.5%12.0y
Lindsell Train13.3%13.9%-0.6%12.0y
Josh Tarasoff15.6%16.4%-0.8%7.3y
Stephen Mandel13.2%14.0%-0.9%12.0y
Christopher Bloomstran13.0%13.9%-0.9%12.0y
Yacktman Asset Management12.8%14.0%-1.3%12.0y
Li Lu13.8%15.0%-1.3%8.9y
First Pacific Advisors12.5%13.9%-1.4%12.0y
Lee Ainslie12.6%14.0%-1.4%12.0y
Vulcan Value Partners12.4%13.9%-1.5%12.0y
AKO Capital13.7%15.3%-1.6%9.3y
Glenn Greenberg12.3%14.0%-1.7%12.0y
Jeffrey Ubben et al12.1%13.9%-1.8%12.0y
Thomas Gayner11.9%13.9%-2.0%12.0y
Bill & Melinda Gates Foundation Trust11.9%14.0%-2.1%12.0y
Chuck Akre11.8%13.9%-2.1%12.0y
Chase Coleman11.8%14.0%-2.3%12.0y
Jensen Investment Management11.7%14.0%-2.3%12.0y
Bill Ackman11.7%14.0%-2.3%12.0y
Richard Pzena11.6%14.0%-2.3%12.0y
Nelson Peltz11.7%14.0%-2.4%12.0y
Kahn Brothers Group11.6%14.0%-2.4%12.0y
Christopher Davis11.4%13.9%-2.4%12.0y
David Rolfe11.5%14.0%-2.5%12.0y
Mairs & Power Funds11.5%14.0%-2.6%12.0y
Bill Nygren11.4%14.0%-2.6%12.1y
Torray Funds11.1%13.9%-2.8%12.0y
First Eagle Investment Management11.0%13.9%-2.9%12.0y
Samantha McLemore11.6%14.6%-3.0%4.3y
Dodge & Cox Funds10.9%13.9%-3.1%12.0y
Warren Buffett10.9%14.0%-3.1%12.0y
Greg Alexander11.8%15.0%-3.2%6.3y
David Katz10.5%14.0%-3.5%12.0y
Harry Burn10.3%13.9%-3.6%12.0y
Bill Miller10.1%14.0%-3.9%12.0y
Sarah Ketterer9.8%13.9%-4.1%12.0y
Third Avenue Management9.6%13.9%-4.4%12.0y
Hillman Capital Management10.3%15.0%-4.7%6.3y
David Abrams8.9%13.9%-5.0%12.0y
Daniel Loeb8.9%14.0%-5.1%12.0y
Dennis Hong11.2%16.4%-5.2%7.3y
Robert Olstein8.7%13.9%-5.2%12.0y
John Rogers8.6%13.9%-5.3%12.0y
Thomas Russo8.5%13.9%-5.3%12.0y
Tweedy Browne8.4%13.9%-5.5%12.0y
Mason Hawkins8.2%14.0%-5.8%12.0y
Prem Watsa8.1%13.9%-5.9%12.0y
Norbert Lou7.6%13.7%-6.1%11.2y
Wallace Weitz6.5%13.9%-7.5%12.0y
Bruce Berkowitz5.4%14.0%-8.6%12.0y
Francois Rochon5.0%14.0%-9.0%11.0y
Tom Bancroft5.0%14.1%-9.1%12.0y
Howard Marks4.6%13.9%-9.3%12.0y
Terry Smith2.7%12.8%-10.0%4.8y
Mohnish Pabrai3.8%13.9%-10.1%12.0y
Seth Klarman2.8%13.9%-11.2%12.0y
Glenn Welling1.0%14.0%-13.1%12.0y
Henry Ellenbogen1.9%15.0%-13.1%6.3y
Alex Roepers-0.3%14.0%-14.3%12.0y
Francis Chou-1.6%13.9%-15.5%12.0y

All 76stock-pickers in the cohort (Greenblatt's Gotham excluded — ~13K-position quant book, intractable to clone). Backtest replicates each manager's 13F on the filing date with a 45-day disclosure lag; rebalanced quarterly to reported weights.

Sources

Frequently asked questions

What is a 13F filing?
A 13F-HR is a quarterly disclosure that institutional managers with over $100M in qualifying US equities are required to file with the SEC. It lists every long US-equity position they held at quarter-end. Managers have 45 days from quarter-end to file, so the data is always at least 45 days delayed. 13Fs don't cover shorts, bonds, derivatives, or non-US securities.
Who counts as a 'super-investor' on Latticework?
We track a curated cohort of concentrated stock-pickers, activists, and quality compounders — managers whose 13F filings carry meaningful signal because they run focused, high-conviction books. The list is sourced from public records and refined by hand: we include names like Buffett, Burry, Klarman, Pabrai, Ackman, Tepper, Greenblatt, and Druckenmiller, and exclude diversified mutual-fund books, quant strategies, and filers without a meaningful concentrated 13F.
How often is the data refreshed?
The pipeline runs daily at 11:30 UTC, polling EDGAR for new 13F filings. Most days nothing new lands; close to the four legal deadlines (Feb 14, May 15, Aug 14, Nov 14), filings arrive in waves and the site updates within hours of each. News mentions refresh daily from Google News.
Is Latticework free?
Yes. No signup, no ads, no premium tier. The optional 'Follow' feature lets you get an email when a specific super-investor files a new 13F, and a Saturday weekly digest summarizes cohort activity. Both are free.
Should I just clone these portfolios?
No. We backtest 11+ of the super-investors at clone-level (45-day disclosure lag, quarterly rebalance). Most underperform SPY this way — not because the managers are bad stock-pickers, but because 13Fs don't show the shorts, options, bonds, foreign positions, and cash management that often define their real strategy. Treat the data as research input, not a recipe.
What makes Latticework different from other 13F sites?
Three things. (1) AI-synthesized analyst-style briefs per investor and per stock, refreshed each quarter — not just raw data tables. (2) A clone backtest with realistic disclosure-lag assumptions and full methodology disclosure, so the 'vs SPY' numbers are defensible rather than marketing. (3) Daily news mentions tied to each manager and stock, plus curated discovery lists (consensus stocks, biggest buys, concentrated bets, etc.) that answer the questions people actually search for. Same SEC public data underneath; different layer of synthesis on top.
Why is there a 45-day lag on 13Fs?
By SEC rule. Managers have 45 days after quarter-end to file. The point is to delay the disclosure long enough that competitors can't front-run an open position. For backtest realism, we assume you trade on the filing date, not the position date — so a Q1 position disclosed May 15 is traded May 15, not March 31.

Not investment advice

13F filings disclose long US equity positions only — no shorts, no bonds, no foreign listings, no options strategy. Past performance does not predict future results. Always do your own diligence.

Names, affiliations & AI content

Investor and company names used on this site refer to publicly- known parties whose 13F filings or securities are described here. Latticework is not affiliated with, endorsed by, or sponsored by any of those investors, funds, or companies. The per-investor and per-stock briefs are generated by AI (Anthropic Claude) from public filing data and may contain errors — read them as machine-produced analyst notes, not verified statements. Full terms, including the trademark/affiliation disclaimer and AI content disclosure, on the Terms of use page.

Built on GitHub.