Does Copying Druckenmiller's 13F Beat the S&P 500?

Published: 2026-09-24 · Data: SEC 13F filings, Q4 2011 – Q2 2026

Stanley Druckenmiller is famous for decades of market-beating returns. So here is a simple question: if you had copied the stocks his family office reports to the SEC every quarter, waiting for each filing to become public, would you have beaten the S&P 500? Over 14 years the answer is yes on returns and no on risk, and the gap between those two answers says a lot about what a 13F can and cannot tell you.

The setup: copying Duquesne the way a real person could

Druckenmiller manages his own money through Duquesne Family Office. Like every manager above the SEC threshold, it files a 13F each quarter listing its US long stock positions. The filing arrives up to 45 days after quarter end, so a copycat is always trading on stale news.

The test is deliberately realistic. Each quarter's portfolio is bought only once that filing would have been public, and held until the next one. Two versions are run: the top 10 positions weighted by size, and the top 50. The benchmark is SPY with dividends over the same months: February 2012 to July 2026, 174 months in total.

The headline result

The top-10 copy compounded 1.3 percentage points a year faster than the index: $10,000 would have grown to about $85,600 against about $73,200. That is a real edge, sustained for 14 years, and it survives the 45-day delay.

The cost is visible in the rest of the table. The copy was about 35% more volatile, and its worst peak-to-trough fall was nearly ten points deeper. Per unit of volatility the index delivered more: 1.05 against 0.85. A regression of monthly returns against SPY gives a beta of 0.93 and about 2.9% a year of alpha. So this is not a leveraged market bet. The excess comes from what the copy holds, just with a much bumpier ride.

Year by year: the edge is lumpy

*2012 covers February–December; 2026 covers January–July.

The copy beat SPY in 10 of 15 calendar years, yet the table does not read like steady outperformance. It reads like two long good stretches separated by a disaster. In 2021 and 2022 combined, the copycat fell about 25 points behind the index. A follower who started in early 2021 would have spent most of the next two years well behind the market, which is exactly when most people quit a strategy.

Then the last three years flipped. From August 2023 to July 2026 the top-10 copy returned 26.1% a year against 19.2% for SPY, but its worst drawdown was -16.3% against -7.6%. Anyone judging "follow Druckenmiller" by the last three years will be very impressed. Anyone judging it by 2021–2022 would have given up. Neither window tells the whole story.

Why the copy is not Druckenmiller

The strategy captures one slice of how he invests, and it helps to be precise about which slice.

• A 13F shows only US long stocks. Bonds, currencies, futures, short positions and cash do not appear. For a manager known as a macro investor, the filing is a partial view by construction.
• It is a quarter-end snapshot. Positions bought and sold within a quarter never show up, and the copy owns each portfolio for about three months regardless of what the manager did in between.
• The lag is structural. Every trade the copy makes is at least six weeks late, and the numbers above already include that delay.

So the fair reading is: the stocks Druckenmiller chose to hold, followed mechanically and late, still produced market-beating returns, with more risk. That is a meaningful result, but it is not the same as getting his track record.

Concentration: 10 names versus 50

A reasonable instinct is to copy more of the book to smooth the ride. The data only partly supports it. The top-50 version was a little less volatile (17.8% against 18.9%) with almost the same worst drawdown, and it gave up 0.7 points a year of return. Risk-adjusted, the two came out almost identical (0.86 against 0.85). Going beyond the top ten mostly diluted the edge without removing the pain of 2022, when both versions lost 28–29%.

How Druckenmiller compares with other famous names

To compare managers fairly you need the same window and clean data. The table below uses one common period, August 2016 to February 2026, and the same top-10 copy method. It includes only funds where, on average, almost all of the reported portfolio could be priced. For earlier years many 13Fs are full of delisted or unidentifiable securities, and a backtest built on what is left of the portfolio would be misleading.

*Himalaya's window starts in August 2017, its first 13F. Calendar-year counts include partial first and last years. Seth Klarman's Baupost is left out because on average about 23% of its reported portfolio could not be priced in this window.

Over the last decade, none of these copies beat the S&P 500 by a wide margin. Tepper's did best, 1.4 points a year ahead with much deeper drawdowns. Druckenmiller's copy was roughly level with the index over this particular window, ending in February 2026. It pulled ahead only when the strong months of 2026 are added (the full-period numbers above run to July). Several famous names lagged the market badly when followed mechanically and late. The question worth asking is never "is this a great investor?" but "does this investor's 13F, followed late, add anything?" For most of this list, over this decade, the honest answer was no.

So, does copying Druckenmiller beat the market?

On raw return, yes: 16.0% against 14.7% a year for 14 years, after the filing delay. On risk-adjusted return, no: the index earned more per unit of volatility, and the copy's worst stretch was far worse than the market's. Whether that trade is worth it depends on whether you could have held through 2021–2022, and most people could not.

The broader lesson is the one that matters for anyone tempted by "copy the billionaires" headlines: judge the filing, not the legend. Some managers' 13Fs have carried real information for decades. Others, however brilliant the investor, have not.

Methodology: portfolios come from Duquesne Family Office's SEC 13F filings (Q4 2011 – Q2 2026) and are applied only after each filing date. The top-10 and top-50 copies are weighted by position value, rebalanced when a new filing appears, and exclude options. The benchmark is SPY total return over identical months. Returns are before fees, taxes and trading costs. In 2012–2013 up to about a third of Duquesne's reported portfolio could not be matched to a price, so those two years are less reliable than the rest. Source data: SEC EDGAR 13F filings, processed by 13Foresight.

Source: 13Foresight analysis of SEC Form 13F filings. 13F data reflects long US equity positions held at quarter-end and is filed up to 45 days after quarter close. This article is for information only and is not investment advice.