Stanley Druckenmiller's portfolio: concentration and risk

Stanley Druckenmiller's portfolio strategy concentrates risk: at 30 June 2026, Duquesne Family Office reported 90 US-listed securities on its 13F, with the ten largest at 53.2% of reported 13F value and Natera alone at 16.6%. The filing provides a dated but partial view of the family office's holdings. It also misses much of what he says he trades: currencies, bonds, commodities and shares listed outside the US.

Key takeaways

1. Concentration is the stated method. In a January 2015 speech, Druckenmiller called diversification "probably the most misguided concept everywhere" and argued for a few large bets watched closely.
2. The 13F is a partial view. In January 2026 he described being bearish on the dollar, rolling copper futures, short bonds and holding big positions in Japan and Korea; currency views, futures, shorts and shares listed outside the US are all outside a 13F.
3. Sizing matters more than hit rate. The lesson he credits to George Soros is about how much you make when right and lose when wrong, and the line is often misquoted as Soros's own.
4. The reported holdings changed substantially in one quarter. Half of the 90 securities at 30 June 2026 were not in the previous quarter's filing, and 23 earlier positions were gone.
5. An exit rule and a crash hedge work differently. Copio seeks maximum expected geometric growth after costs; explicit protection can preserve exposure without relying solely on selling during a fall.

What is in Stanley Druckenmiller's portfolio?

The latest portfolio-wide filing is the 13F of Duquesne Family Office LLC for 30 June 2026, filed with the SEC on 14 August 2026. No 30 September 2026 filing had been published when we checked on 27 October 2026. It lists 95 lines covering 90 securities, with a reported value of about $5.2bn.

At 30 June 2026, as % of reported 13F value: Natera 16.6%; Insmed 5.7%; TSMC 5.4%; iShares MSCI Brazil 5.1%; Amazon 4.6%; STMicroelectronics 4.5%; Invesco S&P 500 Equal Weight 3.7%; YPF 2.7%; CDW 2.7%; BBB Foods 2.3%. Source: Duquesne Family Office 13F-HR, information table, 30 June 2026; chart data (CSV).

Five of the ten combine shares with call options, and the 13F values each call at the market value of the shares behind it, not at its premium. The chart does not show short positions, currencies, bonds, futures or non-US listings.

Call options were 16.4% of reported 13F value, up from 13.0% at 31 March 2026. Calls on four funds that track broad markets, the S&P 500 Equal Weight, Russell 2000, S&P 500 and MSCI Brazil ETFs, made up 9.3%. Natera and Insmed together were 22.3%. Natera has been the largest position in all eight filings from 30 September 2024 to 30 June 2026.

What does a 13F leave out for a macro investor?

A 13F mainly lists long positions in US-listed equities, ETFs and certain listed options, up to 45 days after quarter end. The SEC's Form 13F FAQ says short positions are not reported and shares on non-US exchanges such as Tokyo should not be included. Cash, government bonds, currency positions, futures and private investments are outside its scope.

In a Morgan Stanley interview recorded on 30 January 2026, he described his book in his own words. He said "We still have big positions in Japan and Korea", that Duquesne was "bearish on the US dollar" and that on copper "we just keep rolling the front end", meaning near-dated futures. He added that Duquesne had some gold as "mainly a geopolitical trade" and was short bonds as a hedge that "allows me to hold the other assets I mentioned."

Those comments predate the June filing by five months, and he said in the same interview that he tends to change his mind every three weeks. The interview describes exposures absent from the filing. Teva, the stock he used as his example of Duquesne's process, was 0.4% of reported 13F value at 30 June 2026.

How concentrated is Duquesne, and how fast does it trade?

The ten largest securities have made up between 51.3% and 65.6% of reported 13F value in each of the last eight quarters. The security count ranged from 49 at 31 March 2025 to 90 at 30 June 2026.

Top-10 share of reported 13F value: 65.6% (30 September 2024), 56.2% (31 December 2024), 59.8% (31 March 2025), 51.3% (30 June 2025), 54.3% (30 September 2025), 55.3% (31 December 2025), 58.3% (31 March 2026), 53.2% (30 June 2026). Source: Duquesne Family Office 13F-HR filings on EDGAR, information tables for each quarter; chart data (CSV). The figures measure concentration inside the US-listed long book only. They say nothing about leverage or the size of positions held off the 13F.

Between 31 March and 30 June 2026, 45 of the 90 securities were new to the filing and 23 of the previous 68 had disappeared. New lines were 26.7% of reported value at 30 June, led by call options on Insmed and new shares in Alphabet, Fox, CDW and Bitdeer. Exits included Broadcom, Option Care Health and Intel. Reported value rose from about $3.4bn to $5.2bn over the quarter, so a weight change mixes price moves with trades.

Why does Druckenmiller bet so heavily on his best ideas?

Because he thinks long-term outperformance comes from a few large, well-chosen bets. At the Lost Tree Club in January 2015, he said the mistake most managers make is feeling they "got to be playing in a bunch of stuff." His alternative: "put all your eggs in one basket and then watch the basket very carefully."

The 1992 sterling trade is his best-known example. In the same speech, he said Soros's Quantum fund held about $7bn in August 1992 and he had put $1.5bn into a short position against the pound. After Bundesbank president Helmut Schlesinger's comments in September, he told Soros he planned to sell $5.5bn of pounds, "100 percent of the fund in this one trade."

By his account, Soros called it "an incredible one-way bet" and said the position should be 200%, not 100%. Britain abandoned the European Exchange Rate Mechanism that month, a crisis that Alain Naef, writing on the Economic History Society blog, traces partly to Schlesinger's remark on 15 September 1992.

Quote lists such as Insider Monkey's print "It's not whether you're right or wrong that's important, but how much money you make when you're right and how much you lose when you're wrong" as a Soros quote. In Jack Schwager's The New Market Wizards (1992), those are Druckenmiller's words, describing what he learned from Soros. He repeated the point to Morgan Stanley in 2026: "What I learned from him was sizing." For a different route to concentration, compare Bill Ackman's Pershing Square and Warren Buffett's Berkshire.

How does Druckenmiller decide when to buy, hold and sell?

He looks ahead, times the overall market with liquidity, and exits fast when he is wrong. In 2015 he credited his first mentor with two rules: "you have to visualize the situation 18 months from now", and "It's liquidity that moves markets", not earnings. In 2026 he said: "Most trades I put on, I think in terms of 18 months to three years".

In The New Market Wizards he said: "I never use valuation to time the market." He used liquidity and technical analysis for timing, and valuation to judge how far a move could run. At Lost Tree he said about 80% of his biggest gains came in equity bear markets that followed what he called central bank mistakes.

The horizon does not mean patience with losing positions. He told Morgan Stanley he has "put on a three-year trade that five days later I'm out of and I've reversed." In the 1992 book he recounted going 130% long on Friday 16 October 1987, then selling the whole position and going net short shortly after the open on Monday 19 October.

He also described losing all his young firm's capital in T-bill futures in 1982, learning that "you could be right on a market and still end up losing if you use excessive leverage." One of the two studies that changed his mind that weekend was by Paul Tudor Jones, whose approach is covered in our Tudor Jones case study.

What was Duquesne Capital's record, and why did it close?

Morgan Stanley describes Duquesne Capital Management as delivering roughly 30% annualised returns with no losing years from 1981 to 2010. When he announced the closure in August 2010, the firm oversaw $12bn and was down 5% for the year, according to Bloomberg as reported by Financial Planning. He told Bloomberg that "Managing more than $10 billion seems to challenge my long-term standard." Client money was to be returned in 2011, and he planned to open a family office for some of his wealth, WealthBriefing reported.

He has qualified the record himself. "Everybody says I've never had a down year. That's true. But a lot of it's the luck of the calendar," he said in 2026, describing drawdowns inside those years. These are reported figures, not audited returns we could check, and they cover a fund that took outside money and used leverage.

Can concentration produce a recession-resilient portfolio?

Druckenmiller's defence depends on changing positions when his view changes; concentration itself provides no crash payoff. The 53.2% top-ten share in June 2026 combines shares and calls at underlying share value, so it cannot establish how much investor capital was at risk. Selling a position can limit further losses, but execution gaps and the decision to re-enter affect the result.

Copio judges sizing, exits and hedges by their contribution to expected long-term geometric growth after costs. We believe explicit, cost-controlled convex protection can support continued growth exposure without relying solely on selling during a fall. That is a different mechanism from a discretionary exit, and its value still depends on the covered risk, premiums and reinvestment rule. Our recession-resilient portfolio guide compares these approaches at whole-portfolio level.

What can an individual investor learn from Stanley Druckenmiller?

Druckenmiller's method rests on daily risk control, a research team and instruments few households use.

What share of your portfolio sits in your largest single idea, and what would it cost if that idea failed? Do you know in advance what evidence would make you sell, or would you wait to see if it recovers? Is your portfolio built on what a company or market looks like today, or on a view of where it could be in 18 months?

Do your holdings depend on the same interest-rate and liquidity conditions, even if they carry different names? Our 36 years of asset class correlations and asset classes in four crashes show how far apparently different holdings can fall together.