After a hiatus of more than two years, a legendary Wall Street investor has redirected attention toward Chinese assets.

According to the latest 13F filings, Stanley Druckenmiller's investment platform, Duquesne Family Office, established a new position in Baidu's ADR during the second quarter. This marks the fund's first purchase of Chinese internet stocks since it fully exited its Alibaba ADR holding in the fourth quarter of 2023.

Druckenmiller is an unavoidable name on Wall Street, having been a key figure in the famous 1992 trade that shorted the British pound. Yet, what truly sets him apart is his long-term track record. From 1981 to 2010, his Duquesne fund consistently delivered annualized returns of roughly 30% without a single losing year.

In 2010, Druckenmiller voluntarily closed Duquesne Capital when it managed over $12 billion in assets. This is precisely why his every move is closely scrutinized by the market.

His investment framework is built on a few straightforward principles: macro analysis, trend following, favorable odds, and position sizing. He first identifies the overarching direction, then seeks out the most promising assets to bet on. When his conviction is strong, he is willing to concentrate his firepower; when he is wrong, he quickly admits the mistake.

He once shared a compelling insight: the market is usually right most of the time; the real danger lies in being trapped in the small fraction of time when you are wrong.

The true significance of his renewed interest in Chinese internet stocks is not merely what he bought, but why a top-tier investor who had previously stepped away from this sector for years has now returned.

The answer may be straightforward. Over the past two years, the narrative surrounding Chinese technology assets has shifted. The emergence of DeepSeek, in particular, has demonstrated to global markets for the first time that Chinese AI is not just a "follower," but has developed its own competitive edge in cost, model efficiency, and application deployment.

This rapid progress in China's AI sector is reshaping how overseas capital perceives Chinese tech assets. Increasingly, international investors are moving from a macro-level assessment of China to actively seeking out specific Chinese assets.

Recently, several overseas ETFs have been increasing their exposure to China's chip supply chain. The Roundhill Memory ETF (DRAM), which focuses on the memory chip industry, has adjusted its holdings. As of the close on August 13, Chinese memory company ChangXin Memory Technologies (CXMT) held a 4.46% weight in the fund. GigaDevice, which was first added to the fund in June, now holds a 1.16% weight.

Tema ETFs, a US active management firm, has also announced it will include CXMT in its memory-focused product, the Tema Memory ETF (DISK), with a holding ratio of 8.07% as of August 13's market close.

Goldman Sachs' latest global fund flow report shows that in the month leading up to August 5, global equity funds saw net inflows of $182.774 billion. Among these, emerging market funds attracted $87.243 billion, while China-focused equity funds drew $50.82 billion. Net inflows into Chinese equity funds accounted for nearly 60% of the total inflows to emerging market funds.

Beyond the Baidu position, Duquesne's second-quarter activity included increasing stakes in 16 stocks, reducing positions in 11, initiating new positions in 48, and completely exiting 23 others. The 13F filing reveals that Duquesne's total portfolio value grew to $5.11 billion by the end of the second quarter, up from $3.38 billion at the end of the previous quarter.

AI infrastructure stands out as one of its key investment focuses.

Most notably, Duquesne made a significant new bet on Google, purchasing 336,300 shares of Alphabet during the second quarter, valued at approximately $120 million by the quarter's end. This comes right after the family office had sold its entire position of 385,000 Alphabet shares in the first quarter.

This rapid reversal between buying and selling is quintessentially Druckenmiller: rather than holding onto stocks indefinitely, he continuously reassesses the odds based on new information.

During the second quarter, he also substantially increased his stake in Amazon by 495,800 shares, boosting the position by more than tenfold to 541,600 shares. Amazon is now his 9th largest holding.

He also added 94,000 ADRs of TSMC. Following this increase, TSMC now accounts for 5.4% of his portfolio, making it his second-largest holding.

Duquesne also raised its position in STMicroelectronics by 490,000 shares, bringing the total to 3.1 million shares, valued at $232 million. STMicroelectronics is now the family office's third-largest holding.

As of the end of the second quarter, Duquesne's top ten holdings were: Natera, TSMC, STMicroelectronics, Invesco S&P 500 Equal Weight ETF/CALL, Insmed, iShares MSCI Brazil ETF/CALL, Insmed/CALL, YPF, Amazon, and BBB Foods.

Looking back over Druckenmiller's decades-long career, his investment philosophy is built on four core principles: look at the big picture first, then find opportunities; prioritize favorable odds and be willing to take large positions; admit mistakes quickly and never hold on stubbornly; and when you are right, maximize your profits.

First, "top-down" analysis. Druckenmiller is a quintessential macro investor. He does not start by valuing a specific company, but instead examines the economic cycle, interest rates, central bank policies, and liquidity before seeking out the assets most likely to benefit. He once famously stated: "Earnings don't move the overall market; it's the Federal Reserve Board." In essence, what drives the entire market is often not a company's current earnings, but the Fed and changes in liquidity. Simply put: first see where the water flows, then determine which boat is most likely to rise.

Second, "high conviction, large position." Druckenmiller is not a fan of diversification for its own sake. His logic is that truly great opportunities are rare. One should wait patiently, but when a high-conviction opportunity appears, one must be bold enough to bet big. He believes that if you truly understand an investment, you should put all your eggs in one basket and watch that basket closely. The key here is not the act of "all-in," but the condition that precedes it: if you truly understand it. When he doesn't understand, he prefers to do less, or even nothing at all.

Third, "odds matter more than win rate." This is one of his most central investment tenets. He believes investing isn't about how many times you guess right, but about how much you make when you're right versus how much you lose when you're wrong. An investor can be right only five or six times out of ten, but as long as losses are kept small when wrong and gains are substantial when right, the long-term returns can be extraordinary.

Fourth, "change your mind quickly when wrong." This is the most counterintuitive aspect. The most common mistake retail investors make is seeking reasons to justify their purchase after the fact. Druckenmiller does the opposite. He doesn't sell just because the price drops; he sells when the original rationale for the purchase changes. For him, a stop-loss is not a mechanical price point; the real stop-loss occurs when the investment thesis is broken.

Fifth, "don't get off the train too early when you're right." A major problem for many investors is holding onto losing positions while quickly taking profits on winners. Druckenmiller does the reverse: limiting losses when wrong, and letting profits run when right. He summarized a lesson learned from George Soros perfectly: "The way to build long-term returns is through preservation of capital and home runs." Achieving excellent long-term returns relies on protecting principal and seizing the few opportunities that can generate enormous gains.

His investment framework can be distilled into a simple phrase: get the big picture right, ensure the odds are favorable, and take a large position; if the direction is wrong, exit immediately; if the direction is right, let the profits run.

This approach sounds simple, but the true difficulty lies in execution. It requires an investor to possess both patience and courage; the willingness to make bold bets and the humility to admit errors.

This may be the most valuable lesson Druckenmiller offers to ordinary investors. He is not a man who is always right. On the contrary, he has made significant mistakes. However, his 30-year career proves one thing: the ultimate goal of investing is never to be right every time, but to make enough when correct and lose as little as possible when wrong.

This is the simplest expression of his philosophy: avoid fatal errors and seize the truly great opportunities.