Summary of Key Points
Currently, 80% of trading in the U.S. stock market is dominated by machines and quantitative strategies, which operate using momentum-based tactics of "chasing gains and cutting losses," leading to increasingly extreme market fluctuations (such as a 4% drop on one day followed by a rebound the next). The opportunities for human investors lie in the blind spots of these machines: new developments that start from scratch (for which there is no historical data available to feed into algorithms), supply-side reforms in traditional industries (which machines do not analyze), and taking advantage of the extreme price movements created by machines to trade in the opposite direction.
1. The U.S. Stock Market Has Changed: Machines Control 80% of Trading, with Humans in the Minority
Just over a decade ago, human traders dominated U.S. stock market transactions (80% were done by humans); now the situation has reversed—80% of daily trading is conducted by machines, with only 20% involving fundamental analysis, such as what Warren Buffett does. How do machines trade? They use momentum strategies: they focus solely on short-term data, buying what is rising sharply and selling what is falling rapidly. For example, when news emerged that Broadcom's AI chip business was underperforming expectations, machines immediately sold off in large numbers, causing the Nasdaq to drop by 4.18% in one day, with the semiconductor sector losing $1.3 trillion in value. Human traders simply could not react quickly enough.
2. Machines' Fatal Blind Spot: They Cannot Capture Opportunities That Start from Scratch
Machine decision-making relies entirely on historical data, but new developments that start from scratch (such as a old product suddenly becoming popular due to new technology) lack such data, making them invisible to machines. For instance, in 2022, Micron's stock price was only $40, even though its HBM memory products had been around for years. No one (not even Micron's CEO) anticipated the explosive demand for HBM driven by AI. Machines did not buy into this, but human investors understood the logic and invested heavily, resulting in a 25-fold increase in Micron's stock price to $1000. Moreover, Micron's current price-earnings ratio is less than 7 times—almost the lowest in the AI industry—and supply-demand tensions are expected to persist until 2028. Machines cannot predict this because they lack the relevant data.
3. Supply-Side Reforms: A Concept Familiar to Chinese Investors, but Unrecognized by American Machines
Machines only focus on short-term data and do not analyze structural changes in traditional industries. For example, the U.S. trucking industry (a century-old sector with no direct connection to AI) has seen a 50% increase in the past year. This is similar to China's supply-side reforms in 2015: demand was weak for several years, leading to the collapse of many small companies, while the remaining leaders saw their profit margins rise even without a recovery in demand. If demand recovers, profits could multiply; even if not, reduced supply still ensures steady profits. Machines do not study these changes in "old industries," but Chinese investors are well-versed in this logic (as seen with China National Building Materials Corporation back then) and can seize such opportunities.
4. Why Are Market Fluctuations Becoming More Extreme? The "Positive Feedback Loop" of Reinforcement Learning Is to Blame
The underlying logic of machine trading is reinforcement learning—similar to training a dog: if the behavior (trading) is correct (profitable), a reward is given, and it is repeated in similar situations. The problem is that 80% of machines use similar reward mechanisms, which can interact with each other: one algorithm's decision to chase gains triggers another, reinforcing the same behavior, creating a "positive feedback loop." The same applies to selling off; when one algorithm sells, others follow, leading to market crashes that are 10 times faster than before. For example, if the Nasdaq suddenly drops by 3% one day and then rebounds by 4% the next, it is not humans at work but the machines' reward systems reacting in milliseconds.
5. A Survival Guide for Human Investors in a Machine-Dominated Market:
1. Find Opportunities Unseen by Machines: Innovations that start from scratch (like Micron's HBM) or supply-side reforms in traditional industries (such as U.S. trucks) require on-site visits and conversations with management to uncover; machines cannot access this information.
2. Take Advantage of Extreme Market Movements: Buy when prices are driven down to unreasonable lows by machines, and sell when prices rise too sharply. For example, if Broadcom's stock drops 14% but its fundamentals are sound, it may be a buying opportunity.
3. Avoid Leverage and Over-Trading: Machines' extreme fluctuations can lead to sudden margin calls or liquidation. Maintaining a 30%-40% position is more secure.
4. Use a Cross-Market Perspective: Chinese investors have an advantage in understanding supply-side reforms, which American machines lack. By recognizing similarities between U.S. companies and past Chinese cases (such as ODFL and China National Building Materials Corporation), they can make early investments.
In conclusion, this is not a "human vs. machine" battle but a recognition of reality: much of the volatility in the U.S. stock market today is unrelated to fundamental factors and is driven by machines' reward systems. Understanding this will help you identify opportunities in extreme markets.