PDF Summary:The Man Who Solved the Market, by Gregory Zuckerman

In The Man Who Solved the Market, business journalist Gregory Zuckerman tells the story of Jim Simons, a former mathematician who became one of the most successful hedge fund managers in history. He founded his hedge fund, Renaissance Technologies, in the late 1970s. What set Renaissance apart was Simons’s insight that price fluctuations within financial markets followed recognizable and predictable patterns. When identified, these patterns could be used to strategically buy and sell the right stocks, bonds, currencies, and other financial instruments at the right time.

In this guide, we explore Simons’s story, looking at how his experience as a codebreaker informed his later success as a hedge fund manager; how he leveraged rapid improvements in computer processing power to design algorithms to capitalize on short-term price fluctuations; and how he set the stage for today’s mathematical, large-scale, automated trading.

We’ll also supplement findings from Zuckerman’s book with insights from other financial and trading experts and explore opinions from both critics and supporters of Simons’s model of quantitative investing.

(continued)...

The Opportunities and Risks of Correlation

Identifying patterns and correlations has become a major strategy in modern investing, as investors seek to learn how assets move in relation to each other. Correlations are measured on a scale: A perfect positive correlation between two assets is written as +1 (the two assets always move in the same direction); a perfect negative correlation is written as -1 (the two assets always move in opposite directions).

Portfolio managers often use correlation to assess how diversified their assets are. As a general rule, including assets that have a low correlation to each other reduces portfolio risk: If one asset goes down in value, the uncorrelated asset is likely to go up and offset some of those losses. However, a risk of this approach is its reliance on historical data, which cannot perfectly predict future outcomes. Two previously correlated assets can become uncorrelated; likewise, two negatively correlated assets can become positively correlated, upsetting the risk balance in the portfolio.

For example, under conditions of high inflation, stocks and bonds tend to have a positive correlation—their prices move in the same direction (typically downward). However, under conditions of low inflation stocks and bonds tend to have a negative correlation—when stocks rise in price, bonds fall in price, and vice versa.

Part 2: Launching Renaissance

Although Simons learned valuable skills at the IDA and made critical contacts, he was fired from his position as a codebreaker there in 1968, due to his public opposition to the Vietnam War. At 30 years old with a wife and children, writes Zuckerman, Simons still saw himself fundamentally as a mathematician and academic who thrived in the company of other similarly minded individuals. In this section, we’ll explore how Simons transitioned from an intellectually minded academic to a wildly successful and wealthy captain of finance. Specifically, we’ll look at his time as a mathematics chair on Long Island in the 1970s; his founding of Renaissance in 1978; and the extraordinary, market-beating growth of Renaissance in the 1980s.

Mathematics Chair at Stony Brook

In 1968, Simons took a job as the mathematics chair at the State University of New York at Stony Brook, on Long Island, where he stayed until 1978. There, his connections and keen eye for talent helped him build a first-rate mathematics department, transforming Stony Brook from a little-known public university into a mathematics powerhouse that could hold its own with Ivy League universities. While at Stony Brook, Simons continued his academic work, publishing groundbreaking mathematics papers that cemented his reputation as a leading pure mathematician of his time.

Importantly, writes Zuckerman, Simons also recruited many scholars he’d known at Berkeley to Stony Brook, as well as former IDA colleagues. Many of these hires would in turn become colleagues and investors at Simon’s future hedge fund, Renaissance.

(Shortform note: Simons left a lasting impact on Stony Brook’s reputation and academic standing. According to US News and World Report, in 2023, Stony Brook ranked #30 in the United States for graduate mathematics, ahead of Ivy League universities like Dartmouth. And, although he hasn’t worked for the university in an official capacity for decades, Simons still maintains strong ties with the university: In June 2023, Simons and his wife Marilyn gave $500 million through their Simons Foundation to Stony Brook’s endowment, support scholarships, professorships, research, and clinical care.)

Seeking a New Challenge

In 1978, Simons shocked his friends when he announced his intention to leave academia and start his own fund. Despite his success and popularity at Stony Brook, Simons never quite lost the itch for investing he’d developed as a younger man—whether it had been experimenting with futures contracts in the early 1960s or building mathematical trading models with his IDA colleagues. The rush, the thrill, the attraction of using his quantitative abilities and elite standing within the mathematics world to make money seemed like a way to put pure mathematics into action.

“Mathiness” and the (Mis)use of Numbers

Some in the economics field have decried what they see as the appropriation of advanced mathematics for profit, as Simons did in his career, or to advance ideological agendas. In 2015, economist Paul Romer coined the term “mathiness” to describe what he saw as the misuse of mathematics in economics research. He decried the employment of complicated mathematics to mislead audiences and obscure authors’ true agendas—specifically how statistics and economic models are used in politics.

For example, a politician might campaign on a platform of lowering tax rates while claiming, seemingly paradoxically, that the lower rates will increase government revenue. She may justify this position by pointing to tables and graphs, which lend the appearance of mathematical rigor. This is then used to show that the lower rates will boost productivity across the economy and thus enable the government to collect more overall revenue, even at the lower rate. But the tables and graphs obscure the fact that the economic growth and revenue projections they purport to show are purely speculative and based on the most optimistic reading of the data.

Finding Signals in the Noise

Simons launched his fund in 1978. It sported an atypical look for what would become one of the most successful hedge funds of all time—far from the glittering towers of Wall Street and the glamor of Manhattan, Simons started his operation out of a strip mall on Long Island. And the shabbily dressed, scruffy, academic Simons hardly looked the part of a finance power player.

The fund, initially named Monometrics (the name was changed to Renaissance in 1982), built on the investment model Simons had begun experimenting with back in his IDA days. It focused on buying or selling currencies at the right time based on the model’s predictions of when they were most likely to rise or fall in value—predictions derived from quantitative analysis and statistical modeling of historical price data.

It was, as Zuckerman writes, the logical extension of Simons’s work as a mathematician and codebreaker: finding patterns, structure, signals, and meaning in the seemingly random data. Based on the condition or phase of the market at a given time, the model could create a probability distribution mapping out the likelihood of different sets of subsequent outcomes. From this, the model could assess which assets or commodities were likely to increase in value over a certain period of time and which were likely to decrease in value.

Alfred Winslow Jones and the Invention of the Hedge Fund

Although Simons and his team were certainly early hedge fund pioneers—and among its most successful practitioners—this form of investing had existed for decades. In 1949, Australian investor Alfred Winslow Jones started what many scholars call the world’s first hedge fund. Like Simons, Jones came from a non-financial background (in fact, he’d even been a spy like Simons), yet he was able to perceive insights that eluded traditional investors. Jones argued for a systems-based approach to investing that foreshadowed the hedge fund strategies that investors like Simons, George Soros, and Ray Dalio honed decades later.

Jones’s strategy was to create a market-neutral portfolio—in other words, one that would be “hedged” against the volatility of the market. He achieved this by buying securities his model predicted would outperform the market, and short-selling (effectively betting against) securities he believed would sell below market value. Like Simons later, his strategy was predicated upon buying and selling the right mix of assets at the right times, rather than allowing his portfolio to simply rise or fall with the fortunes of the market. And Jones’s “hedged” strategy paid off handsomely: A 1966 article in Fortune showed that Jones’s fund had outperformed the top-performing mutual fund by 44% over the preceding five years.

Early Success

Zuckerman writes that, within a few years, the mathematical modeling strategy paid off for Simons: Although still flying largely under the radar, the fund attracted clients and investors, growing by tens of millions of dollars. Simons used rigorous mathematical modeling and the nascent power of computing to test and refine his models as they absorbed new data—the building blocks of what, decades later, would be called machine learning. And in this effort, he tapped the minds of his mathematician colleagues—people like Lenny Baum, who built the original fund algorithm that directed the fund to buy or sell certain assets based on observed price movements; and James Ax, who helped build the fund’s computer trading system that enabled the firm to execute those timely trades.

The Rise of Computers

The timing of the rise of Simons’s fund in the 1970s and 1980s coincided with the extraordinary, exponential growth of computer processing power during this time. The principle behind this growth is known as Moore’s law, which states that the speed and capabilities of computer chips double every two years due to chipmakers’ increasing ability to fit more transistors on semiconductor chips. Since its promulgation in 1965, Moore’s law has largely held true. Simons and his team successfully rode the computer processing wave as they were getting Renaissance off the ground, with computing power increasing a trillion-fold from 1956 to 2015.

And it wasn’t just the finance industry that computing power was revolutionizing. One study of labor and residential data from 1970 to 2000 shows a significant shift in the skills required for new jobs across all industries after 1980—with this shift beginning at the same time that the personal computer became ubiquitous. Computer-related jobs became the most in-demand and computer-related industries saw the fastest labor growth. By 2000, all the fastest-growing jobs were related to computers.

Increasing the Trade Volume

By the mid-1980s, Simons and Renaissance were getting closer to a truly automated trading system due to cheaper and more powerful computers and access to new microdata. These enabled the fund to capitalize on intraday trading fluctuations, broken down by hours and minutes, which optimized the algorithms and made them more robust. And these intraday fluctuations pointed to the need for the fund to massively increase its volume of daily trading to capitalize on these short-term movements—to make the model agile and responsive enough to buy and sell assets within minutes or seconds.

Zuckerman writes that Simons also believed more frequent trading would reduce risk: With more trades, there would be less risk associated with each individual trade. The negative effects of a “bad” trade would be minimized, since the model didn’t need to “win” every trade to make a profit—it just needed to win the majority of them.

The key to the strategy’s success was learning to trust the model’s predictions based on the computer’s analytics of historical data. It wasn’t necessary to know anything substantive about the underlying bonds, commodities, or currencies being traded or why their prices fluctuated the way they did: What mattered was the integrity of the data and the reliability of the patterns identified by the algorithm. If the algorithm could detect a non-random pattern and make bets that paid off more often than not, the fund would invest even if it followed no apparent economic logic.

(Shortform note: Simons’s focus on short-term price movements presaged the rise of what would become known as high-frequency trading (HFT). In Flash Boys, Michael Lewis describes HFT as a type of electronic trading platform that uses automated computer algorithms to quickly buy and sell large quantities of stocks. These HFT algorithms gather market data and use this information to buy and sell stocks, completing this process in microseconds. This was a very different strategy from the time-honored method used by traditional mutual funds, which typically buy a large, diversified bundle of stocks and hold on to them for the long term. The speed advantage of the HFT algorithms enables them to accrue billions of dollars in tiny profits from trades that traditional investors and traders can’t capitalize on.)

Part 3: Simons Hits His Stride and the Quant Era Begins

The Renaissance strategy proved prescient as computer power increased in the 1990s, writes Zuckerman. Since the computers could process more data, the fund’s models became stronger, faster, and more comprehensive. By the early 1990s, returns started hitting upwards of 70%—beating the market by a wide margin. By 1993, the fund was managing over $280 million. In this section, we’ll explore how Renaissance used machine-learning principles to refine its trading code and how Simons’s mathematics-based investing strategy eventually took over the finance industry.

(Shortform note: Renaissance’s exceptional growth during the 1990s may have been enhanced by the overall strength and productivity of the American economy during this time. Particularly as the decade drew into its second half, productivity growth surged to 2.5% per year—compared to 1.5% in the early 1990s. Economic historians attribute much of this productivity boom to investment in IT (across all sectors, not just in tech-intensive companies like Renaissance). Indeed, IT investment grew from 3% of GDP in 1991 to 4.9% by the decade’s end. As a consequence, innovation more than doubled in the second half of the 1990s.)

Mercer and Brown Join Renaissance

In 1993, writes Zuckerman, Simons hired computer scientists Peter Brown and Robert Mercer, renowned for their groundbreaking work in IBM’s language recognition unit. Simons recognized that these two scientists had experience and insight that could be of great value to Renaissance. This was because language recognition software depended on the same recognition of “states” in language as Renaissance’s model depended on recognizing market “states.” Just as a certain sequence of price movements could yield reasonable predictions about the next price movement, so could a certain sequence of words yield a reasonable prediction about the next word.

With their expertise in designing big systems, the pair were tasked with establishing a single trading code to handle Renaissance’s stock-trading business, constructing an optimal set of stock holdings given the fund’s risk appetite and financial resources. Best of all, Mercer and Brown designed the code based on machine learning principles: It would be adaptive, able to synthesize new information on the fly. The code was vastly more complex than the company’s previous code by several orders of magnitude, but it proved a winner for Renaissance: The fund was managing $900 million by 1997 and $5 billion by the mid-2000s.

The Growth—and Risks—of Automated Trading

Algorithmic trading of the kind pioneered by Simons, Brown, and Mercer would sweep the financial world. Today, algorithmic trading accounts for around 60-73% of overall US equity trading. Despite its dominance, some financial experts observe that the spread of automated trading threatens to make the overall financial system riskier. Because financial markets (such as the stock market, the bond market, or the foreign exchange market) are often closely linked, algorithms that extend across markets can, in mere microseconds, inadvertently spread risk from one market to the next in a sort of “domino effect”—far faster than human traders could.

Further, because of the speed at which these algorithms trade, a single flawed algorithm can quickly lead to catastrophic losses. In 2012, Knight Capital lost $440 million in 45 minutes due to a poorly written trading algorithm that executed millions of faulty trades in about 150 stocks. Ironically, despite claims about the “efficiency” of automated trading, Knight’s slip-up gave rival traders the opportunity to exploit the company’s tailspin as employees helplessly watched their algorithm spin out of control.

The Quant Era on Wall Street

By the mid-2000s, the quant era had dawned on Wall Street. Rigorous mathematical models that analyzed the entire market were now the only way to beat the market. The financial world had at last taken notice of the emerging powerhouse fund run by the unusual collection of dressed-down mathematicians, computer scientists, and academics that was beating Wall Street at its own game. When Simons stepped down from Renaissance in 2009, he had amassed a personal fortune of $11 billion and rewritten the rules on Wall Street. His fund’s record of performance was untouchable, having consistently outperformed the market over nearly 30 years.

And in doing so, writes Zuckerman, Simons remade Wall Street in his image. The big firms that scoffed at his approach in the late 1970s and early 1980s had all to some degree or another adopted his quantitative and mathematics-based strategy by the dawn of the 2020s. From Fidelity to Merrill to the major banks, global financial institutions were investing the Renaissance way: absorbing megadata, building robust machine learning models to anticipate barely perceptible price movements, and creating computer models to automatically trade at scale based on those models.

Zuckerman writes that the datafication of everything will only make quantitative investing more powerful and dominant in the years and decades ahead. Data, after all, is the cornerstone of this investment strategy: And more data means more to analyze, more patterns to identify, and more tiny price fluctuations to capitalize on.

Does Data Collection Threaten Our Privacy?

In The Age of Surveillance Capitalism, Shoshana Zuboff describes this drive for data collection by big business as a dark new frontier for the free enterprise system. She writes that this new form of capitalism—the titular surveillance capitalism—is defined by companies harvesting data about our behavior, making predictions about our future behavior using that data, and selling those predictions for profit. The idea is that serving people’s needs is less profitable, and therefore less desirable, than selling predictions about their future behavior.

Zuboff warns that the ultimate goal of surveillance capitalism is to create a society in which our free will is replaced by behavioral conditioning that encourages predictable and machine-like patterns of behavior. At stake is our most basic right to privacy—our ability to exercise consent and control over our personal data and to not have the most intimate information about ourselves packaged and commodified.