More on this book
Community
Kindle Notes & Highlights
Read between
July 18 - July 31, 2023
In recent years, Renaissance has been scoring over $7 billion annually in trading gains. That’s more than the annual revenues of brand-name corporations including Under Armour, Levi Strauss, Hasbro, and Hyatt Hotels. Here’s the absurd thing—while those other companies have tens of thousands of employees, there are just three hundred or so at Renaissance.
Early on, Simons made a decision to dig through mountains of data, employ advanced mathematics, and develop cutting-edge computer models, while others were still relying on intuition, instinct, and old-fashioned research for their own predictions.
In trading, as in mathematics, it’s rare to achieve breakthroughs in midlife. Yet, Simons was convinced he was on the verge of something special, maybe even historic.
Unlike his rivals, Simons didn’t have a clue how to estimate cash flows, identify new products, or forecast interest rates. He was digging through reams of price information.
They posited that the market had as many as eight underlying “states”—such as “high variance,” when stocks experienced larger-than-average moves, and “good,” when shares generally rose.
The whys didn’t matter, Simons and his colleagues seemed to suggest, just the strategies to take advantage of the inferred states.
Simons and the code-breakers proposed a similar approach to predicting stock prices, relying on a sophisticated mathematical tool called a hidden Markov model. Just as a gambler might guess an opponent’s mood based on his or her decisions, an investor might deduce a market’s state from its price movements.
Until then, investors generally sought an underlying economic rationale to explain and predict stock moves, or they used simple technical analysis, which involved employing graphs or other representations of past price movements to discover repeatable patterns.
Simons and his colleagues were proposing a third approach, one that had similarities with technical trading but was much more sophisticated and reliant on tools of math and science.
He rarely wore socks, even in the frigid New York winters, a practice he would continue into his eighties. “I just decided it takes too much of my time to put them on,” Simons says.
A decade later, theoretical physicist Edward Witten and others would discover that Chern-Simons theory had applications to a range of areas in physics, including condensed matter, string theory, and supergravity. It even became crucial to methods used by Microsoft and others in their attempts to develop quantum computers capable of solving problems vexing modern computers, such as drug development and artificial intelligence.
The odds weren’t in favor of a forty-year-old mathematician embarking on his fourth career, hoping to revolutionize the centuries-old world of investing. Indeed, Simons appeared closer to retirement than any sort of historic breakthrough.
Today, though, Baum’s algorithm, which allows a computer to teach itself states and probabilities, is seen as one of the twentieth century’s notable advances in machine learning,
Baum fit the stereotype of an absentminded professor—once, he came to work with half a beard, explaining that he had become distracted thinking about mathematics while shaving.
The inconsistencies bothered Straus. He hired a student to write computer programs to detect unusual spikes, dips, or gaps in their collection of prices. Working in a small, windowless office next to Ax and down a spiral staircase from Simons, Straus began the painstaking work of checking his prices against yearbooks produced by commodity exchanges, futures tables, and archives of the Wall Street Journal and other newspapers, as well as other sources. No one had told Straus to worry so much about the prices, but he had transformed into a data purist, foraging and cleaning data the rest of the
...more
Ax concluded it was time to bring in someone with experience developing stochastic equations, the broader family of equations to which Markov chains belong. Stochastic equations model dynamic processes that evolve over time and can involve a high level of uncertainty. Straus had recently read academic literature suggesting that trading models based on stochastic equations could be valuable tools.
stochastic differential equations were his specialty. These equations can make predictions using data that appears random; weather-forecasting models, for example, use stochastic equations to generate reasonably accurate estimates.
Market prices are sometimes all over the place, though. A model dependent on running simple linear regressions through data points generally does a poor job predicting future prices in complex, volatile markets marked by freak snowstorms, panic selling, and turbulent geopolitical events, all of which can play havoc with commodity and other prices.
Correlations from one period to the next shouldn’t happen with any frequency, at least according to most economists at the time who had embraced the efficient market hypothesis. Under this view, it’s impossible to beat the market by taking advantage of price irregularities—they shouldn’t exist. Once irregularities are discovered, investors should step in to remove them, the academics argued.
Simons was a mathematician with a limited understanding of the history of investing, however. He didn’t realize his approach wasn’t as original as he believed. Simons also wasn’t aware of how many traders had crashed and burned using similar methods. Some traders employing similar tactics even had substantial head starts on him. To truly conquer financial markets, Simons would have to overcome a series of imposing obstacles that he didn’t even realize were in his way.
Professor Lo and others argue that technical analysts were the “forerunners” of quantitative investing. However, their methods were never subjected to independent and thorough testing, and most of their rules arose from a mysterious combination of human pattern recognition and reasonable-sounding rules of thumb, raising questions about their efficacy.4
Simons agreed with Berlekamp that technical indicators were better at guiding short-term trades than long-term investments. But Simons hoped rigorous testing and sophisticated predictive models, based on statistical analysis rather than eyeballing price charts, might help him escape the fate of the chart adherents who had crashed and burned.
Over time, these specialists became known as quants, short for specialists in quantitative finance.
There were good reasons to be skeptical of the “computer people.” For one thing, their sophisticated hedging didn’t always work so perfectly. On October 19, 1987, the Dow Jones Industrial Average plunged 23 percent, the largest one-day decline ever, a drop blamed on the widespread embrace of portfolio insurance, a hedging technique in which investors’ computers sold stock-index futures at the first sign of a decline to protect against deeper pain. The selling sent prices down further, of course, leading to even more computerized selling and the eventual rout.
Mandelbrot’s work would reinforce the views of trader-turned-author Nassim Nicholas Taleb and others that popular math tools and risk models are incapable of sufficiently preparing investors for large and highly unpredictable deviations from historic patterns—deviations that occur more frequently than most models suggest.
Partly due to these concerns, those tinkering with models and machines usually weren’t allowed to trade or invest. Instead, they were hired to help—and stay out of the way of—the traders and other important people within banks and investment firms.
In 1964, Thorp turned his attention to Wall Street, the biggest casino of them all. After reading books on technical analysis—as well as Benjamin Graham and David Dodd’s landmark tome, Security Analysis, which laid the foundations for fundamental investing—Thorp was “surprised and encouraged by how little was known by so many,” he writes in his autobiography, A Man for All Markets.7
Thorp’s trading formula was influenced by the doctoral thesis of French mathematician Louis Bachelier, who, in 1900, developed a theory for pricing options on the Paris stock exchange using equations similar to those later employed by Albert Einstein to describe the Brownian motion of pollen particles. Bachelier’s thesis, describing the irregular motion of stock prices, had been overlooked for decades, but Thorp and others understood its relevance to modern investing.
The Morgan Stanley traders became some of the first to embrace the strategy of statistical arbitrage, or stat arb. This generally means making lots of concurrent trades, most of which aren’t correlated to the overall market but are aimed at taking advantage of statistical anomalies or other market behavior. The team’s software ranked stocks by their gains or losses over the previous weeks, for example. APT would then sell short, or bet against, the top 10 percent of the winners within an industry while buying the bottom 10 percent of the losers on the expectation that these trading patterns
...more
investors often tend to overreact to both good and bad news before calming down and helping to restore historic relationships between stocks.
Laufer made an early decision that would prove extraordinarily valuable: Medallion would employ a single trading model rather than maintain various models for different investments and market conditions, a style most quantitative firms would embrace. A collection of trading models was simpler and easier to pull off, Laufer acknowledged. But, he argued, a single model could draw on Straus’s vast trove of pricing data, detecting correlations, opportunities, and other signals across various asset classes. Narrow, individual models, by contrast, can suffer from too little data.
Straus and others had compiled reams of files tracking decades of prices of dozens of commodities, bonds, and currencies. To make it all easier to digest, they had broken the trading week into ten segments—five overnight sessions, when stocks traded in overseas markets, and five day sessions. In effect, they sliced the day in half, enabling the team to search for repeating patterns and sequences in the various segments. Then, they entered trades in the morning, at noon, and at the end of the day.
Laufer’s five-minute bars gave the team the ability to identify new trends, oddities, and other phenomena, or, in their parlance, nonrandom trading effects.
Straus and others conducted tests to ensure they hadn’t mined so deeply into their data that they had arrived at bogus trading strategies, but many of the new signals seemed to hold up.
Certain trading bands from Friday morning’s action had the uncanny ability to predict bands later that same afternoon, nearer to the close of trading. Laufer’s work also showed that, if markets moved higher late in a day, it often paid to buy futures contracts just before the close of trading and dump them at the market’s opening the next day.
In the 1970s, Israeli psychologists Amos Tversky and Daniel Kahneman had explored how individuals make decisions, demonstrating how prone most are to act irrationally. Later, economist Richard Thaler used psychological insights to explain anomalies in investor behavior, spurring the growth of the field of behavioral economics, which explored the cognitive biases of individuals and investors.
Among those identified: loss aversion, or how investors generally feel the pain from losses twice as much as the pleasure from gains; anchoring, the way judgment is skewed by an initial piece of information or experience; and the endowment effect, how investors assign excessive value to what they already own in their portfolios.
Investors overreact to stress and make emotional decisions. Indeed, it’s likely no coincidence that Medallion found itself making its largest profits during times of extreme turbulence in financial markets, a phenomenon that would continue for decades to come.
“What you’re really modeling is human behavior,” explains Penavic, the researcher. “Humans are most predictable in times of high stress—they act instinctively and panic. Our entire premise was that human actors will react the way humans did in the past … we learned to take advantage.”
It was self-evident that the surest way to score huge sums in the market was by unearthing corporate information and analyzing economic trends. The idea that someone could use computers to beat these seasoned pros seemed far-fetched.
Frey’s models usually just focused on whether relationships between clusters of stocks returned to their historic norms—a reversion-to-the-mean strategy.
Constructing a portfolio of these investments figured to dampen the fund’s volatility, giving it a high Sharpe ratio. Named after economist William F. Sharpe, the Sharpe ratio is a commonly used measure of returns that incorporates a portfolio’s risk. A high Sharpe suggests a strong and stable historic performance.
“Any time you hear financial experts talking about how the market went up because of such and such—remember it’s all nonsense,” Brown later would say.
There would be no corners of the code accessible only to top executives; anyone could make experimental modifications to improve the trading system. Simons hoped his researchers would swap ideas, rather than embrace private projects.
By 1997, Medallion’s staffers had settled on a three-step process to discover statistically significant moneymaking strategies, or what they called their trading signals. Identify anomalous patterns in historic pricing data; make sure the anomalies were statistically significant, consistent over time, and nonrandom; and see if the identified pricing behavior could be explained in a reasonable way.
One strategy with enduring success: betting on retracements. About 60 percent of investments that experienced big, sudden price rises or drops would snap back, at least partially, it turned out.
Soon, researchers were tracking newspaper and newswire stories, internet posts, and more obscure data—such as offshore insurance claims—racing to get their hands on pretty much any information that could be quantified and scrutinized for its predictive value.
The Medallion fund became something of a data sponge, soaking up a terabyte, or one trillion bytes, of information annually, buying expensive disk drives and processors to digest, store, and analyze it all, looking for reliable patterns.
If there was a newspaper article about a shortage of bread in Serbia, for example, Renaissance’s computers would sift through past examples of bread shortages and rising wheat prices to see how various investments reacted,
Watching for patterns in how stocks traded following earnings announcements, and tracking corporate cash flows, research-and-development spending, share issuance, and other factors, also proved to be useful activities.
