The Man Who Solved the Market by Gregory Zuckerman

Buy the Book: Print | eBook

Gregory Zuckerman tells the story of Jim Simons and the brainpower behind Renaissance Technologies and the Medallion Fund — the most successful hedge fund ever.

The Notes

• The Medallion fund averaged a 66% annual return from 1988 to 2019!
• Simons first got into trading after his first wedding. He put $5,000 of wedding gift money into United Fruit Company and Celanese Corp. He was quickly bored of the lack of price movement. His broker suggested soybeans. Simons bought 2 futures contracts, made several thousand, got a rush from the quick gains, and was hooked.
• He joined an intelligence group (the IDA) as a codebreaker in 1964. It assisted the NSA. The job was to indentify patterns in huge amounts of data. The researchers had free rein to worked on what they wanted. The group’s credo was: “Bad ideas is good, good ideas is terrific, no ideas is terrible.”
• Simons developed a trading system at the IDA that looked past typical fundamental financial data. It focused on variables like “high variance” to predict short-term market behavior.
• He treated the market as a chaotic system.
• Markov Chains: sequences of events where the probability of the next event is independent of past events. It’s like a coin flip, each flip is independent of the last. Only with a Markov chain, it’s impossible to predict the next event with absolute certainty. Only educated guesses can be made.
• A hidden Markov chain is driven by unknown variables. We see the results of the chain but not the variables that explain the progression through the chain. Some believe markets are hidden market chains.
• Simons started Monemetrics in 1978 to trade currencies. He recruited Leonard Baum to help develop a trading system. The early system worked, making millions.
• The success pushed them to add U.S. Treasury bond futures and commodities to the system. Losses piled up. It was a massive failure.
• The Monemetrics system was crude and dependent on Simons and Baum’s instincts to make trades. Basic behavioral errors like fear and greed cost them money.
• “I don’t want to have to worry about the market every minute. I want models that will make money while I sleep. A pure system without humans interfering.” — Jim Simons
• “If you make money, you feel like a genius. If you lose, you’re a dope.” — Jim Simons
• Simons hired Sandor Strauss, a math Ph.D. He built a custom database of historical prices. He combined data sources and cleaned it until they basically had more accurate data than anyone else. It was a huge advantage.
• They collected every piece of data they could — including lunar phases and sunspots — to test its viability. Most tests failed.
• He hired James Ax to develop more sophisticated trading models with their new data. It led to stochastic equations which model dynamic processes that involve a high level of uncertainty.
• “The goal was to invent a mathematical model and use it as a framework to infer some consequences and conclusions. The name of the game is not to always be right, but to be right often enough.” — René Carmona
• Simons hired Rene Carmona. He was the first to recommend a machine learning approach. Instead of standard equations, the system reads the data and proposes buy-sell decisions. It worked. Except, they didn’t know why the system made the decisions it did.
• James Ax overrode the model and bought Eurodollar futures in 1987, which helped avoid the crash and earn a double-digit return that year. They still relied on instincts and luck!
• In 1988, Simons created a new fund called Medallion based on a new strategy. Elwyn Berlekamp was put in charge in 1989. The new trading system started in late 1989 with $27 million.
• Medallion’s average holding time ranged from a day in a half to a week in a half and was profitable almost every day.
• “Scientists are human, often all too human. When desire and data are in collision, evidence sometimes loses out to emotion.” — Brian Keating
• Berlekamp pushed for shorter-term trades because the long-term trades were less successful. He hoped to recreate the fund in the image of a casino. If they had a small statistical edge, the law of large numbers would tilt the odds in their favor. The small edge was based on numerous recurring patterns that were faint but noticeable thanks to their trove of data.
• “Their goal remained the same: scrutinize historic price information to discover sequences that might repeat, under the assumption that investors will exhibit similar behavior in the future.”
• During the Iraq invasion of Kuwait, gold and oil prices jumped in price. Simons wanted to override the system and buy gold. Emotion got the best of him again. Berlekamp didn’t give in.
• Berlekamp wasn’t convinced the new model would continue to achieve high returns. Simons disagreed — patterns existed in the data which would inform their computer models to identify markets trends.
• Simons bought out Berlekamp in late 1990 and converted everyone else’s stake into shares of Renaissance Technology.
• Benoit Mandelbrot argued markets have fractal patterns. Fractals are complex patterns that occur in dynamic systems. His “theory suggested that markets will deliver more unexpected events than widely assumed, another reason to doubt the elaborate models produced by high-powered computers. 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.”
• “A model is a simplified version of reality, like a street map that shows you how to travel from one part of the city to another. If you got them right, [you] could then use the rules to predict what would happen in new situations.” — Ed Thorp
• The Medallion model was a single trading model that essentially incorporated several models for different investments and market conditions into one. The simpler and easier approach would have been to run each model independently, but a single model benefited from their massive amounts of data. It also made it easier to add new investments and models later on. The model would later be adapted to learn from trades that couldn’t be executed so it could self-correct by searching for other trades to tilt the portfolio where it needed to be.
• Advances in computing made it easier to dissect the intraday data into smaller and smaller slices to uncover oddities to trade around. First, the team cut the trading day in half, then quarters, and eventually down to 5-minutes slices. In those 5-minute slices, they found patterns in morning trading that predicted afternoon trades. They found patterns between pairs of investments.
• “It wasn’t immediately obvious why some of the new trading signals worked, but as long as they had p-values, or probability values, under 0.01 — meaning they appeared statistically significant, with a low probability of being statistical mirages — they were added to the system.”
• They uncovered so many potential trades, their next challenge was finding the optimal trade and how much to bet on it. Their “betting algorithm” was dynamic, adapting in real-time and adjusted the fund’s holdings to changes in the market.
• “We’re mediocre traders, but our system never has rows with its girlfriends — that’s the kind of thing that causes patterns in markets.” — Nick Patterson
• “What you’re really modeling is human behavior. 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.” — Kresimir Penavic
• “It continued to identify enough winning trades to make serious money, usually by wagering on reversions after stocks got out of whack. Over the years, Renaissance would add twists to this bedrock strategy, but, for more than a decade, those would just be second order complements to the firm’s core reversion-to-the-mean predictive signals.”
• “An employee boils it down succinctly: ‘We make money from the reactions people have to price moves.’”
• “Any time you hear financial experts talking about how the market went up because of such and such — remember it’s all nonsense.” — Peter Brown
• The team read hundreds of research papers but when they tested a paper’s proposed strategy proposed, most failed to live up to the published results.
• “Simons insisted on a different approach — Medallion would have a single, monolithic trading system. All staffers enjoyed full access to each line of the source code underpinning their moneymaking algorithms, all of it readable in cleartext on the firm’s internal network. 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.”
• “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.”
• By 1997, over half the signals discovered were one’s they couldn’t completely understand. They used it anyway unless it was completely nonsensical. “It’s not that they wanted trades that didn’t make any sense; it’s just that these were the statistically valid strategies they were finding. Recurring patterns without apparent logic to explain them had an added bonus: They were less likely to be discovered and adopted by rivals, most of whom wouldn’t touch these kind of trades.”
• “If there were signals that made a lot of sense that were very strong, they would have long-ago been traded out. There are signals that you can’t understand, but they’re there, and they can be relatively strong.” — Peter Brown
• “We ask, ‘Does this correspond to some aspect of behavior that seems reasonable?’” — Jim Simons
• “When you’re down by half, people figure you can go down all the way. They’re going to push the market against you… You’re finished.” — Vinny Mattone re: John Meriweather and LTCM
• “For Patterson and his colleagues, the LTCM collapse reinforced an existing mantra at Renaissance: Never place too much trust in trading models. Yes, the firm’s system seemed to work, but all formulas are fallible. This conclusion reinforced the fund’s approach to managing risk. If a strategy wasn’t working, or when market volatility surged, Renaissance’s system tended to automatically reduce positions and risk.”
• “LTCM’s basic error was believing its models were truth. We never believed our models reflected reality — just some aspects of reality.” — Nick Patterson
• The Dotcom bubble bursting caused Medallion to hemorrhage losses thanks to a momentum strategy that bought any shares that rallied the previous week. After turning off the momentum strategy, the fund was backing to making gains.
• The goal was more data in early 2001. They collected trade orders, annual/quarterly reports, insider trades, government reports, economic predictions, news stories, insurance claims, and more. They added a terabyte of data annually in search of more patterns.
• Medallion made 150,000 to 300,000 trades per day to purposely avoid impacting market prices. They added signals in foreign markets. The addition of foreign markets reduces the funds market correlation and smoothed out returns. Its Sharpe ratio in 2003 was 6.0!
• Medallion added basket options in the early 2000s which were options on a basket of stocks. “The fund’s computers told the banks which stocks to place in the basket and how they should be traded… All day, Medallion’s computers sent automated instructions to the banks, sometimes an order a minute or even a second. After a year or so, Medallion exercised its options, claiming whatever returns the shares generated, less some related costs.” The banks carried most of the risk since they held the stocks. The most the fund could lose was the premium on the options.
• Simons worried about the fund’s performance suffering if it got too big. He started returning all the investors’ funds in 2003.
• To capitalize on their success, Simons started a new fund that would be “long-term” in nature. They settled on a model that held investments for a month or longer that would beat the market by a few points per year.
• The 2007 Quant Crash caused severe losses for both funds. Simons overrode the model to reduce exposure.
• By 2010: “Medallion still did bond, commodity, and currency trades, and it made money from trending and reversion-predicting signals, including a particularly effective one aptly named Déjà Vu. More than ever, though, it was powered by complex equity trades featuring a mixture of complex signals, rather than simple pairs trades, such as buying Coke and selling Pepsi. The gains on each trade were never huge, and the fund only got it right a bit more than half the time, but that was more than enough.”
• “We’re right 50.75 percent of the time…but we’re 100 percent right 50.75 percent of the time. You can make billions that way.” — Robert Mercer
• “‘The inefficiencies are so complex they are, in a sense, hidden in the markets in code,’ a staffer says. ‘RenTec decrypts them. We find them across time, across risk factors, across sectors and industries.’ Even more important: Renaissance concluded that there are reliable mathematical relationships between all these forces. Applying data science, the researchers achieved a better sense of when various factors were relevant, how they interrelated, and the frequency with which they influenced shares. They also tested and teased out subtle, nuanced mathematical relationships between various shares — what staffers call multidimensional anomalies — that other investors were oblivious to or didn’t fully understand. ‘These relationships have to exist, since companies are interconnected in complex ways,’ says a former Renaissance executive. ‘This interconnectedness is hard to model and predict with accuracy, and it changes over time. RenTec has built a machine to model this interconnectedness, track its behavior over time, and bet on when prices seem out of whack according to these models.’”
• “‘There is no individual bet we make that we can explain by saying we think one stock is going to go up or another down,’ a senior staffer says. ‘Every bet is a function of all the other bets, our risk profile, and what we expect to do in the near and distant future. It’s a big, complex optimization based on the premise that we predict the future well enough to make money from our predictions, and that we understand risk, cost, impact, and market structure well enough to leverage the hell out of it.’”
• “It’s a very big exercise in machine learning, if you want to look at it that way. Studying the past, understanding what happens and how it might impinge, nonrandomly, on the future.” — Jim Simons
• Broad Lessons:
  • Simons’s true talent was recruiting math and physics Ph.D.’s that were creative problem solvers. He borrowed the idea from his time at the IDA, which looked for the smartest, most creative types, as opposed to finding someone to fill specific skills. He collected some of the greatest minds around.
  • “The firm’s success is a useful reminder of the predictability of human behavior. Renaissance studies the past because it is reasonably confident investors will make similar decisions in the future. At the same time, staffers embrace the scientific method to combat cognitive and emotional biases, suggesting there’s value to this philosophical approach when tackling challenging problems of all kinds.”
  • “Another lesson of the Renaissance experience is that there are more factors and variables influencing financial markets and individual investments than most realize or can deduce. Investors tend to focus on the most basic forces, but there are dozens of factors, perhaps whole dimensions of them, that are missed.”
  • “For all the unique data, computer firepower, special talent, and trading and risk-management expertise Renaissance has gathered, the firm only profits on barely more than 50 percent of its trades, a sign of how challenging it is to try to beat the market — and how foolish it is for most investors to try.”
  • The goal of trading models is to eliminate emotions that lead to mistakes. Yet, despite Simon’s insistence to “trust the model,” it’s difficult to completely let go. Several times the fund was losing money, emotion got the best of him, and he overrode it. Each time it worked out. Was it luck or skill? In either case, models aren’t perfect but neither are people.

Buy the Book: Print | eBook

Or read other book notes.

The post The Man Who Solved the Market by Gregory Zuckerman appeared first on Novel Investor.