The Man Who Solved the Market

Jim Simon's father, Matty, was a sales manager for 20th Century Fox, then at a shoe store... Reagan's dad sold shoes. Feynman's dad sold uniforms.

Graduated MIT at 20.

Got arrested in Mexico for disturbing the peace during his trip "Buenos Aires or Bust" where he and friends rode from Boston on mopeds.

Returned to MIT for graduate studies but then went to Cal for his PhD. Got it at 23.

Accepted a research position at Harvard in 1963 where he also taught, but then left for the Institute for Defense Analyses in 1964 where they offered him more money to be a codebreaker. They hired doctorates for their brainpower and ambition, not experience, and it was more of a think lab. He cracked Soviet codes during the Cold War and gained experience applying math to algorithms. He still wanted to make more money though. Avid poker player and small-time trader. At the IDA, he tried to leave with 2 others to start a company to electronically trade and research stocks. He and 3 others at the IDA published an internal, classified paper for the IDA called "Probabilistic Modes for Prediction of Stock Market Behavior" that proposed a method of trading claiming 50% annualized returns.

In 1969, the US was getting involved with the Vietnam War, and due to the confidentiality of the IDA, nobody knew what they were working on and anti-war protesters often harassed employees. Simons stated that he would not be working on any government work until the war was over and would only focus on his personal research. He just said this to sound cool but really he kept working on government stuff. However, this interview was published and they fired him from the IDA.

He was offered to be the head of the math department at Stony Brooks University at age 30 and he took it. He recruited a Cornell mathematician named James Ax, who was well known and prestigious name to have. He divorced his wife, Barbara, who agreed they married too early.

He made a breakthrough with Shiing-Shen Chern, the professor he followed to UC Berkeley, only to realize he was on leave. In 1974, Chern and Simons published "Characteristic Forms and Geometric Invariants". Chern-Simons invariants proved useful in various areas of math. A decade later, it was discovered it applied to range of areas in physics, and was widely cited. In 1976, he won the Mathematical Society's Oswald Veblen Prize in Geometry, the highest honor in the field, for his work with Chern.

In 1978, at 40 years old, he left academia to start his own investment firm focusing on currency trading. He named his firm Monemetrics. He got Lenny Baum, a co-author of the IDA research paper and mathematician who spent time discerning hidden states and making short term decisions in chaotic environments. The perfect partner. Lenny was an even more successful codebreaker than Simons. Harvard undergrad and PhD in math. He and Lloyd Welch developed an algorithm to analyze Markov chains and the Baum-Welch algorithm provided ways to estimate probabilities and parameters of complex sequences with little more information than the output of the process. It was one of the more notable advances in machine learning of the 20th century.

Once Simons convinced him to join, it didn't take long for Baum to develop an algorithm to direct Monemetrics to buy currencies if they moved a certain level below their trend line and sell if too far above. From there they set up a hedge fund called Limroy. Simons then recruited James Ax to leave academia and join them as well. He now had two esteemed mathematicians working with him to unlock the secrets of the markets.

He then hired Greg Hullender, a 19 year old on the verge of getting kicked out of Cal Tech for running an unauthorized high stakes trading operation out of his room. He was a programmer and would help code algorithms and result tracking. He wanted to fully automate the trading so he could make money in his sleep. He only stayed a little before going back to school.

Baum would hold onto investments too long. He was too optimistic. He just couldn't bear to sell. It ended up being the reason for their split.

James Ax was still working but he was a bull in a China shop. Always flaring out on others and getting in fights. He also lost in poker a lot which would leave him fuming since he was a genius but just had an easy tell. He became one of the nation's top backgammon players. As recommended they fill Hullender's place with Sandor Strauss, a computer specialist at the math department of where else, SUNY. Strauss became obsessed with gathering data. He got not only open and close data, but tick data, trade data, historic commodity-price data.

In 1985, Ax and Strauss wanted to leave the cold Northeast, and so Simons agreed to let them move to Huntington Beach. They set up a company called Axcom which Simons had 25% in.

They looked for patterns in everything. Lunar and solar cycles. Weather for coffee prices in Brazil, public sentiment for futures prices, and more. Nothing was landing though.

After little success in California, they brought on Rene Carmona, a professor at UC Irvine who specialized in stochastic differential equations. Equations that make predictions using data that appears random. He suggested they use computers to search for relationships in all the data Strauss amassed to find instances of similar trading environments, and how the prices reacted.

Elwyn Berlekamp: "When I heard MIT didn't have a football team, I knew it was the school for me". He was a research assistant at Bell Laboratories. Started his PhD at MIT in EE, finished at UC Berkeley when they offered him a teaching position. He became an expert on decoding digital information. Cofounded a branch of math called combinatorial game theory and wrote a book called Algebraic Coding Theory. His work got the attention of the IDA, where he was recruited and met Simons, through they didn't hit it off because Simons was so into finance and poker and Berlekamp had no interest in that. But when he started a cryptography company, he had Simons join the board, and noted he was a great listener and contributor.

In 1988, after selling his company and making some money, he became intrigued in the challenge of trading and Simons convinced him to go to Huntington Beach to help out and learn from Ax. Ax ignored most of Berlekamp's suggestions. He suggested better trade sizes for more confident trades and also shorter term trades. Data wasn't quite good enough for intraday though. Axcom was scoring double digit returns and word started to spread. Edward Thorp, a pioneering quant trader, showed interest in investing, but Simon's chain smoking was so brutal he cancelled the meeting.

In 1988, they named their hedge fund Medallion, in honor of the prestigious math awards each had received. The fund started off very rocky. 30% losses and mounting pressures from investors. Ax and Simons bickering on trading strategy and power. Berlekamp, tired of the fighting, offered to buy Ax's stake in the firm. He thought he could turn Axcom around. Ax sold him 40% of Axcom and kept 10%. Ax moved to San Diego and chilled out a lot. Turning his life around and reconnecting with his sons after 15 years.

Berlekamp wanted to turn around the Medallion fund and do more short term trades like he had been suggesting. Scrutinize historic price information to discover sequences that might repeat, under the assumption that investors will exhibit similar behavior in the future.

They let their data point them to anomalies and they'd trade them. Patterns on days of the week for example. Floor traders liked to go home at the end of the trading week holding few or no futures contracts, just in case bad news over the weekend tanked their holdings. They'd get back in their positions after the weekend. Medallion fund bought them when they sold on Friday and sold them back to them on Monday. "We're in the insurance business" Berlekamp told Strauss. Strauss rejoined after Ax left.

They found the Deutschmark trended longer than normal currencies. They had a mix of tradable effects. Simons pushed them to not only look for big patterns but also subtle anomalies others had overlooked. "Ghosts" they called them. So quick and unnoticeable to most investors. Don't know why they're there, but they are and the trades worked.

A big advantage they had was more data. Strauss collected intraday before computers could even handle it and so they already had it when computers improved.

In 1990, Medallion fund had 55.9 percent returns. Charging 5 and 20. Simons called Berlekamp from New York constantly and started to annoy Berlekamp, who was enjoying teaching at Berkeley and didn't want to move to NY, which Simons wanted. Berlekamp suggested Simons buy him out, which he did for 6x what Berlekamp bought it from Ax.

Got rejected by lots of investors, despite strong returns, because people viewed them as too risky. Laufer took over trading strategies. Starting with the decision to employ a single trading strategy instead of various different models for different investments and market conditions. This took advantage of their swarms of data. They then broke trading days into 5 minute bars as a way to view patterns. Certain trading "bands" from Friday morning had uncanny correlation to the same afternoon, spawning a trading strategy.

More mathematicians were hired as they sought to expand. Nick Patterson, who could have his own book about him, could code and actually read about business occasionally, and was a top codebreaker at the IDA and British govt. He grew up a strong chess player, winning at coffee shops and said no to the mob who offered to stake him in a hustling scheme. He helped them with the problem of slippage. The amount your own trades affect the price. They called the difference in theoretical price and real price The Devil. They developed an algorithm to calculate how much they were missing out on and then directed certain trades to futures exchanges to reduce the market impact of each trade.

"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" - Penavic.

Medallion grew and grew, finally getting some investors and had 600M. The size caused more slippage and their signals got weaker as competitors caught up. To grow, they'd need to move into the stock market. They had only traded in equities so far.

Simons wanted to consolidate and move everyone to Long Island, NY. Strauss, who was in Berkeley, protested the move and eventually quit and sold his share in Medallion and Renaissance... ouch.

Simon's son died in a tragic biking accident when he was 34, devastating Simons and Barbara.

At the time, fundamental investors were killing it. Peter Lynch with Fidelity. George Soros and Drunkenmiller with the Quantum Fund. Bill Gross the Bond King. These guys were making a fortune by unearthing corporate information and analyzing economic trends so the idea someone could use computers to beat these pros seemed unlikely.

Patterson liked to hire smart people who weren't already on Wall Street and were unhappy in their current position. That's how he found Peter Brown and Robert Mercer at IBM. Brown and Mercer were part of the team at IBM that developed early NLP algorithms for speech recognition. Brown was the one who convinced IBM to hire the Carnegie Mellon team that was programming a computer to play chess. The Deep Blue program that eventually defeated Kasparov in 1997.

The IBM group was described as intellectual warfare. They would rip each other to shreds then go play tennis together. Locker room environment, but for intellectuals. The speech team used the Baum-Welch algorithm, which gave Renaissance some credibility when they reached out to recruit them since Simon's original partner was Lenny Baum. Brown and Mercer joined in 1993 as Renaissance was struggling to expand to the stock market.

They worked on the stock trading system which had impressive returns with small capital but profits disappeared when they increased AUM for some reason. They poured over the code and couldn't figure out why. Finally, Magerman, another IBM sider, discovered a statically typed variable for a price instead of a dynamically typed one. It fixed it.

Simons encourages an open office of collaboration. Most of the researchers there had already made contributions to their fields with independent research and now got to work with other heavyweights. Very few cared about the money even though they were all making bank. Parking lot was filled with crappy old cars. One was embarrassed to admit he flew first class. The atmosphere was informal, and academic, yet intense. It appealed to a lot of academics who enjoyed working on hard problems and Simons cultivated a sense of camaraderie. Paid group vacations with spouses.

Everyone was encouraged to find trading strategies. If their signals met certain measures of statistical strength, they were comfortable wagering on them. Even if they couldn't fully explain why. These were also less likely to be adopted by rivals. By 1997, more than half their strategies were non intuitive. Over time they usually found reasonable explanations. Some signals were too reliable to be ignored so they employed them.

Mercer would always bring politics to talk at lunch and enjoyed speaking his conservative views to his liberal colleagues.

In January of 2000, the Medallion had its highest month, 10.5 percent return. Then March 10, 2000 the tech bubble burst and it lost 90 million in a single day. Then 80 the next. 70 the next. They were freaking out and couldn't figure out why it was happening. Their losses approached 300 million when they finally decided to ditch a signal that was bleeding money that failed in a bear market. They started to rebound and saw gains immediately after. By end of 2000, Medallion soared 99 percent even after charging clients 5 percent invested and 20 percent of gains. It now had 4 billion in AUM.

"There's no data like more data" - Mercer.

They started collecting data on everything. Developing algorithms to measure a company's presence in news reports. Medallion rose 33% in 2001. S&P 500 averaged 0.2 percent over 2000 and 2001. Rival hedge funds averaged 7.3 percent.

Brown and Mercer's system worked so well that researchers could test and develop new algorithms and plop them into their existing, single trading system. New employees began identifying predictive signals in markets all over the world. Adding foreign markets smoothed out their portfolio's returns and raised their Sharpe ratio to 6.0 in early 2003. 7.5 in 2004. There was nearly no risk of the fund losing money over a whole year. $5 B portfolio.

They supercharged returns using basket options, which allowed them to borrow more. In 2002, they managed 5B, but controlled more than 60B of positions. Gained 25.8 percent despite the S&P 500 losing 22.1 percent. The basket options also enabled them to be eligible for long term capital gains tax (20%) instead of short term gains (39.5%) because they would exercise the option after a year. It was legal, technically. The IRS said they improperly claimed profits from the basket options as LTCG and paid 6.8 B less in taxes than they should have. Renaissance challenged this finding and they are still disputing it.

Simons bought a machine that sucked cigarette smoke from the air because Magerman got mad at him for ripping darts all day indoors.

In 2002, a Ukrainian named Alexey Kononenko climbed his way through the company, making an estimated 40 million a year. He was a dick though and called others deadwood who should be fired. Even saying Brown and Mercer should be out. It divided the company on whether to keep or fire him. 2 other employees left and stole secrets. Nick Simons, Jim's son, drowned in Bali, devastating Jim. Magerman hated Kononenko.

By 2005, Medallion sported 38.4 percent returns over the previous 15 years after fees. To get all his rich academic employees to focus on the firm again, he gave them a fresh new challenge: a mega-fund holding investments for long periods. Renaissance Institutional Equities Fund, RIEF. They predicted it could have $100B in assets. By 2007, it had 37B invested, making it one of the biggest hedge funds in the world.

RIEF didn't do that well though. Medallion on the other hand returned 86 and 82 percent in 2007 and 2008 respectively. It was killing it, managing 10B. Averaged returns of 45% after fees since 1988. Simons decided to pass the torch of Renaissance to Brown and Mercer.

He retired and could now spend more time on his 220 foot, $100M yacht, Archimedes. Named for Greek mathematician and inventor. So sick.

He got into philanthropy, becoming the largest private donor in the field of autism research, $100M, and launched Math for America, giving stipends to New York's top math teachers, encouraging them to stay teaching instead of going into private industry.

"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" - Mercer.

Mercer started spending his money on supporting Republican causes. He didn't trust the government and thought it should have less control. His daughter, Rebekah, became the face of the family support their causes. In 2011, they met Andrew Breitbart and were intrigued in his news organization. He introduced them to Steve Bannon, former Goldman Sachs banker, who drew up the term sheet for the Mercer family to purchase 50% of Breitbart News for $10M. In 2012 Breitbart collapsed and died at age 43. Bannon became the site's executive chairman.

Mercers invested in Cambridge Analytica and became Trump's largest financial backers. They hated Hillary. They brought Kellyanne Conway to lead a super PAC in opposition of Hillary. Robert and Rebekah Mercer convinced Trump to hire Bannon, Conway, and Bossie to lead the campaign. Trump went to Mercer's party as president elect and thanked him in front of all guests.

Simons eventually asked Mercer to step down as co-CEO because they were getting so much flak about Mercer's political actions and morale was low.