Jim Simons' Medallion – The Money Management Industry's Black Box
Jim Simons is one of the most successful hedge fund managers in history. Founder of the investment company Renaissance Technologies and the Medallion fund – the best-performing and most closed fund in modern financial history. Let's try to collect information from different sources and figure out what the secret of success is. After all, the approaches to money management used by Simons will be useful to everyone who manages money in the stock market.
Simons and team
Jim Simons' ancestors moved to the United States from the Russian Empire in the late 19th century. He himself was born in 1938 and was interested in mathematics since childhood. He studied at the Massachusetts Institute of Technology (MIT), completed his doctorate at Berkeley, and taught at Harvard. Since the mid-1960s, Simons worked on deciphering secret codes at the US Institute for Defense Analyzes (IDA), from where he was fired for criticizing the Vietnam military campaign. After which, in 1968, he headed the mathematics department at Stony Brook University for 10 years.
“I wanted to do math since I was 3 years old. I thought about numbers and shapes.” – Jim Simons
In 1978, at the age of 40, Simons left academia to pursue trading. His colleagues and family thought he was making a huge mistake – why give up a promising career in mathematics to try to beat the market?
When Jim Simons created Renaissance Technologies and first applied his mathematical skills to investing in the early 1980s, fundamental analysis was still king in the financial markets. This was back when personal computers were just beginning to emerge, and technical analysis was not nearly as popular as it is now. Most technical analysts still had to go through the painstaking process of manual charting. The term “quant trading,” which refers to computerized trading strategies created from the quantitative analysis of large amounts of data, did not even exist.
Drawing on his experience and skills as a mathematician and codebreaker, Simons looked at the market in a fundamentally different way. He decided to approach trading by creating mathematical models using algorithms and automated trading, something that is commonplace at large investment firms these days but was completely unheard of back then.
At that time, there were two popular schools of thought: a) the market is efficient – all the information is already in the prices and it is impossible to beat the market; and b) if you were someone special, like Warren Buffett or Peter Lynch, you could gain an advantage by talking to company executives and endlessly poring over their books.
But Jim Simons did not believe in any of these approaches and decided to look for patterns in the movement of market prices. If he can identify repeating patterns in the market, he can make a lot of money.
“The efficient market theory is correct in that there is no obvious inefficiency. But we are looking at anomalies that may be small in size and short in duration. We make our forecast. Then, shortly thereafter, we reassess the situation and revise our forecast and our portfolio. We do this all day. We are always in and out, out and in. So we depend on activity to make money.” Jim Simons
Simons began recruiting mathematicians and data modelers he knew from his days at the Institute for Defense Analyzes (IDA) and Stony Brook University. But for 10 years he received returns at the average market level, until he finally attracted algebraist James Ax to work. He refined existing models based on correlations that computers looked for in all the data Simons had accumulated. Together with Ax, the company Axcom Ltd. was created, which grew into the Medallion fund in 1998. In 1989, after a decline in the value of assets, a conflict arose between Simons and Ax, and the latter left the company.
Simons brought in the consulting mathematician Alvin Berlekamp of Berkeley, who bought out Ax's stake and changed the algorithm for trading short-term patterns, increasing the volume of High Frequency Trading. Mathematicians decided that it will be easier for algorithms to predict events in the market if the periods of holding positions are reduced as much as possible.
In 1990, the third year of its operation, the fund showed a return of 59%.
The next surge of talent, much of which remains the core of the company to this day, came from a team of mathematicians at IBM's Thomas Watson Research Center working on problems in speech recognition and machine translation.
Shortly before this, two of them, Bob Mercer and Peter Brown, came to IBM management in 1993 with a bold proposal. They asked to be allowed to build models that would manage the company's then colossal pension fund. IBM management ignored them, asking: What can computational linguists know about investment management?
In 1993, Simons hired Mercer and Brown. This partnership proved to be strong and very successful. Medallion began to simultaneously create positions with thousands of instruments.
“We quickly realized that the world of finance was a different environment than IBM. He is ruthless. Either your models work better than others, and you make money, or they don’t, and you go broke. That kind of pressure really forces you to concentrate.” Brown said.
Simons was 50 years old before he found success in trading and 60 before he managed to attract significant investment in the fund. Today, Simons' fortune is estimated at more than $20 billion. His story is a lesson in resilience, perseverance and self-belief.
results
Renaissance's Medallion hedge fund specializes in diversified, systematic trading using customized quantitative models derived from statistical analysis of historical price patterns. As they say, “his activities are one of the wonders of the modern financial world. It's the technological equivalent of a license to print money.” For 30 years, starting in 1988, the Medallion fund averaged 66% per annum. Or 39% after deducting the commission that the fund keeps for itself.
“Renaissance is the commercial version of the Manhattan Project. They are the pinnacle of quantitative investing. No one is even close.” —Andrew Lo, professor of finance at MIT Sloan School of Management and chairman of AlphaSimplex, a quantum research firm.
To outsiders, it is a mystery how Medallion can maintain such an average annual return over a long period. “Even after all these years, they easily outperform any attempt to copy their model,” says Philippe Bonefoy, a former Medallion investor who later co-founded Swiss-based quantum firm Eleuthera Capital.
Net return on investment in the Medallion fund (after commission)
Secrets of success
People
The success of the Renaissance, of course, is inextricably linked to the people who created it, improved it and supported the Medallion models.
Of all its employees, a third have doctorates, not in finance, but in fields such as computer science, physics, mathematics and statistics. Renaissance has been called “the best physics department in the world” and avoids hiring anyone with even the slightest hint of Wall Street involvement.
“I've always said the secret of the Renaissance is that we don't hire people with MBAs” – Alvin Berlekamp (co-founder of the Medallion Foundation)
Another likely driver of Medallion's success is Renaissance's high level of employee motivation and retention, which allows the firm to keep its proprietary algorithms and strategies secret.
Information
There are many more factors influencing the market than we ordinary people realize. Statistical methods allow us to identify subtle and multidimensional cause-and-effect relationships between factors and the market. Since its inception, the Renaissance has collected vast archives of data on everything from economic indicators to weather. The quantity and quality of this data played a key role in enabling the algorithms to discover subtle and undetected predictors of market patterns.
Simons used “big data” about 20 years before anyone else. He would go to the Federal Reserve in Manhattan and obtain decades-long price data for a wide variety of assets. He later even analyzed data going back centuries, all the way back to the 1700s. Simons once considered the possible impact of sunspots and lunar phases on markets. Or, for example, if a newspaper article appeared about a bread shortage in Serbia, Renaissance's computers would sift through past examples of bread shortages and rising wheat prices to see how prices react.
One day, the team improved its predictive algorithms by developing a fairly simple measure of how many times a company was mentioned in the news feed—regardless of whether the mentions were positive, negative, or even pure rumor.
Good backtesting gives a fund a huge advantage. People trade in markets and their assessments of situations, emotions and reactions create price patterns that will be repeated in the future. The Medallion Foundation is like a sponge, absorbing terabytes of information, purchasing expensive equipment to digest, store and analyze the data, looking for reliable patterns.
“Price movement patterns are not random. However, they are quite close to random” – Jim Simons
Signals
When the IBM team came to Renaissance, Medallion had already been generating relatively high returns for some time. In those early days, anomalies were easy to detect and exploit. A Renaissance scientist once noted that the closing times for Standard & Poor's options and futures were 15 minutes apart. He turned this part into a profit-generating machine for some time, said one of the fund’s former investors.
Another quant found correlations within markets broken down into small time periods. He started looking at things like, “If gold goes up on Friday afternoon, what is it doing on Monday morning?” By looking at these types of correlations, he finally found repeating patterns from which to build reliable trading systems.
The quants squeezed out every last drop. Taken together, such deviations brought in serious money – first millions, and then billions.
But the Fund really took off when it started focusing on stocks. Initially the focus was on bonds, commodities and currencies. However, the fund needed much more assets to grow.
With Mercer and Brown joining the team, a completely new system was coded for more instruments, which was essentially a complex version of statistical arbitrage.
Today, Medallion’s strategy includes holding thousands of long and short positions and uses approximately the following algorithm of actions:
evaluate financial instruments based on complex mathematical models and proprietary algorithms;
they find two different tools, one of which is relatively expensive, and the other is relatively cheap;
they buy one and sell the other, betting that at some point prices will return to the proper level.
“We have something like 90 PhDs in physics and mathematics who are just looking for such signals all day long. We have 10,000 processors constantly running, processing data in search of signals.” Brown said
Bob Mercer once said that Medallion is right 50.75% of the time. And he added, “When we are talking about millions of transactions, billions can be made this way.”
Simply put, the Medallion fund makes money in much the same way as a casino. The house has a slight edge, but small wins can add up to big profits over time.
“If you're potentially making hundreds of thousands or millions of trades, even a small profit per trade adds up to a lot of money,” explains Campbell Harvey, professor of finance at the Fuqua School of Business.
This is true. Medallion has turned a $100 investment into the fund in early 1988 into $400 million in 2020. By comparison, $100 invested in the stock market in 1988 would have grown to $2,000 over the same time period with dividends reinvested.
Today Medallion uses dozens of “strategies” that work together as one system. The code on which the fund operates consists of several million lines, according to people close to the company. Different teams are responsible for different areas of research.
“We look for historical data, looking for anomalous patterns that we wouldn’t expect to find by chance. Once we find them, we check for statistical significance. What we do is not going anywhere. We can have bad years, sometimes we can have a terrible year. But the principles we follow will remain. Profitable trading comes down to finding a trading system with a positive expected value and managing risk through position sizing.” – Jim Simons
Discipline
Once you have created a profitable trading system, you must trade it consistently and with discipline. No trading system will work unless it is followed long enough to allow it to “take” the statistical advantage on which it is designed. All systems have drawdowns or losing streaks, even those used by the Medallion fund. All traders have times when they want to stop acting on their trading plan.
Medallion differs from most in that it does not deviate from its trading plan. “They're scientists and mathematicians, they don't necessarily feel as emotional as you and I, they're kind of a different breed… If you rely on models and the scientific method, then you can beat the market.” – Gregory Zuckerman of The Wall Street Journal .
“The computer has its own opinion, and we slavishly follow it.” – Simon
The Renaissance team views the narratives that most investors cling to to explain price movements as strange and even dangerous because they create misplaced confidence that market situations can be adequately understood and future prices predicted. A Renaissance employee once said that if it were up to him, the shares would have numbers rather than names, so that investors would be less pressured by the story behind the company.
Fluctuations and drawdowns
Diversification into stocks was a big breakthrough for Renaissance and allowed the Medallion fund to increase its size to $10 billion. One of the key advantages of the stock market is the large number of instruments available. Zuckerman mentions that Renaissance may have 4,000 long and 4,000 short positions in the stock at any given time. Trading such a huge number of assets has a profound impact on portfolio risk and drawdown. The figure below illustrates the impact of trading a large number of uncorrelated assets on portfolio volatility.
The greater the number of instruments in the portfolio, the less fluctuations in the value of the portfolio, which means the lower the risks. Wide diversification allows the fund to reduce overall portfolio risks and increase leverage, attracting money on the most favorable terms
Leverage
The world-famous example of Long-Term Capital Management (LTCM) is a textbook example of very smart people (Nobel laureates) not being smart enough to understand risk. LTCM earned excess profits using up to 25x leverage. However, a lack of understanding of the risks involved led to his catastrophic death in 1998. Renaissance, on the other hand, has demonstrated a deep understanding of risk for over 30 years. Simons' team has developed a very good solution for using leverage.
The ability to use borrowed funds allows the fund to maintain extremely high returns. A system with a yield of 0.015% per day for a year will give a modest 4%. However, the same system with access to 12.5x leverage generates a compound annual return of 60%.
Weak spots
Since the 1970s, the idea of passive (index) investing has been gaining popularity in the money management market. As part of it, investors give up trying to choose stocks on their own in order to beat the market and create a portfolio in full accordance with the structure of the selected indices. That is, they create a portfolio that is a smaller copy of the market as a whole. With a passive approach, decisions are made by the index itself, and the manager simply copies the index portfolio and receives the average market return. (on the channel Trading Phronesis we have already discussed the reasons for the growing popularity of passive investing)
Today, the share of funds transferred to index funds has reached 50% and continues to grow. This trend could have a negative impact on Medallion, since the fund essentially profits from people making their own decisions and reacting incorrectly to various market situations.
“The whole genius of Renaissance Technologies and Jim Simons is that there are patterns in the market that we humans don't really pick up on, but machines do. They use incredible data… But what happens if market behavior changes? At this point there are enough dentists and people like me trading in the market that they [Ренессанс] enjoy. But what if everyone switched to passive investing? The very nature of the market may change.” – said Zuckerman.
The second weakness of the fund is market liquidity. Simons identified almost from the start that a fund's overall size could affect its performance: “too much money kills returns.” Simply put, Renaissance trading systems become crowded in the market and their efficiency decreases. For now, Simons caps Medallion's assets under management at between $10 billion, only double what it was 10 years ago. Profits are not reinvested and are distributed every six months.
Conclusion
The big secret of the Medallion Fund is that there is none. Success comes down to doing everything right, using the highest level of competence in all the components that make up proper trading. This leads to unsurpassed results.
Data: Renaissance has a huge amount of high-quality data on everything that can remotely influence price movements. They were among the first to collect and explore data archives.
Machine learning: Renaissance applied its statistical learning algorithms very early on to identify predictive data and statistically significant repeating patterns in the market. Their early adoption of machine learning in the stock market contributed enormously to their success.
Liquidity restrictions: Each asset is traded within existing liquidity restrictions to allow Medallion to enter and exit the market without undue exposure or risk.
Stealth: Positions are entered and exited in such a way as to hide their activity from the market.
Exact costs: When the statistical advantage is small, having an accurate estimate of transaction costs improves the identification of opportunities that are worth pursuing.
Discipline: Another important lesson from Jim Simons is that profitable trading is built on the basis of mathematics, but a person must have the discipline to follow the mathematics without allowing emotions to influence and “turn off” a profitable system.
Diversification: Medallion reputedly trades 8,000 assets long/short with an average trade duration of approximately 2 days. Applying a market neutral approach to such a large asset base significantly reduces volatility and drawdowns.
Leverage: All of the above best practices underlie access to significant leverage. Zuckerman estimates Medallion trades to average between 12x and 20x leverage. Medallion's unleveraged historical returns are similar to the S&P 500. However, it is the efficient use of leverage that has made the fund great.
Thanks for “likes!
Telegram channel link: https://t.me/TradPhronesis
There are links to other articles that also healthy familiarize 🙂
