It has been said that a good investor must always strive to crush his most cherished beliefs. Well, during Berkshire Hathaway’s 2021 annual shareholder meeting Warren Buffett and Charlie Munger crushed one of mine. What do you think of quants? Jim Simons’ Medallion fund has done 39% net of fees for three decades which, proves that it works. But they were very smart. Yes they got very rich. Very, very smart.
Very smart and very rich, yes. And very high grade, by the way. Jim Simons.
But we are not trying to make money trading stocks. We don’t think we know how to do it. Charles Darwin used to say that any time he found evidence that contradicted his previous convictions he had to write it down in the first 30 minutes because otherwise, the mind would reject the evidence for cherished beliefs. Well, how about reading a whole book and making a video about it? During the last few years, I’ve read tons of books on personal finance and investing, and I settled down on a conclusion that value investing and fundamental analysis is the way to go, while day trading and studying price charts is just pure bogus. Or, you know, bots trying to sell you something. Enter Jim Simons.
Jim Simons is the world’s richest mathematician. Forbes estimates his wealth to be at a staggering $24.6b currently. He gained most of this wealth through conquering the world of trading by starting the “quant revolution” with his company Renaissance Technologies and the Medallion fund. The Medallion fund has the best track-record of any hedge fund in history.
It has averaged a 62.9% return per year before fees, and 37.2% net of fees, verses 11% for the S&P 500. That is the most impressive investment record I’ve ever heard about, it’s even better than Warren Buffett’s, although it has been accomplished with a much smaller capital.
So, should you abandon value investing to become a day trading quant? Let’s not get ahead of ourselves here, but we’ll get to that. In this video, we shall take a closer look at the Medallion fund’s success and reveal a few of its secrets. You’ll have to stay put for that, but I think that a quote from agent Smith from the movie The Matrix may be used to set the stage: “Never send a human to do a machine’s job.” This is a top 5 takeaways summary of The Man who Solved the Market.
Written by Gregory Zuckerman. And this is The Swedish Investor, bringing you the best tips and tools for reaching financial freedom through stock market trad… investing. Takeaway number 1: From $0 to $25b Before getting into the how’s of Simons’s incredible success, let’s first have a look at the what’s. What did Jim Simons do to go from $0 in 1938 to $25b in 2021? Simons was born in 1938. Early on, he began to read a lot. I know, surprise, surprise! Jim Simons is another one of those successful people who read a lot.
To be honest I do not think that the medium through which you consume information matters to much, what matters is that you strive for more knowledge. Simons enrolled at MIT in 1955 and was even able to skip the first year of mathematics thanks to his extensive high school curriculum. He decided early on what he wanted from life - coffee, cigarettes and lots and lots of maths. Simons was quite the adventures type. Together with two friends he decided to go from Boston to Buenos Aires, riding scooters. They named the trip “Buenos Aires or Bust”.
Well, they busted in Bogota in Colombia, but it must have been a crazy trip nonetheless. While getting his PhD at the university of California, Berkley, Simons had his first dabble in stocks. He got up early each morning to drive to Merrill Lynch’s office in Los Angeles, just in time for the opening of the Chicago exchange. He would stand there to watch prices flash by and make a few trades. Simons got his PhD by age 23, in 1961.
At age 26 he got a job as a code-breaker at a US intelligence unit, targeting old Soviet Russia. Simons learned something important about hiring people here. The unit worked very well while primary focusing on hiring people for their creativity, ambition and brainpower, rather than any specific expertise or education. Another important lesson from this place was its motto: “Bad ideas is good. Good ideas is terrific.
No ideas is terrible.” In 1968 Simons published a mathematical paper on something which I find quite difficult to pronounce, let alone understand: “Minimal Varieties in Riemannian Manifolds”. To this day, the paper has been cited 1722 times, which counts as an incredible success for a paper on geometry.
Also in 1968, Simons was asked to build and lead a maths department at Stony Brook University. It’s been said that the extroverted mathematician will look at your shoes during a conversation rather than his own. Well, Simons was extroverted, period. And he had an unusual knack for leading his fellow mathematicians. In 1978 Simons had had enough of the world of academia though.
He wanted new challenges and solving the market, conquering the world of trading, was something which no one had done before, which sparked his enthusiasm. He called his first hedge fund “Monemetrics”, which was a play of words combining “money” and “econometrics”. Simons was hinting that he would use math to analyse the financial markets and score big time. Simons utilized his unusual combination of being an exceptionally skilled mathematician himself while possessing some incredible leadership and interpersonal skills to hire and get the most out of many fellow mathematicians. He realized early on that he wouldn’t solve this puzzle by himself.
In fact, the book might as well have been called “The Men Who Solved the Market”, but you’ll hear more about these people and their contributions later, because that’s more about the how’s than the what’s of this incredible trading success. In 1982 Simons renamed the company to Renaissance Technologies, a name that it holds to this day. In 1988 Simons launched the Medallion fund, which is the most successful hedge fund of all time in terms of returns on capital. While others were still relying on instinct and intuitions for their trades, Simons employed automated algorithms, tons of data and advanced mathematics, but again we are getting ahead of ourselves.
Medallion didn’t charge the usual rip-off fee of 2/20 that other Wall Streeters did, no, no. They charged 5/20, eventually raising that to 5/44! This is insane numbers, but in Renaissance’s case, it proved to be worth it. In 1990 the Medallion fund had its first year surpassing a 50% return.
It scored as high as 77.8% before fees for the twelve months. Simons kept up his leadership skills. He created a culture of unusual openness at Renaissance.
Moreover, he used smart monetary incentives, where people were paid bonuses, but only if the company reached certain levels of profit. This money was paid out over many years to keep people in the firm. Renaissance had almost no employee turnover. Simons also had an important role to play in the hiring process.
He wanted people who had little or no connection to Wall Street and generally accepted business dogmas. In 1993 the Medallion fund closed to outside investors, from now on it was only available to employees of Renaissance and their families. In the year 2000, Medallion had its first year exceeding a 100% return, achieving a stunning +128.1%. In 2003, stocks had officially taken over as the most important trading instrument of the firm from previously having focused on currencies, commodities and bonds. In 2005, Jim Simons received a personal gain of $1.5b, which was the highest compensation
among any hedge fund manager that year. Simons retired as CEO of Renaissance in 2009 and handed over the role to two of his colleagues - Robert Mercer & Peter Brown. Simons stayed as Chairman, but eventually left that post too, just recently in 2021, but he remains on the board of directors. He earned a cool $2.6b with his financial stake in the company in 2020, reaching an estimated personal wealth of $24.6b. Okay, let’s now get into how Jim Simons (and his colleagues, I should add!) was able to achieve these stellar results in the Medallion fund.
Takeaway number 2: Medallion is a short-termpredictive algorithm The secret to Renaissance’s and the Medallionfund’s success has been to employ tons of data and advanced mathematics to develop automated trading algorithms. Renaissance was one of the pioneers of using machine learning and applying it to the world of investing. Today, this black box algorithm is an exceptional short-term predictor of market movements. The Medallion fund holds on to positions for an average of a day or so, but sometimes as little as minutes or seconds. It executed 150,000 – 300,000 trades per day back in 2000, and probably even more of them today. One employee expressed that Medallion’s goal is the following: “[To] scrutinize historic price information to discover sequences that might repeat, under the assumption that investors will exhibit similar behaviour in the future.”
Simons understood quite early on that the stock market moves because of a complex process with many, many inputs. Some of these inputs may be difficult or even impossible to understand. They may not be related to traditional fundamentals such as earnings, dividends, interest rates or similar, but there may be some other, more obscure reason for certain moves. However, eventually, they will all be reflected in pricing data, so Simons decided to study that data. What type of human behaviour is it that Medallionis able to take advantage of? Well, to Simons, it didn’t matter as long as the patterns reappeared with a certain degree of statistical significance, but it can be interesting to speculate a little. Medallion’s profits probably stem from human biases and misjudgements.
We may have a few of the suspects in Daniel Kahneman’s famous book Thinking Fast and Slow Loss Aversion – people hate losing more than they like winning - Anchoring – one’s judgement is skewed based on previous prices and experiences And - The Endowment Effect – you like what you already have more than what is objectively sound One of the core strategies ever since the inception of the fund has been to rely on mean reversion. Early on, back in the 80s & 90s, the Medallion fund used simple linear regression models to plot, for example, the price of crude oil vs the price of gasoline. If you look at enough data points you can spot a trend line, a linear relationship between the two assets. When gasoline is cheap compared to oil, you’ll buy gasoline and short oil and vice versa. Then you wait for the prices to go back to “normal”, reverting to the mean. Today, Medallion uses a technique called “statistical arbitrage” which is about identifying a small set of quantifiable market-wide factors that best explain certain stock market movements.
If, for instance, Exxon tends to move in tandem with petroleum prices and interest rates, Renaissance identifies that. Then they bet on the stocks that have moved the least in the direction of their market-wide factors while betting against those that have moved more than the factors predicted. Again, reversion to the mean.
Today these relationships often consist of multiple variables and the relationships no longer have to be linear, so they are often difficult to identify for the naked eye. To identify such relationships the Medallion fund needed data. Mountains of data.
Takeaway number 3: Medallion requires TONS of data One of the former CEOs of Renaissance Technologies, Robert Mercer, said it best: “There’s no data like more data.” This became something of a mantra at Renaissance. Any data that could be quantified and was deemed to have some potential for predictive value was gathered. Newspaper stories, internet posts, insurance claims, nothing is too obscure. An employee named Sandor Straus noticed the need for data early on, if Simon’s wish for a fully automated algorithm was to become reality. Back then, having more data than your competitors meant buying books from the World Bank and magnetic tape from various exchanges.
The data went back as far as WW2. Straus collected more than even Simons thought was necessary, among other things, he started to collect the intraday tick prices, betting that it would become useful to them at some point. Straus even began to model data itself for a while. There were gaps in the data at certain periods due to unexpected circumstances, such as when a major flood had suspended Chicago trading. Sometimes, modelling data was possible, just like it is possible to sometimes determine the shape of a missing jigsaw puzzle piece.
Today more than 300 people are employed at Renaissance and they have more than 30 people with PhDs with the primary focus of cleaning up different data feeds so that they have the best data available for making short-term predictions. While studying this much quantitative information, Medallion must be careful as to not run into data overfitting. If you look at enough data you are bound to find some signals that seem statistically significant just by pure chance. For example, a quant investor called David Leinweber had identified that US stock returns could be predicted by combining the yearly butter production of Bangladesh, the cheese production of the US and the population of sheep in Bangladesh and the US (true story!).
For this reason, Medallion always starts to trade new signals with smaller amounts of cash, gradually ramping up the capital committed as profits roll in. Takeaway number 4: Medallion is based on advanced mathematics As I said before, this book might as well have been called “The Men who Solved the Market” as Simons definitely couldn’t do this alone. He employed various mathematicians who were specialists within their fields. Among the men who solved the market were: Lenny Baum, who was an early employee that helped Simons with something called Hidden Markov Processes. James Ax, who held the largest stake in Medallion’s predecessor, Axcom. Ax was a great algebraist and exceptional at exploring correlations.
René Carmona helped incorporate some stochastic differential equations to the models. He began applying so called Kernel Methods to analyse patterns in the data sets. He was the first one to implement a full blackbox approach, where they allowed the computer to teach itself which patterns were most important. Elwyn Berlekamp, who helped with advanced game theory. He even founded a branch of mathematics called combinatorial game theory.
Berlekamp had also worked for John Larry Kelly Jr., the creator of the Kelly criterion. The Kelly criterion determines how large or small a certain trading position should be. It can be used for value investing too, by the way, something I’ve discussed previously in a summary of The Dhandho Investor by Mohnish Pabrai. And then there were of course Robert Mercer and Peter Brown who were appointed co-CEOs of Renaissance in 2009. They both brought even more experience with Hidden Markov Processes to the group, but most of all, they were exceptional computer scientists.
Simons headhunted both of them from IBM’s former speech recognition team. Takeaway number 5: Don’t try this at home! Here’s a little thought experiment for you. Since 2003, the Medallion fund has primarily been trading in stocks, although they also do trades in commodities, currencies and bonds. Also since 2003, Medallion has returned an average of 73.7% per year, before fees,
while the S&P 500 has returned on average 10.6% per year. The returns could be explained by the fact that Medallion uses leverage, but to say that that explains the full overperformance, I think would be a little bit foolish. No, Medallion is also just on the right side of trades.
This leaves one questioning: Who does it “take” these profits from? Who is on the other side of the trade? According to a 2019 CNBC article, index investors control nearly half the US stock market. If you are an index investor and you’ve been investing without buying and selling too much since 2003, you would have received those 10.6% in average returns. Medallion cannot have “stolen” profits from the index investors, as they will get the average almost by definition. Then we have the long-term investors who invest based on fundamentals and hold over longer time periods. That’s the Warren Buffetts of the investing world. This represents an area where Renaissance and its mathematicians haven’t been able to produce any above-average profits yet.
In 2005, Renaissance founded a new fund called Renaissance Institutional Equities Fund, or RIEF, to take in outside capital without risking the profits of Medallion. Supposedly, this fund would make similar long-term predictions as the Medallion fund has been doing successfully for the short-term ones. To this day though, it hasn’t been able to do that. That’s a 9.1% return on average for all the full years since RIEF’s inception,
verses 9.9% for the S&P. The forecasting methods that Renaissance uses are similar to weather predictions – very useful for saying what is likely to happen in the coming hours or perhaps days, but not useful for longer time periods than that. Therefore, it is unlikely that it’s the acolytes of Warren Buffett who have been the pray of Medallion either.
Who’s left? It is the traders. Renaissance’s profits stem from fellow speculators, both large and small. The people who do not have a trading algorithm which trades without being| influenced by emotions, who do not have access to the same amount of data, who do not have access to some of the greatest mathematicians of our time, but who decides to take a short-term gamble in the stock market, nonetheless. The people who buy a course in day trading from a “guru” on Udemy and draw trendlines on a head and shoulder pattern. Simons came to this conclusion himself, that these must be the people who Medallion is “taking” profits from, and it is good for you to know as an investor too.
Do not think that you can do this at home. The only thing that will happen is that you’ll hand over your hard-earned money to Renaissance or some of the other quants. What was it that agent Smith said now again? “Never send a human to do a machine’s job.” So: - Jim Simons became one of the richest people on the planet through his Medallion fund - The fund is a short-term algorithm which scrutinize historic price information to discover sequences that might repeat, under the assumption that investors will exhibit similar behaviour in the future - To do this, the fund requires tons of quantitative data - Moreover, Simons needed the help of a few of the world’s greatest mathematicians - Finally, don’t think that you can do this at home. As a smaller and private investor, I’d still opt for the value investing approach of Warren Buffett over a day trading, technical analysis approach, any day.
Value investing is an area where the machines haven’t caught up to us yet. If you want to learn how to invest in a Warren Buffett way, one of the best methods for learning his secrets would be to study his most important investments of all time. In this video, I’ve summarized the 25 most important ones.
Sure, it’s a long video, but hopefully there’s a lot of meat in there. Cheers guys, hope to see you again soon!
2021-05-20