The Greatest Geometric Balancers: Renaissance Technologies, Part I

Renaissance Technologies is the ultimate Rorschach test in the investing world.

Factor investors think they are the best at implementing factors.  Machine Learning advocates think they are the best at applying machine learning.  Day traders think they are the best day traders. Trend followers think they are the best trend followers. Operations people think they hire the best employees.  Mathematicians think its because they hire mathematicians.  Managers think it’s their culture of sharing.¹

Everyone thinks Renaissance Technologies (nicknamed RenTec) does what they do, they just think they do it better.  Of course I’m not immune to this illusion either. 

I think Renaissance Technologies is the greatest Geometric Balancers.

The Man Who Solved The Market

Renaissance Technologies’ investing approach sometimes confuses people. They go against typical investing recommendations and practices. However, these practices don’t seem strange to me and I’d like to explain why through quotes from Gregory Zuckerman’s book The Man Who Solved the Market. This post is my interpretation from various public sources of information about what makes the company so successful.

What is Renaissance Technologies?

Renaissance Technologies was founded in 1978 by James Simons.² The firm has gone on to become the greatest investment company the world has ever seen.

Since 1988 Renaissance’s flagship Medallion hedge fund has generated average annual returns of 66 percent, racking of trading profits of more than $100 billion. No one in the the investment world comes close. Warren Buffet, George Soros, Peter Lynch, Steve Cohen and Ray Dalio all fall short.

Gregory Zuckerman, The Man Who Solved the Market, page xvi.

Jim Simons

Jim Simons is a world class mathematician known for creating the Churn-Simons form. He spent the early part of his career as a code breaker for the Department of Defense during the cold war. Essentially he was trying to extract useful information out of seemingly random signals. Later he lead the mathematics department at Stony Brook University.

Simons came from a different world and enjoyed a unique perspective. He was accustom to scrutinizing large data sets and detecting order where others saw randomness.

Gregory Zuckerman, The Man Who Solved the Market, Chapter 3, page 44

But Simons always had an investing bug.

Leonard Baum

While code breaking, Simons worked with mathematitian Leonard Baum. He co-developed the Baum-Welsh algorithm which uncovers hidden order in series of information.

The Baum-Welsh algorithm proved a way to estimate probabilities and parameters within these complex sequences with little more information than the the output of the process.

Gregory Zuckerman, The Man Who Solved the Market, Chapter 3, page 47

Simons felt Baum was the ideal partner to help him break the code of investing. So in 1978 they began trying to solve the markets together.

Early Efforts

Skipping over all sort of interesting detail which you should really read in the book, Simons and Baum realized that investing markets were similar to cryptography. They had plenty of experience exploring random looking communication signals and using their equations to extract order out of the confusion.

So Simons and Baum set out to use the same tools to uncover that order in investing markets and profit from it. They began with currency and commodities markets.

It worked, kind of…

RenTec spent a lot of time gathering mountains of data and analyzing that data with their sophisticated predictive models.

Many of the tactics they tried focused on various momentum strategies, they they also looked for potential correlations between commodities.

Gregory Zuckerman, The Man Who Solved the Market, Chapter 3, page 57

Generalizing, their strategies worked most of the time. But sometimes they didn’t, and when they didn’t the drawdowns were painful. They had some good years and some bad years. Their overall performance was good, but nothing extraordinary.

I suspect their predictive methods were useful, but they clearly were not enough to generate consistent profits.

Baum ultimately retired, but they they kept poking at investing, hiring more brilliant mathematicians, gathering more and more data, and further refining their systems.

In 1988 they launched the infamous Medallion Fund. It didn’t go well.

Medallion Fund.

After 16% returns their first year things turned south in 1989. The fund fell 30% from its peak. Investors were concerned. Employees were worried the firm would close.

Their fund had been losing money for months and was now down nearly 30% from the middle of the previous year, a staggering blow.

…

Simons was spending more time on his various side business than he was on trading: his heart didn’t seem to be in the investment business. Strauss and his colleagues become convinced Simons might shutter the firm.

“It wasn’t clear Jim had any faith,” he says, “And it wasn’t clear if we would survive or fold.”

Gregory Zuckerman, The Man Who Solved the Market, Chapter 5, page 100-101

I believe the key to understanding what makes RenTec so successful today is understanding what turned them around at this time.  Before 1989, they didn’t have spectacular performance. They struggled at times and were very close to closing. Their models to predict future returns simply were not enough to generate consistent profits. There was nothing overtly special about their operation.  But in 1989 that changed and the company never looked back.  So what caused this sudden reversal of fortune?

In 1989 Elwin Berlekamp took over the investment decisions of the Medallion fund. 

Elwin Berlekamp

Elwin Berlekamp studied electrical engineering under Claude Shannon at MIT.  During the summers he worked at Bell Labs under John Kelly, creator of the Kelly criterion.

Elwin Berlekamp comes from the Shannon-Kelly tree of investing and would have been well versed in the concepts I write about on this blog.  He understood how to size bets to maximize geometric returns. 

Kelly’s formula had grown out of Shannon’s earlier work on information theory. Spending evenings at Kelly’s home playing bridge and discussing science, math and more. Berlekamp came to see the similarities between betting on horses and investing in stocks, given that chance plays a huge role in both. They also discussed how accurate information and properly sized wagers can provide one with an advantage.

Kelly’s work underscored the importance of sizing one’s bets, a lesson Berlekamp would draw on later in life.

Gregory Zuckerman, The Man Who Solved the Market, Chapter 5, page 92

Now like Kelly and Shannon (and Ed Thrope), he went into academia and become a professor at the University of California. In 1988 Simons invited Berlekamp to visit his trading division, called Axcom at the time.

Berlekamp’s Recommendations

Simon’s asked Berlekamp to review their trading strategy.

For all the brainpower the team was employing, and the help they were receiving from…others, Axcom’s model usually focused on two simple and commonplace trading strategies. Sometimes it chased pries or bought various commodities that were moving higher or lower on the assumption that the trend would continue. Other times, the model wagered that a price move was petering out and would reverse, a reversion strategy.

Gregory Zuckerman, The Man Who Solved the Market, Chapter 5, page 95

So for all their fancy equations, Renaissance was pretty much just trend following and mean reverting.

After looking over the strategies, Berlekamp made various recommendations which were ultimately ignored.

Berlekamp began sharing his suggestions. He told Ax that Axcom’s trading models didn’t seem to size trades properly. They should buy and sell larger amounts when their model suggested a better chance of making money. Berlekamp argued precepts he had learned from Kelly. “We ought to load up here,” Berlekamp said one day.

Gregory Zuckerman, The Man Who Solved the Market, Chapter 5, page 96

When the odds are strongly in your favor, you strike big. It’s a statement on bet sizing within the Kelly criterion.

This is also the only time I saw one of Berlekamp’s recommendations directly linked to Kelly, but as as were about to see, many of them have their roots in Kelly.

Shorten Holding Periods

Elwin looked at other aspects of the trading strategy, and the recommendations were also ignored.

Berlekamp discovered other problems with Axcom’s operations. The firm traded gold, silver, copper and other metals, as well as hogs and other meats, grains, and other commodities…..and Axcom often held on to investments for weeks or even months at a time.

That’s a dangerous approach, Berlekamp argued, because markets can be volatile. Infrequent trading precluded the firm from jumping on new opportunities as they arose and led to losses during extended downturns. Berlekamp urged Ax to look for smaller, short-term opportunities–get in and get out.

Gregory Zuckerman, The Man Who Solved the Market, Chapter 5, page 97

There are three key points here. One, markets change. The market properties today may not be the same a month later. You are far more likely to develop accurate “measures” of the market if you take them frequently and act on them frequently.

Its easier to predict tomorrow than a month from now.

Second, Berlekamp said holding onto an investment for a long time was dangerous “because markets can be volatile”. This is a nod to volatility drag. Over time the compound growth of a portfolio will degrade. The longer you hold a volatile investment, the further it will degrade.

How do you limit this volatility drag?

Rebalance As Often As You Can.

Third, the other reason for shortening holding timeframes comes from the rebalancing benefit. You want to rebalance as often as you can. Anyone building an investing system directly from Shannon and Kelly’s own ideas is going to know this.

The surface of returns is the highest when rebalancing as frequently as possible (see this post).

In investing markets, you accomplish this by shortening your holding periods. Shorter holding periods give the investment less time to experience volatility drag. After each holding period, you sell, and then purchase the next, properly balanced, portfolio.³

Shorter holding periods allow you to “rebalance” more often, and therefore receive the maximum compound growth rate.

Strange for Trend Following

One confusing thing here is that shortening timeframes doesn’t mesh with the traditional views of trend following. Trend followers believe in letting the winner run. Hold them for as long as you can and don’t sell them. Berlekamp reviewed RenTec’s strategies, saw that they were trend following, and then recommended they shorten holding periods. “Get in and get out.”

This view is violently opposed to the idea of letting the winners run. The entire concept of buying and selling more frequently to shorten holding periods–and routinely selling winners–is still not well understood, but it’s a foundational idea of the greatest investment firm ever.

Berlekamp Comes Onboard

These recommendations were ignored, and as mentioned earlier, soon after the fund fell 30% from previous highs. But Berlekamp thought he could fix their issues with his recommendations, so he bought into the firm and took over the investment responsibilities.

Berlekamp hadn’t worked on Wall Street and was inherently skeptical of long held dogmas developed by those he suspected weren’t especially sophisticated in their analysis. He advocated for more short-term trades…

Their goal remained the same: scrutinize historic price information to discover sequences that might repeat, under the assumptions that investors will exhibit similar behaviors in the future. Simons views the approach as sharing some similarities with technical trading. The Wall Street establishment generally viewed this type of trading as something of a dark art, but Berlekamp and his colleagues were convinced it could work if done in a sophisticated and scientific manner–but only if they focused on short term shifts father than longer term trends.

Gregory Zuckerman, The Man Who Solved the Market, Chapter 6, page 108

A few key points here.

One, much respect for being skeptical of long held wall street dogmas.

Two, I believe most people interpret the part about using historic prices to discover repeating sequences as a reference to returns. It may be. But I strongly suspect it’s just as much a reference to volatility and especially correlation repeating in the near term.

There is far more evidence that historic price information predicts future volatility and correlation than it does prices. Since Berlekamp’s philosophy came from Shannon and Kelly, he knew the importance of volatility–and in a portfolio, correlation–to compound growth. Therefore, don’t just take this statement as being only about returns; it’s likely about volatility and correlation too.

Three, the part about doing it “in a sophisticated and scientific manner” is clearly about sizing investment with the Kelly criterion and not just “feeling out” how much to invest and for how long.

Law of Large Numbers

Berlekamp followed his recommendations and shortened the holding periods by increasing trading frequency to tighten up the variability of the average compound growth rate they received.

Berlekamp also argued that buying and selling infrequently magnifies the consequences of each move. Mess up a couple times and your portfolio could be doomed. Make a lot of trades however and each individual move is less important, reducing a portfolio’s risk.

Berlekamp and his colleagues hopped Medallion could resemble a gambling casino. Just as casinos handle so many daily bets that they only need profits from a bit more than half of those wagers, the Axcom team wanted to fund their trade so frequently that it could score big profits by making money on a bare majority of its trades. With a slight statistical edge the law of large numbers would be on their side, just as it is for casinos.

Gregory Zuckerman, The Man Who Solved the Market, Chapter 6, page 108

Now this is key: this is a justification for buying and selling more frequently. Its an argument about diversifying through time. The casino reference isn’t about simultaneous bets, its about bets in series.

The law of large numbers argument doesn’t apply through time, unless you are focusing on the geometric average. The law of large numbers doesn’t apply to absolute returns when gains compound through time.⁴

110% * 90% = 99%, not 100%

Berlekamp would have known this. This is pure Shannon/Kelly philosophy.

How many people do you think read Zuckerman’s quote and think: “Yes lots of simultaneous bets just like a casino, and just like typical diversification recommendations.” But this is a very different concept about creating multiple bets through time, and it’s not a typical diversification statement.⁵ It’s a statement about how to maximize compound growth through rebalancing.

And of course the reference to a casino is another hat tip that Berlekamp is applying a Kelly’s gambling and betting framework to maximize geometric growth.

It Worked

Berlekamp’s strategy worked from day one.

The firm implemented it’s new approach in late 1989 with the $27 million Simons still managed. The results were almost immediate, startling nearly everyone in the office. They did more trading than ever, cutting Medallion’s average holding time to just a day and a half from a week and half, scoring profits almost every day.

Gregory Zuckerman, The Man Who Solved the Market, Chapter 6, page 113

Medallion scored a gain of 55.9 percent in 1990, a dramatic improvement on its 4% loss the previous year.

Gregory Zuckerman, The Man Who Solved the Market, Chapter 6, page 115

This quote undersells their excellence. The 55.9% gains is after fees. Before fees, they produced 77.8% return. And you can see Berlekamp dug the fund out of its deep losses in 1989 as well.

Even though Elwin Berlekamp retired soon after 1990, he put the foundation in place, and Renaissance never looked back. Elwin Berlekamp merged Simon’s cryptography skills with Shannon’s demon and the Kelly criterion to shape the greatest investment strategy the world has ever seen.

In Part II we will examine how RenTech evolved further and read between the lines in the challenges they overcame to learn more about the nature of their strategies. There’s plenty more as we’re only halfway through the book.

Footnotes:

1-I think the management skills of James Simons are very underrated. He’s led this company for over 40 years, with a changing cast of characters through the first half. It takes enormous skill to do that and essentially bring out the best in your employees. You can see it too in his leadership of the math department at Stony Brook. I find most of the reasons people give for RenTec’s greatness exaggerated, but not Simon’s management prowess.

2-They changed the name from Monemetrics to Renaissance Technologies in 1982.

3-Now you don’t actually have to actually sell if you still want to hold the asset. You can simply sell (or buy) it back to the proper portfolio allocation, which is essentially what I do.

4-This is because the returns are skewed when they compound. And when the become skewed, the arithmetic return doesn’t exist.

5-This is also essentially the key point Ergodicity Economics is trying to put back into economics.