Business

“King Of Quants” Questions Machine Learning: “Can Machines Learn Without Humans?”

Eric Sorensen is one of the last standing men of his generation. As one of the early quants, Sorensen, now 71, is the oldest quant researchers running a large quantitative investment firm — one currently with over $46 billion under management. As head of Boston-based PanAgora Asset Management, he looks at trends in artificial intelligence, machine learning and other quantitative advancements and he sees repeating patterns in history. Today’s challenge, he told ValueWalk, is sifting through the hype to recognize meaningful advancements.

Eric Sorensen
mohamed_hassan / Pixabay

There is very little new theory in finance, but new techniques

“There is very little new theory,” Sorensen said, “but there are new techniques.” Take artificial intelligence, for instance. While computer-based advancements in data collection, processing and pattern recognition are changing the shape of financial analytics, they are methods to analyze information, not itself a new theory on how finance works.

Get The Full Ray Dalio Series in PDF

Get the entire 10-part series on Ray Dalio in PDF. Save it to your desktop, read it on your tablet, or email to your colleagues

Q4 hedge fund letters, conference, scoops etc

“Do you really think machines learn?” His point is that machine learning is very close to the statistical analysis used when he was pioneering the craft nearly 40 years ago, albeit with advanced tools, computational speed and much deeper levels of data analytics. But the machines themselves don’t learn.

“Machine learning was called statistics when I first started,” Sorensen said when reflecting on a life transformed much like statistical analysis has morphed with common tools such as Microsoft Excel having given way to more robust computational techniques. But like counterparts at quantitative investment firm D.E. Shaw, PanAgora has grown significantly both intellectually and in regard to assets under management.

####

As the US went off the gold standard under the Nixon administration and inflation would ultimately spike to near 23% when President Jimmy Carter took control in the White House, “there was an explosion in academic finance in 1970s,” Sorensen noted, speaking with the tone of a professor and looking every bit the part.

In 1960s and 70s, economists William Sharpe, Jack Treynor, John Lintner and Jan Mossin debated particulars surrounding the Capital Asset Pricing Model (CAPM), the Nobel Prize winning theory that measured portfolio risk and reward, it changed the way investors thought about portfolio construction. Another ground-breaking financial paragon at the time was Efficient Market Theory, championed by University of Chicago’s Eugene Fama who presumed that markets had all relevant information and were entirely efficient and thus investors should expect to do better than index averages. Fama’s theory was extended to most recently claim that financial advisors are largely unnecessary and documented market inefficiencies, such as the existence of market price momentum, was an impossibility due to market efficiency.

These two theories were diametrically opposed to what where then considered “anomalies,” studies that showed that markets can be inefficient and mis-price assets based on several factors, including information asymmetry. Sandwiched in between these competing theories was the study of econometric techniques, which would eventually define Sorensen’s quantamental approach. Econometrics uses statistical theory and mathematics to develop and explain economic theory and Sorensen would implement dramatic advancements in combining traditional fundamental analysis methods with a quantitative approach.

Ultimately arriving at this destination would not be done in a straight line, but rather he would travel down a twisting, often inefficient road to get to the point. As part of the twisted road, he built the first quantitative equity research team on Wall Street at Solomon Brothers, and would later build one of the world’s largest quantitative investment management companies.

Sorensen learned hedge fund management techniques while training high-performance jet pilots

The day he graduated undergrad in economics from the University of Oregon in June 1969, he was commissioned as a lieutenant in the US Air Force – and in the afternoon of that same day he got married. This ultimately lead to the development of thought process proclivities that appear throughout his life and shape key operating methods behind his success.

In US Air Force, he started training T38 (also called F5) high-performance jet pilots. On the surface, it may seem as if training performance jet pilots and developing one of the world’s leading quantitative trading firms may be very different. But both use precision mathematical calculations and combine the most sophisticated computational technology with the elite of elite human minds. Both trading and being in the cockpit of a performance jet require a cold, calculating demeanor under intense pressure along with the ability to quickly assimilate information and transform it into decisive action. But he also made a discover about his inner nature that would likewise shape his life.

“I like to teach,” Sorensen realized. “There rules and limits and you have to be able to navigate within them,” he said when considering commonalities between a high-performance supersonic jet pilot and investment manager. He would train a pilot, watch how they learned and managed situations, then let go. “I learned about empowering people. You make judgements and decisions (about students). They must be able to do it themselves. When I got onto Wall Street I did the same thing.”

But before he would enter the capitalistic zoo, he would first study it.

From 1974 to 1986 he completed his Ph.D and relatively soon became Professor of Finance and Department Head at the University of Arizona where he looked at economics with a mathematical mind. Following this mindset, he specialized in the econometrics of asset pricing, publishing dozens of journal articles. Ultimately, the siren song of Wall Street beckoned.

“Academics are about solving problems,” he noted, pointing to a purpose than just studying issues. He was among the first academics to take a quantimental view of the market literally decades before University of Chicago professor Lin William Cong coined the term. It was at this moment Sorensen realized he wanted more than just understanding how the financial engine worked.

####

In the 1980s President Ronald Regan’s “Morning in America” gripped the nation just as the lead character in the movie “Wall Street,” fund manager Gordon Gecko, extolled the financial world to think “greed is good.” This is when Sorensen made his move into finance, but it wasn’t due to greed. “When (academics) came into finance, they realized they could conduct research, solve problems and make money at the same time. It was a perfect fit.”

He started as an equity analyst and between 1986 and 2000, Sorensen rose to become the Global Head of Quantitative Research at Salomon Brothers and Citigroup CIB. It was here he first found opportunity to apply econometrics towards profitable research. Years later after taking the helm of PanAgora, he encouraged his colleague George Mussalli, the CIO, Equity Markets, to look at the information used to analyze bank earnings because more could be done. The team approached the Federal Deposit Insurance Corporation (FDIC) and asked for all the information they had on the banks. To their surprise, reams of paper came back, which they quickly engineered receiving it digitally and then conducted analysis to find patterns that gave them an investment edge, making PanAgora’s equity team among the first to discover truly alternative data. He would also be first to discover the common limits, when eventually the alternative information was more widely distributed and then it became less valuable.

He applied a quantimental view to all meaningful aspects of a stock, digitizing and ranking everything from the value a member of the board of directors has on the company to key executives and the firm’s moat and distinct market advantage. Sorensen was starting to use statistics to solve a host of financial problems but at the time “quant” was a dirty word. “People would say, ‘You are a quant. You are a black box. You overfit.’ In the 1980s these were derogatory terms.” But that wouldn’t last long. “Today they are glamour terms.” Near this time Eddie Qian, the firm’s multi-asset CIO, developed and coined the term “risk parity” but “Bridgewater was better at marketing it,” Sorensen says.

After Salomon in 2000 he served as Partner and Chief Investment Officer, Structured Equity, at Putnam Investment Management and ultimately took charge at PanAgora in 2004 and would be called the “King of Quants” by Pension & Investments magazine in 2007, just before the global financial crisis would create the ultimate economic anomaly. From that point of being called “king,” assets under management would swell by more than 300% and the world that started to unfold near the turn of the century would finally sprint into a full embrace of quantitative finance. As he looks on today’s market panoply, with 10 PhDs and even a medical doctor on the staff looking for investment ideas, what is most interesting is not new theoretical thinking but the tools and resources. “Data sets today are extraordinary,” he said, remembering his early days when PanAgora’s FIDC data would be discovered by others, thus reducing effectiveness. Today he sees a larger investment world where crowding occurring with too many hedge fund managers going after a limited number of opportunities, but they do exist.

“In last few years, big data -- or smart data, as I like to say -- has exploded,” he said. “There are opportunities if you can get better data.” But when those opportunities are crowded, they become less valuable. For Sorensen, quantitative, computer-based systems should never run exclusive on autopilot. Finding those unique opportunities will take creativity and combining the best and brightest humans with computers is an area where supply will always be relatively limited, but demand will likely never cease.

This article first appeared on ValueWalk Premium