The Struggles of Quantitative Investing
When being smart and skilled is not enough.
The Opening Bid
The only truly quantitative course that I took in business school, a seminar that compelled its attendees to write computer programs to evaluate financial-markets data, was filled with ardent engineers. Never mind those pesky accounting classes or, God forbid, marketing seminars that required tedious essays. This was real analysis! My classmates attacked their assignments, convinced that by the quarter's end they would discover a market anomaly.
Once upon a time, such aspirants would have been rare. In the early days of investment analysis, the field attracted the sort of people who excelled at contract bridge. (Today, the game would be poker.) They had a good head for numbers, but they weren't mathematicians. More important than calculus was the ability to understand income statements and industry trends.
So described Ben Graham and John Templeton, and so still describes Warren Buffett (who remains an avid bridge player). In 1963, when a group of investment professionals created the initial Chartered Financial Analyst exam, they designed the test to identify those who owned such skills. Receiving the CFA designation required thorough subject-matter knowledge, but only eighth grade math.
The Numbers Revolution
Gradually, things changed. Thanks to the spread of computers, investment researchers could collect and process much more information. They began to search for investment patterns that could be exploited. The natural starting point was technical analysis, which evaluated the behavior of stock prices. It was much easier to collect and analyze daily stock prices than it was to sift through cumbersome, and difficult to obtain, fundamental evidence. From thence were born "head and shoulders" charts, along with various other curiosities.
Admittedly, identifying (mostly spurious) patterns didn't require advanced degrees. The next development did, however: quantitative investing. In the 1980s, big data arrived. Initially among institutional firms, and then through mutual funds, came strategies that crunched unprecedentedly large numbers. Extracting the mountain's lessons was not a task for conventional portfolio managers. Investment organizations began to hire "quantitative" analysts.
As company analysis became more structured, so as well did asset allocation. Professional investors began to employ optimization models. Some also offered "tactical asset allocation" portfolios, which applied science (or at least the attempt) to the long-standing practice of market-timing. No longer would portfolio managers' decisions be guided solely by experience and instinct. They would now be supported by rigor.
Decidedly Mixed Results
The benefits of these innovations were fleeting. The early quantitative investors did fare well. When I joined Morningstar, in 1988, my fellow researchers informed me that investment companies that had invested heavily in technology, and that had built major databases, possessed a competitive advantage over their slower-moving peers. That was indeed the case--until it was not. By the early 1990s, the quants had regressed to the mean. They had lost their edge.
Whether the first technical analysts profited from their work, I cannot say. By the late '80s, their time had come and gone. However, I can attest to the failure of tactical asset allocators. With rare exception, their funds were never good. Currently, allocation funds that contain the word "balanced" in their names hold more than $500 billion in assets, while those that call themselves "tactical" have only $10 billion. The discrepancy is justified; over the past decade, the balanced funds have gained 1.7 percentage points more per year, while being less volatile.
In recent years, several investment companies have adopted artificial intelligence routines. (To my knowledge, the first public fund to do was Fidelity Disciplined Equity, all the way back in 1988. The fund came out of the gate well, but after a few years it too came back to earth.) While Morningstar doesn't formally track which funds use artificial intelligence, it's obvious from the U.S. equity fund leaders list that the tactic hasn't been notably successful.
Quantitative investors have faced two challenges. One is that their reach sometimes exceeds their grasp. Predicting short-term movements of the global financial markets, including equities, bonds, and currencies, may simply be past the ability of any investment routine, no matter how sophisticated. Tactical allocators, to cite the most prominent example, may never be good enough to complete their assigned duties.
The bigger problem, though, is competition. Each year, U.S. universities grant an aggregate 25,000 Ph.D.s in the fields of finance, engineering, physics, mathematics, and computer science. Most of those graduates don't become professional investors--but if only 1% of them do, that makes 250 each year. Over time, that number swells to several thousand quantitative hopefuls, all vying for their investment fortunes.
That is too much company. To earn high grades in my MBA course, the engineers needed only to beat out the seminar's lightweights. In the investment business, though, outdoing the also-rans is insufficient. To earn excess profits, quantitative managers must find opportunities that elude even those with similar training. They are unlikely to achieve that goal by writing better equations. Most likely, their results will depend upon obtaining an informational advantage.
Which is very difficult to accomplish these days, with seemingly every detail of companies, the economy, and the financial markets recorded on the internet cloud. To be sure, some prominent quantitative investors do gather proprietary information, which they then feed into their models. But they typically do so as high-frequency traders, moving into and out of positions in micro-seconds. Because of regulations, such strategies cannot be used within publicly available funds.
The accomplishments of high-frequency traders, as well as of other investment approaches that can be regarded as "market making" (as with Ken Griffin'sCitadel organization and Renaissance Technologies' Medallion Fund), demonstrate that quantitative investing can be profitable. However, whether the technique can be harnessed for retail investors is unclear.
A few years back, I thought that artificial intelligence might break through. So far, though, that has not been the case. While artificial intelligence has dramatically improved chess programs, the investment markets behave differently than a board game that has fixed rules and no element of chance. As with ordinary, human-inspired intelligence, the artificial version thereof may require special information to succeed. Great thoughts alone may not suffice.
John Rekenthaler (firstname.lastname@example.org) has been researching the fund industry since 1988. He is now a columnist for Morningstar.com and a member of Morningstar's investment research department. John is quick to point out that while Morningstar typically agrees with the views of the Rekenthaler Report, his views are his own.