New Frontier’s Richard (left) and Robert Michaud have abandoned their family chess game in pursuit of global asset management
Image: Michael Prince for Forbes
Among academicians, few have made a real-world impact as far-reaching as Harry Markowitz, the father of modern portfolio theory. Markowitz devised his seminal theory as a 23-year-old PhD student at the University of Chicago in 1950 and won a Nobel Prize for it in 1990. For generations, it has guided money managers in picking stocks that have the highest possible returns given a set of risk assumptions. In fact, Markowitz’s mean-variance optimisation, also known as the “efficient frontier”, is embedded in most software that financial advisors use to create portfolios today.
But there’s a big problem lurking at its core: The mathematical tools Markowitz used are actually too precise, so in the real world of stocks and bonds—where information is much blurrier—it occasionally produces wild outcomes.
Enter New Frontier Advisors, a PhD-heavy Boston-based asset manager that has been using low-cost exchange-traded funds to one-up Markowitz. For more than a decade, its clients have beaten equity- and bond-market benchmarks by 100 to 200 basis points after fees, and assets have swelled to almost $3 billion.
“Markowitz’s theory makes all kinds of sense, consistent with informed investor behaviour,” says New Frontier founder Richard Michaud, 75, a Boston University-educated mathematician. “But when you try to use it, you can’t.” Michaud cites the bizarre portfolios that sometimes result from a pure Markowitz equation. The Nobel laureate’s efficient-frontier math, for example, might produce a portfolio with a 25 percent position in Brazilian stocks and 30 percent in microcaps, since the assets are uncorrelated. But most advisors would consider such heavy allocations crazy. To fix this problem, analysts place limits on the percentage of any security type in the portfolio. And that defeats the purpose of using Markowitz’s elegant theory in the first place.
“Almost all investment-management software has a dirty little secret underneath: It’s not actually letting the theory dictate the portfolio,” says Richard’s son Robert, 44, who started coding for his father at the age of 12 and dropped out of a doctoral programme in finance at UCLA’s Anderson School to join New Frontier in 2001. “It’s forcing the user of the software to tell the software what portfolio they wanted to get anyway.”
Instead of simply plugging reliably unreliable estimates into the computer, the Michauds layer on thousands of Monte Carlo simulations—a mathematical technique that uses randomness to create a cloud of possible outcomes for individual stocks and collections of securities. Then they use more advanced math to find the portfolios expected to perform best across alternative future outcomes. This forms the basis for a new set of optimal risk-adjusted portfolios.
Michaud started probing the flaws in Markowitz’s theory as an analyst in the 1970sNot everybody agrees with the Michauds’ methods. “What they’re doing is in theory wrong,” says John Liechty, a professor at Penn State. But even Markowitz concedes New Frontier’s potential. “I conjectured that I could produce a mechanical methodology that would beat Michaud at his own game. But he won,” says Markowitz, 89, who considers Michaud a friend despite their long-running disagreements.