This article originally appeared in Institutional Investor on October 26, 2017.

“The problem with modern portfolio theory is that it’s not very modern,” says one endowment chief.

If you listen carefully to what the leading investors are saying about asset allocation, you’ll hear a subtle shift in thinking.

Asset allocation is commonly said to be the most important determinant of investment results. Yet thought leaders I’m speaking to are moving away from the confines of asset classes and are embracing idiosyncratic manager risk — a shift that is heresy to conventional thinkers in this industry and antithetical to the wave of inflows to passive vehicles.

Today’s investment portfolios tend to get organized by either assets or risk factors. Allocation by assets became the primary modus operandi following the release of Yale University endowment chief David Swensen’s seminal tome on the subject in 2000. He provided the research and logic necessary for governance boards to broadly adopt well-diversified, multi-asset-class portfolios that extended beyond the standard stock/bond mix at the time. The subsequent market selloff from 2000 to 2002 offered ex post evidence of the success of the endowment model, and most institutions were off to the races.

Now factor exposures have come into vogue, following academic research outlining the long-term outperformance of small, value, momentum, and quality factors, as well as the development of cost-effective products to express such factors.

A small subset of allocators, including Verger Capital CEO Jim Dunn, have eschewed asset allocation in favor of risk allocation. Dunn explained on a recent podcast that “the problem with modern portfolio theory is that it’s not very modern.” His approach analyzes the factors that drive a manager’s returns and optimizes across factors rather than across asset classes. The resulting portfolio tends to look quite different from one that starts with allocations to asset classes and fills buckets and sizes managers accordingly.

These core approaches are not mutually exclusive. Allocators employing an asset allocation framework spend a lot of time calculating underlying risk factors, ensuring they are comfortable with the aggregate exposure created by their managers. Similarly, risk-based allocators also look top down to ensure the boat isn’t tipped too far toward a single asset category.

Back in 2003, the late Peter Bernstein wrote that asset prices were sufficiently expensive that holding diversified, long-term allocations in a policy portfolio would be insufficient to meet the spending needs of institutions. He hinted at opportunism or market timing as potential necessary evils to fill the gap.

Looking back 14 years later, Bernstein was either very wrong or very early. While the 2008 crisis briefly gave credence to his beliefs, the subsequent near-decade-long bull market has rewarded those holding fast to asset-allocation-based policy portfolios. In fact, the basic 60/40 stock/bond asset allocation mix has significantly outperformed more modern asset allocation and risk factor frameworks. As a result, active managers and allocators alike have never been more scrutinized.

Ever the contrarians, leading allocators increasingly share Bernstein’s earlier outlook following this prolonged period of market strength. Top allocators don’t believe that passive exposure to diversified asset classes will get them where they need to go. That necessitates a change in thinking.

In our conversations, long-standing chief investment officers Andrew Golden at Princeton University and Scott Malpass at Notre Dame both spoke about loosening the bonds of asset classes in favor of the dogged pursuit of alpha.

For example, the largest single allocation in the Princeton endowment is with a manager of a highly concentrated portfolio of fallen-angel biotech stocks. At 4 percent of the endowment and 40 percent of its U.S. equity portfolio, the individual manager’s idiosyncratic returns will drive Princeton’s results far more than its asset allocation to U.S. stocks or the growth-risk factor created by the exposure.

Notre Dame has simplified its asset class mix to just three — public equity, private equity, and opportunistic — and is letting its best manager’s ideas fill the portfolio bottom up. When Princeton and Notre Dame shift their emphasis toward the unique skills of managers and away from the buckets in which those managers participate, they bet on people and skills far more than on assets or risk factors.

These prominent investors contend that alpha generation by their managers will be required to make up for deficits left by broad asset classes and risk factors in the current market environment. The asset allocation structure remains a piece of the puzzle, but one that matters more for governance and communication than for driving results.

This manager-centric approach directly contrasts with the surge in fund flows toward indexation and the massive scrutiny of fees charged by active managers. Passive investment emphasizes low-cost ownership of the very assets leading allocators are deemphasizing in their thought process. Leading thinkers are willing to pay for perceived talent, and are tweaking their huge pools of assets accordingly.

Average allocators will still lose to the market return by the amount of fees they pay, but does that mean they should give up trying if they believe the market return won’t be sufficient to pay the bills?

Only one side is likely to be right. My money is on these leading allocators.