Since early this year, I've been arranging the sessions for my upcoming Business & Wealth Management Forum, coordinating presentations by industry heavyweights Bill Bengen, Harold Evensky, Don Phillips, Michael Kitces, Mark Tibergien and-well, you get the idea. The goal is to get all of these deep thinkers to go a little deeper than you typically see at these gatherings and explore topics you don't hear enough about. Along the way, I've found there's still a lot of unexplored terrain in our professional body of knowledge.

For instance? Investing, where modern portfolio theory is not very different in the way we apply it than it was when Dwight Eisenhower occupied the White House. How is that possible? Unlike Harry Markowitz when he was writing his seminal research, advisors today live in the Information Age. We know things Markowitz only suspected: that during normal markets, correlations among assets tend to move around unpredictably within a surprisingly tight spectrum of numbers, and then behave very differently whenever investors are paralyzed with fear or excited. Interestingly, the same is true about measures of investment volatility.

We also now know that there is a significant connection between current valuations and long-term future returns. As Roger Gibson (profiled on page 57) has shown, future performance has been much less exciting when we invest during a period when P/E ratios are high than when they are low. When people bravely invested during periods when the S&P 500's P/E10 (the normalized price/earnings ratio over the past 10 years) was below 12, their average annual return over the next 10 years was a princely 14.88%. Investors who put their money into the index when its P/E10 was more than 20 received an average of just 4.68% a year over the ensuing decade. The initial P/E ratio seems to be telling us a lot more about expected returns than the historical averages.

But how do we capitalize on this information? One of the conference presenters I've been talking to, Jerry Miccolis of Brinton Eaton Wealth Advisors, spent 25 years working at Towers Perrin before he moved into the financial services world. As a consultant to business managers, he routinely used the mathematics of something called enterprise risk management, which is the corporate management equivalent of modern portfolio theory. In fact, enterprise risk management was created from the work by Markowitz. Today, it's used by corporate managers and consultants to evaluate the real-world riskiness of "portfolios" of a company's divisions or product lines. But unlike modern portfolio theory, enterprise risk management has been evolving along with new computer systems and mathematical tools.

As an example, Miccolis says no corporate manager would measure risk as a single static measure in a dynamic marketplace. "In the corporate world," he says, "there is broad understanding that the degree of risk and the nature of the risks change as the marketplace changes. And when you look at investments, what do you find?" he asks rhetorically. "Exactly the same thing."

He was taken aback when told a single number normally describes the correlation of price movements between, say, large-cap U.S. stocks and emerging-market bonds. Why would modern portfolio theory inputs use single numbers for all the correlations when they clearly move around, and converge when markets do what they did in 2008? A corporate manager with enterprise risk management training is naturally curious about how the correlations between different product lines will change over time, and what market conditions have, in the past, caused which kinds of changes.

Miccolis is attempting to translate enterprise risk management mathematics into a more sophisticated version of modern portfolio theory and discovering that our investing world has developed a lot of strange taboos. "In the investment markets, we are learning that when volatility rises, it tends to stay high-what they call volatility clustering," he says. "If portfolios are entering a period of high volatility, one might respond by taking some risk off the table." In the enterprise risk management world, that is common sense. In the investment world, such behavior is labeled market timing.

Miccolis believes we can do a better job of forecasting expected equity and bond performance, and create modern portfolio theory models that use a spectrum of numbers rather than a single expected return. For our volatility and correlation inputs, we can use models instead of numbers, and the models would distinguish between normal markets and periods like the fall of 2008, when everybody is trying to unload their investments in a panic. The goal is to build sturdier portfolios, and also make constant adjustments, just the way a corporate manager or consultant would.

Are there other holes in our body of knowledge? Recently, we've seen data showing the efficient frontier is kind of a silly idea. If you draw one for every decade, using retrospective return and volatility numbers, you get a whole bunch of fishhook graphs in different parts of the risk/return space, with different shapes and slopes. Invest precisely along last decade's frontier and you're likely nowhere near the one that fits the current decade.

Clearly, either the frontiers or the markets aren't as efficient as they pretend to be.Is there a systematic way to push a diversified portfolio above that moving target?

The most interesting thinker I've found on this issue is Gary Miller of Frontier Asset Management, who's co-credited with inventing factor analysis (along with Nobel laureate Bill Sharpe). Miller has created a remarkably geeky empirical approach to building portfolios, which allows (in a crude oversimplification) millions of different combinations of investments to "compete" with each other for supremacy on his computer.

The process starts with factor analysis, which (once again, I'm simplifying) regresses the daily or weekly movements of each fund manager's portfolio with different combinations of assets. When Miller finds a very close match, it means he's found something close to the precise asset mix the manager is investing in. He does this for many thousands of managers, then looks for the few whose fund has outperformed the index returns for their asset mix. As Milleradmits cheerfully, it doesn't matter why they've consistently bested this customized index. "If we pick you to input into our portfolio-creating process," he says, "it means you are doing something better than anybody else, even if we're not quite sure what that is."

Then Miller creates forward-looking real return expectations for 16 different asset classes, which takes into account current valuations in an interesting way: The higher the valuation, the more assumed downside risk. An algorithm adds the various risks under normal markets so that clients, depending on their risk tolerance, will get a higher or lower downside risk number, which changes month to month according to changes in valuations and shifts in a rather complex market anxiety measurement. Instead of using historical numbers, Miller inputs correlations taken from periods of major market movements. The idea is to find the current efficient frontier by a process that might be described as blindly groping for it initially and refining as you go.

Then, in an effort to punch through the efficient frontier into higher risk/return territory, Miller inputs the best 120 funds he finds in the factor analysis process and creates what he describes as "a zillion" different combinations. The computer sorts the combinations in up, down and in-between markets, looking for the best aggregate returns over all three-returns that look like they can beat the underlying asset allocation index performance.

Miller sees this as a fancy version of outsourcing. "We don't pick stocks or bonds, and we don't try to evaluate the market's short-term movements," he says. "We try to do a great job of asset allocation and identifying smart managers, and then we let them do what they do best. What should our growth/value orientation be? We let the managers tell us and be part of the process."

There's certainly a lot more to this, which I plan to explore in detail when our conference attendees get face-to-face with Miccolis, Miller and a host of other thinkers in the investing, practice management and technology spaces. Do these represent better ways to build portfolios than what we have now?

Maybe yes, maybe no. But even if they don't, Miccolis and Miller and a host of others are opening up very interesting possibilities that we, as a profession, need to investigate on behalf of our clients. I'll be taking a lot of notes, and I hope that sometime next month I'll be a lot smarter than I am now.

Bob Veres writes and publishes the Inside Information service at For details about the Business & Wealth Management Forum, set for Oct. 13-15, visit