Much will stay in money market accounts for years, and some will go to bonds. But at some point, equity will seem attractive again and the dollars will come flooding back. And it makes sense, even for mechanical investors, to think carefully about how to build better, more diversified portfolios than the ones they started with before the recent rout.

For the past few years, I’ve been happy to let the SuperModel screens do the heavy lifting on diversification. From 1998 through 2000, that meant almost no diversification at all, and the SuperModel portfolios have ended up looking largely like thinly disguised technology growth-stock funds. To be sure, there were different flavors of tech, like software, data storage and networking -- but those are distinctions without a difference when the market decides it hates one big group of stocks. Very few names from the drug, utility, energy or consumer cyclical sectors have topped the screens long enough to enter our portfolios and round out performance.

If anything then, the recent unpleasantness has cast into sharp relief the potential benefits of purposefully spreading the risk of a portfolio out among sectors and styles whose expected returns have little correlation to one another. But how do you do that and stay mechanical and unemotional? Mechanical-investing orthodoxy, as passed down by experts like author/money manager James O’Shaughnessy, essentially prohibits messing with the screens.

Another way to diversify

Carr Bettis, a leading quantitative stock researcher, finance professor and hedge fund manager based in Scottsdale, Ariz., suggested one way out of the dilemma. He suggested that a statistically valid methodology for diversifying the SuperModel portfolios would be to take the first three or four names from the Redwood and MVP Growth results from the predominant sector, and then “skip down” the lists, in order to get a more representative sample of names. For instance, if the first seven Redwood Growth names are techs followed by companies from the utility, health-care, retail and energy sectors, then he suggests you would take the first three techs, skip the next four, then take one each from the other sectors.

This is the methodology that I used to create the Rip Van Robot portfolio, detailed in my Oct. 18 column “Rip van Robot: Investing’s Mr. October”). That portfolio was still tech-heavy, with 12 names from that suddenly benighted sector. But it could have been worse had I not reached farther into some of the screens to diversify with such names as Southwest Airlines (LUV, news, msgs), Coastal (CGP, news, msgs), Tenet Healthcare (THC, news, msgs), UnitedHealth Group (UNH, news, msgs) and Paychex (PAYX, news, msgs). The 20-stock portfolio is down 3.59% through Tuesday, which is about one percentage point worse than the Nasdaq Composite ($COMPX). Almost all the gainers have been non-techs.

No-brainer

The notion of overt sector diversification should come as a no-brainer to anyone schooled in Modern Portfolio Theory, the Nobel Prize-winning concept introduced by Harry Markowitz in a Journal of Finance article in 1952. The theory explains how risk-averse investors can build portfolios that optimize market risk vs. expected returns -- and it quantifies the advantages of diversification. To summarize glibly, MPT suggests that out of a universe of risky assets, one should make an effort to define a portfolio that rests on an “efficient frontier” that perfectly balances maximum expected gain at any given level of risk. MPT, in fact, is the concept that launched passive index funds, and it is the bedrock methodology behind the construction of most institutional portfolios.

In a minute I’m going to suggest a painless new way to achieve MPT nirvana with the help of a new Web site called FinPortfolio, but first let me spend two more minutes explaining why it’s necessary.

You may be wrong

Kenneth L. Fisher, a money manager, author and market historian in the San Francisco Bay area, believes MPT-oriented diversification makes sense because it forces you to acknowledge that your investment views may be mistaken. “Even when you’re right for some period of time, you don’t want to continue to bet too heavily because you know you will eventually be wrong,” he said. “In my business we always know we may be wrong, so every strategy is balanced with a counter-strategy.”

Fisher suggests that anyone interested in building a better portfolio in coming months should consider four things:

1. Pick a reasonable benchmark e.g., the S&P 500 ($INX) or set of benchmarks to manage against -- not your day-trading brother-in-law's bloated claims.
2. If you're using a set of benchmarks, assign each of the components an expected return over a specific time period (e.g., 12% over 12 months).
3. Construct portfolios that blend investments you believe have similar return expectations over those time periods, but are short-term negatively correlated (e.g., add drug and food stocks to your tech-stock holdings).
4. Always remember you may be wrong.

Fisher said that “what private investors get wrong the most is that they forget about picking a reasonable benchmark -- so they go off on a wild goose chase, seeking a level of return that makes no sense.”

Don't be a collector

Why does this happen? He says that private investors tend to function too much like collectors, gathering only things that they like and not the things they don’t like. “People tend to collect only value or growth stocks, big stocks or small stocks, tech stocks or drug stocks. And they don’t understand people who do the opposite -- much like the person who passionately collects antique cars doesn’t understand the guy who passionately collects stamps. But the way to become a star manager is to collect well between categories. You need to learn to become agnostic -- and never believe one category is permanently better than any other. It’s not true in history or in theory. And one reason it’s wrong in the equity market is that when underwriters see a huge demand for one type of stock, they go and create supply to match it -- an effort that ultimately brings prices into equilibrium. Whether it’s Internet stocks, optical-networking stocks or biotech stocks, supply is always going to get matched up against excess demand -- and prices will fall across the board.”

To blend MPT with the SuperModels as an experiment, I visited the new Web site FinPortfolio.com -- which was launched recently by a couple of former Goldman Sachs executives. I registered for free, clicked the "MyPortfolio" link on the left navigation menu and set up a mock fund with 10 SuperModel stocks from the Redwood and MVP Growth screens. I used the "Portfolio Optimization" link at the site to access software that analyzes funds against a set of MPT algorithms, then clicked through to a page that showed my SuperModel "fund" fell 12% short of the Efficient Frontier.

To fix the imbalance, I followed FinPortfolio's simple on-screen instructions to adjust my portfolio to reach a higher Sharpe ratio -- which is roughly return divided by risk. To achieve that “better” figure, its software dynamically showed me how to underweight -- or even throw out -- certain names on my list and overweight others. My original Sharpe ratio was a good 1.9, but the application showed that I could get an excellent 2.2 by tossing out four of my stocks and moving the money into the rest. The two lists are shown in the table below.

Fine Print

There are a slew of other cool tools at FinPortfolio.com besides the MPT Optimization. One of my favorites is the Asset Analysis page, which lets you see how the returns of up to 15 stocks correlate with each other. This does what my Buddy Stock spreadsheets attempted to do (“Choosing stocks by the buddy system”), but with a lot less effort. … Ken Fisher writes a terrific bimonthly column for Forbes; he called the tech-stock debacle correctly in late February in a column titled "1980 Revisited -- Tech Stocks are in a Late-Stage Bubble." It was no ordinary hatchet job on tech stocks, since he had been bullish on them for years. He’s written a couple of good books too, including "Super Stocks" and "100 Minds That Made the Market."… For more about Modern Portfolio Theory, check out the book "Equity Management: Quantitative Analysis for Stock Selection" by Kenneth Levy and Bruce Jacobs, with a foreword by Harry Markowitz (see links at left).