Investing may not have an obvious finale, but there's always a beginning. The question of how to begin doesn't receive the attention it deserves, but no one should confuse popularity with relevance in finance.

Portfolio design should start with an objective review of the risk and return possibilities, even though the crowd thinks otherwise. Just as a football coach or corporate manager must weigh the strengths and weakness of his team before making critical decisions, strategic portfolio analysis should commence with an impartial look at the implied payoffs for risk among the major asset classes.

Surveying the future with absolute objectivity and clarity isn't possible, of course, but we should still make an effort to assess the markets with as little bias as possible. To the degree we succeed, we have a useful frame of reference for adjusting the portfolio mix to satisfy each client's objectives, risk tolerance, etc. Developing a neutral outlook also provides some context for any short-term forecasts we care to make.

What's a neutral outlook? One answer comes from equilibrium-based estimates of risk premiums. These are assumptions about future prices in a world where supply matches demand. Supply and demand are constantly in flux, but the process of making long-term equilibrium forecasts is helpful for judging how our assumptions about risk translate into asset allocation decisions.

The rationale for making equilibrium projections comes from modern portfolio theory. Citing MPT may sound naive in the wake of 2008's financial cataclysm, but the recent attacks on finance theory are misguided. For example, some critics dismiss MPT by claiming that it demands a mindless acceptance of buy-and-hold asset allocation based on extrapolating historical returns into the future. MPT says nothing of the kind. In fact, the founding document in MPT-Harry Markowitz's 1952 "Portfolio Selection" paper-states clearly that some degree of looking ahead and estimating risk and return is vital.

The past isn't irrelevant, but neither should it be used as a shortcut for the hard work of designing and managing asset allocation through time. The solution is blending current and historical market information with expectations. That's no silver bullet, of course. MPT's output is directly related to the skills of the person wielding it-garbage in, garbage out, as they say. But at least we know where to begin.

MPT tells us that the market portfolio is optimal for the average investor for the infinite future. Over the long haul, an unmanaged, passive asset allocation of the major asset classes is likely to generate middling returns and risk relative to a wide spectrum of active portfolio strategies and individual asset classes. That leads us to the fundamental investment challenge: deciding how to customize Mr. Market's passive asset allocation.

Why customize? The short answer is that no one is average and everyone has a finite time horizon. But before we start second-guessing the market portfolio, we need something approaching an impartial yardstick for projecting the long-run outlook for major asset classes. In short, we must calculate equilibrium risk premiums.

With robust estimates in hand, we can begin analyzing the markets and customizing the asset allocation. To the extent that our intermediate outlook differs from predictions of long-term equilibrium risk premiums, we have a reference point for adjusting asset allocation. What if you don't have a strongly held view at the moment for equities? The fence-sitting implies holding the passive market-cap weights for asset classes until more convincing evidence sways your view.

Robert Litterman, chairman of the Quantitative Investment Strategies Group at Goldman Sachs Asset Management, summarizes the reasoning in Modern Investment Management: An Equilibrium Approach (Wiley, 2003). "We need not assume that markets are always in equilibrium to find an equilibrium approach useful," Litterman writes. "Rather, we view the world as a complex, highly random system in which there is a constant barrage of new data and shocks to existing valuations that as often as not knock the system away from equilibrium."

These departures from equilibrium provide strategic-minded investors with guideposts for exploiting the apparent opportunities, he continues. Financial markets have a "center of gravity" that's defined by the equilibrium between supply and demand, Litterman advises. As a result, investors should take advantage of the deviations from equilibrium. "Understanding the nature of that equilibrium helps us to understand financial markets as they constantly are shocked around and then pushed back toward that equilibrium."

The first rule in forecasting equilibrium performance for the various asset classes: Avoid the trap of trying to predict returns directly. A superior approach is inferring risk premiums from the interaction of investor risk aversion, market volatility and the correlations of returns. Developing robust estimates is still prone to error, of course, but the odds for success are better, if only slightly, by analyzing risk compared with trying to forecast returns outright. For example, financial research reveals that the ratio between any two risk premiums generally matches the ratio of correlation between those premiums. In turn, that tells us something about how to think about expected return. Simply put, reverse engineering risk premiums by measuring and analyzing risk is a more reliable way to anticipate return.

Using the formulation outlined by money manager and asset allocation expert Gary Brinson in The Portable MBA In Investment (Wiley, 1995), the equilibrium risk premium estimate for a given asset class is calculated as:

Sharpe ratio  x  ?i  x  ?im

with the variables defined as

Sharpe ratio = the price of risk, or the asset's return over the risk-free rate divided by the asset's annualized volatility (standard deviation of return). This is a proxy for the general level of risk aversion, which is derived from investors' utility functions. Higher levels of risk aversion equate with a greater demand for return per unit of risk, and vice versa. In equilibrium, this relationship is proportional across assets and applies to the market portfolio overall.

?i = standard deviation (volatility)
of asset i

?im = correlation of asset i with portfolio m

As a basic example of estimating risk premiums, imagine that the market portfolio is comprised of three assets: domestic stocks, foreign stocks in developed markets and intermediate-term U.S. Treasurys. (In practice, strategists should define the market portfolio as broadly as possible.) As proxies, we'll use the S&P 500 (U.S. equities), MSCI EAFE (foreign equities) and a five-year Treasury bond.

To keep the following example manageable, we'll simplify the analysis. Let's begin with the recent history of U.S. stock market volatility. Historical data offers a practical starting point, but the past should only be a guide. Consider that standard deviations rose sharply in 2008 amid the dramatic sell-off. That compares with relatively low volatility for equity returns during 2004-2007, a period of generally rising asset prices. During those four years, annualized monthly standard deviations of large-cap U.S. stocks in each calendar year ranged from roughly 6.5 to a bit over 10 before jumping to more than 14 for 2008, according to Ibbotson SBBI 2009 Classic Yearbook (the source for all subsequent data cited unless otherwise noted).

We know too that volatility runs in cycles. That suggests that volatility will fall from 2008's lofty level in the years ahead. Let's project that domestic equity volatility will drop moderately to a standard deviation of 12 on a calendar year basis for the foreseeable future.

Foreign equities tend to exhibit even higher volatility from a U.S. dollar-investor perspective, and so we assume a standard deviation of 18, or slightly above the reported figure of roughly 17 for 1998-2008.

As for bonds, volatility for intermediate government securities has fallen in recent years, in part because interest rates dropped to record lows. But interest rates can't fall much further, if at all, and so investors should be suspicious of the recent lull in fixed-income volatility.

For 1999-2008, intermediate Treasurys posted an annualized standard deviation of 5. Yet the 8.8 volatility of the 1980s looks more reasonable going forward. Why? Interest rates were rising in the 1980s, in part because the Federal Reserve tightened monetary policy in the early part of that decade to fight inflation. Although the inflation outlook at the end of 2008 was quite low, bordering on deflation, one could reason that higher rates were coming as the economy rebounds. If so, the trend would likely elevate inflation to something approaching historically average levels. In turn, let's estimate a volatility of 7 for intermediate government bonds, or modestly higher from the reported 5 for the asset class during 1999-2008.

For correlation, consider the ten-year record through 2008 for each of the three asset classes relative to a market-cap-weighted portfolio of the trio, as determined by market values at 1998's close. The resulting correlations for 1999-2008 are 0.97 for U.S. stocks, 0.96 for foreign developed stocks and -0.29 for intermediate government bonds. Correlations tend to rise during severe periods of selling and so the reported numbers are somewhat skewed to the high side for equities in the wake of 2008's dramatic losses.

Meanwhile, due to the rush into the safe harbor of Treasurys last year, the normally low correlation for bonds relative to stocks has fallen further to a modestly negative reading. Anticipating a more conventional relationship for stocks and bonds in the years ahead, let's forecast slightly lower correlations for equities and slightly higher correlations for government bonds. As such, we estimate correlations with the market portfolio of 0.9 for both U.S. stocks and foreign stocks and 0.2 for Treasurys.

Finally, we need to make an assumption about the general level of investor risk aversion, as represented by the Sharpe ratio. This can be thought of as the market's demand for return per unit of risk (volatility). After studying history and reading the literature on risk aversion, we might anticipate a Sharpe ratio of 0.25 for the portfolio.

With the necessary estimates in hand, we can use the equation above to calculate an implied equilibrium risk premium for each asset class:

U.S. Stocks: 2.7%
Foreign Stocks: 4.1%
Intermediate Treasurys: 0.4%

Remember, these are risk premiums and so we need to add an expected risk-free rate to calculate total return estimates. For instance, if we assume that three-month Treasury bills (a common proxy for the risk-free rate) will match the long-term historical inflation rate of 3% for the foreseeable future, our equilibrium total return estimate for U.S. stocks becomes 5.7% (a 2.7% risk premium plus a 3.0% risk-free rate).

To calculate the equilibrium risk premium for our simple three-asset class portfolio, we add up the weight-adjusted return estimates. Using relative market cap weights as of September 2009, we project an expected risk premium of roughly 3.1% for this portfolio. If we add a 3% risk-free rate, the portfolio's projected total return becomes 6.1%.

How can we use these equilibrium estimates of risk premiums to build a portfolio? One option is running the forecasts for each asset class through portfolio optimization software to determine the optimal asset mix. But this approach has pitfalls and shouldn't be used in isolation. In particular, small changes in assumptions can lead to extreme results in the standard mean-variance model. One solution is using what's known as the Black-Litterman model for taming extreme outputs by integrating an investor's views with equilibrium estimates.

We can use our forward-looking equilibrium forecasts as neutral views of the future. A long-term investor with no particular outlook, or one who deems himself average, should consider the equilibrium projections as the basis for asset allocation.

Alternatively, if you have definite expectations about one or more of the asset classes, those forecasts are the inspiration for altering the neutral weights.

For instance, let's say you're quite a bit more bullish on U.S. stocks for the next three to five years compared to the equilibrium risk premium projection. Adapting that view into the asset allocation in the example above translates into raising the U.S. equity allocation over the passive market cap weight of roughly 41% (as of September 2009) of the total portfolio. A more bearish outlook suggests a below-41% asset allocation for U.S. stocks.

Inevitably, different investors will come to different conclusions about the future. Even your own view on what's coming will change over time as new information arrives. In short, no one forecast should be considered definitive. What seems reasonable today may look unrealistic tomorrow. We should also assume that our forecasts, no matter how thoroughly researched, will be less than perfect. That suggests holding a multi-asset-class portfolio to hedge this risk. As our confidence in our projections increases, so too might our willingness to deviate from the passive asset allocation implied by market capitalization.

We also need to think about the real-world implications if we're expecting, say, a 3% risk premium for the portfolio and end up with less or even a small loss over some period of time. If that possibility is unacceptable, perhaps it's time to go back to the drawing board and reassess expectations, risk tolerance and the asset allocation choices. Indeed, the example above is only a simple illustration of projecting risk premiums, and a deeper analysis may yield different results.

Nonetheless, the future is still uncertain and projecting equilibrium risk premiums won't change that fundamental obstacle. But routinely performing the analysis for the major asset classes and the portfolio overall promotes a deeper level of insight when it comes to changing Mr. Market's asset allocation to match your client's needs. That's hardly the last word on anticipating the future, but it's a reasonable place to start.