The quest for Camelot pales in comparison to the insatiable desire to produce alpha. Alpha has various technical definitions, but most simply defined, it is considered excess return above the return of the most relevant benchmark.

Ironically, if a fund beats its benchmark, the resulting excess performance can be referred to either as alpha-or as tracking error. Something bad (tracking error) from one perspective is good (alpha) from another. Pragmatically speaking, it is a tracking error if a fund underperforms its benchmark and alpha when it outperforms.

The never-ending search for alpha has led advisors and investors into some unlikely places, a few of them dark and scary indeed. But instead of searching for alpha in individual funds, it might be better to look for it at the portfolio level instead. Research shows that a diversified portfolio of beta ingredients-ETFs that mimic indexes-will produce a surprising amount of alpha.

STAYING RELEVANT

Correctly calculating alpha is a tricky job, mostly because it requires a relevant benchmark. While that may sound like a straightforward requirement, it's really not. For example, many actively managed funds contain a mixture of assets in a wide variety of weightings and combinations.

Take a growth and income mutual fund that contains stocks across various market capitalizations, an assortment of bonds and some cash for liquidity. This type of fund represents the beginning of a diversified investment portfolio because it contains several different asset classes, not just one asset class such as all stocks or all bonds.

But here's the rub: Virtually all benchmark indexes currently available are single asset-class indexes, meaning they are entirely a stock index or a bond index. Precious few indexes today incorporate multiple asset classes.

Consider the dominant indexes: the S&P 500 includes only large-cap U.S. stocks; the Russell 2000 includes only small-cap U.S. stocks; the MSCI EAFE includes only stocks from outside the United States; the Barclay's Capital Aggregate Bond Index includes only U.S. bonds. The list goes on.

Only in recent years have we begun to see the development of multi-asset class indexes (those that include various types of equities as well as fixed-income components). Most of these newer generation multi-asset indexes have been developed to benchmark the performance of target-date funds, which in nearly all cases are multi-asset portfolios.

Nearly all equity or fixed-income indexes are useful for benchmarking individual single-asset mutual funds (or ETFs), if and only if the fund and index have similar portfolio characteristics. The typical index is inappropriate as the benchmark for a multi-asset mutual fund. Moreover, nearly all indexes are inappropriate benchmarks for even the most basic investment portfolio, which should include at least several different asset classes (meaning several different types of mutual funds).

Let's cut to the chase. Shouldn't the whole investment portfolio be the real measure of performance, rather than micro-analyzing all the individual components within the portfolio? After all, portfolios are a bit like salsa. Several common ingredients in salsa are rarely eaten alone, but when combined skillfully, they enhance one another. The whole is tastier than the parts. Portfolios are similar to salsa in this regard. The whole (portfolio) is greater than the sum of the parts (individual funds in the portfolio).

The bottom line is that alpha should be a portfolio measure, rather than an individual fund measure. That is, does the portfolio as a whole generate a return greater than the sum of the individual ingredients?

PORTFOLIO ALPHA

An exchange-traded fund (ETF) typically provides a pure index exposure. (Active and leveraged ETFs are a different story.) By themselves, these ETFs are not alpha producers because they mimic indexes. Indexes don't produce alpha; they are a pure beta exposure. So technically most ETFs should have an alpha of zero and a beta of one. (In fact, the alpha of an ETF should be slightly negative to account for the expense ratio.)

Let's take 12 ETFs and combine them into a portfolio that covers all major asset classes. So as not to play favorites, all 12 ETFs are equally weighted in the portfolio and rebalanced to equal weighting at the end of each year. This is a well-diversified portfolio. But will it produce alpha? Said differently, can a portfolio of pure beta ingredients produce alpha at the portfolio level?

The following analysis shows that when combined appropriately, non-alpha-generating ingredients can produce alpha at the portfolio level. The challenge is calculating a portfolio-level alpha. Doing so requires a comparable, relevant benchmark that has similar portfolio asset allocation characteristics. As already noted, multi-asset indexes are few in number, and the ones that do exist have limited performance histories, making comparisons hard.