Yes, an actively managed portfolio of exchange-traded funds (ETFs) can generate risk-adjusted excess returns above those of its individual funds-a positive alpha. And, yes, it can be achieved with a wide variety of asset classes, including both large- and small-cap stocks, both growth and value stocks, both domestic and international stocks and both developed and emerging market stocks.

Strictly speaking, efficient markets theory contends that stock price changes follow a random pattern, i.e., subsequent stock price changes can be treated as identically distributed random variables. This theory asserts that everything knowable about a stock is imbedded in its price, and that price changes are due to new information randomly entering the market. Serious followers of business and market cycles, however, recognize that cycles persist for a period of time, that different economic segments benefit to different degrees over the course of a cycle and that segments benefiting the most become more highly valued. While the lengths and amplitudes of cycles cannot be exactly determined, their persistence influences stock prices considerably.

Devising strategies to profit from cyclical opportunities requires that two conditions be met. First, the means to easily implement the strategy must be available; next, it must be tested over a sufficient amount of time to determine its reliability. Fortunately, the first condition is easily satisfied by the use of exchange-traded funds. These are ideal because they are based on indexes representing broad classes of securities because they are broad in variety; and because they are highly liquid, which allows investors to take tactical positions with them.

The second back-testing condition is not as easily met because of the relatively short history of ETFs. If actual ETF performance data are unavailable, it is necessary to substitute the historical performance of the indexes upon which these ETFs are based. Fortunately, this is perfectly acceptable since ETFs track very close to their indexes, and conclusions derived from statistical testing are likely to be valid.

To devise potentially profitable active strategies, it is necessary to recognize when a cyclical change is taking place. This is accomplished by monitoring changes in business and market conditions, looking at changes in fundamental and technical trends, and finding out why one asset class is more likely to be valued over another.

Large- Vs. Small-Cap Stocks

Large-cap stocks are generally characterized as having significant revenues, earnings and dividends; broadly diversified product lines; multiple distribution channels; a large geographical base; substantial financial assets; and liquid reserves. They are usually considered sound investments by investors seeking dividend income, modest capital appreciation and "safe havens" in times of economic stress.

In contrast, small-cap stocks typically have small but significant revenue and earnings potential; meager or nonexistent dividend histories; narrow product lines; limited distribution channels; regional markets; and frequently chronic capital shortages. They are typically considered attractive investments to those who are seeking above-average capital appreciation either through internal growth or acquisitions, things that are most likely to occur during periods of economic expansion.

Given the characteristic differences between large- and small-cap stocks, there are times when large-cap stocks should be owned by investors usually favoring small-cap stocks and vice versa.

For the nearly 20-year period from February 1988 through October 2007, a portfolio that alternated between large- and small-cap stocks whenever one category had greater trailing-two-month returns than the other was more profitable than a simple buy-and-hold strategy in either asset class.             Trailing-two-month returns are used in this strategy to smooth short-term fluctuations and identify the returns' underlying trends.
This strategy produces a compound annual return of 12.5%, which is 0.6% above the Russell 1000 Index's 11.9% and which is 1.3% above the Russell 2000 Index's 11.2% (Figure 1). The strategy has a greater annualized standard deviation, at 15.0%, than the Russell 1000 stocks' 13.6%, but is less volatile than the Russell 2000's 17.5%. The strategy provides a comparable return per unit of risk as measured by its Sharpe ratio, 0.53, which is lower than the Russell 1000's 0.55, but is significantly above the Russell 2000's 0.38.

In one important aspect of risk, downside vulnerability, the strategy's maximum drawdown of 39.8% recovered in 24 months, compared with the Russell 1000's drawdown of 45.1%, which took 49 months (Figure 2). The Russell 2000 had two significant drawdowns over the period, one was 29.8% and recovered in only 16 months and another was 35.1% and recovered in 15 months. The strategy's greater returns account for the shorter recovery period.