How Moving Average Strategies Can Really Work
October 28, 2014
by Jerry A. Miccolis, CFA®, CFP®, FCAS, CERA, Marina Goodman, CFA®, CFP® and Rohith Eggidi
Has including ESG become a necessity for investors?
Advisor Perspectives welcomes guest contributions. The views presented here do not necessarily represent those of Advisor Perspectives.
In the August 19, 2014 issue of Advisor Perspectives, Paul Allen explored the universe of moving average crossover (MAC) strategies in his article “Do Moving Average Strategies Really Work?” In his thorough and even-handed analysis, Allen concluded that MAC strategies can effectively decrease periodic drawdowns in portfolios but can materially underperform during bull markets. In this article, we propose how to improve MAC strategies so that they may perform better during bull markets and still provide protection during bear markets.
All or nothing
Allen’s research assumed a single investment in the S&P 500 index. A buy signal meant 100% investment in the index, and a sell signal meant 100% in cash. His results demonstrated the well known “whipsaw effect,” to which virtually all MAC strategies are vulnerable. Specifically, if a modest downturn in the index is followed by a quick recovery, the strategy may suffer most of the initial loss but then exit the market without benefiting from the early days of the subsequent gain. Since this is a common feature of market behavior during bull markets, simple MAC strategies tend to underperform in such markets.
One way to improve the MAC strategy, in our view, is to recognize that the all-or-nothing approach represents a false choice. It is not necessary to confine your exposure in the equity markets to a single index. Once you embrace more than one exposure, you have more choices when you get a sell signal. That is, getting a sell signal does not necessarily mean being completely out of the market. For example, you could split the S&P 500 index into its underlying industry sectors and have independent MAC signals for each. You might also choose to broaden the investment universe to global equities and have signals for its different components, such as U.S. equities, international developed markets, and emerging markets. Ideally, the components should not have very high correlations with one another so that their respective signals are not identical.
A less extreme example
In this article, we focus on using the ten Global Industry Classification Standard (GICS®) sectors of the S&P 500 index. Strictly for illustration, we first explore a fairly rudimentary strategy. A buy signal is generated for each GICS sector when the moving average for its total return index over the last 50 days is greater than or equal to the moving average over the last 200 days. This is also known as the “golden cross” strategy. Sectors are equally weighted, with a maximum 25% allocation. If fewer than four sectors have a buy signal, then the remaining allocation goes to cash. The results are shown in Graph 1, and key returns are shown in Table 1. Note that the start date of 10/8/1990 is based on the first possible date that signals can be generated, given the availability of sector data.
As seen in the graph and the table, the sector-rotation MAC strategy did much better during bull markets than the MAC strategy that invested only in the S&P 500 and cash. It did not do as well as a buy-and-hold strategy during the bull markets of 10/1990 – 6/2000 and 3/2009 – 8/2014, but it outperformed during 12/2002 – 12/2007. One could more nearly align the sector rotation MAC performance with a buy-and-hold strategy during bull markets by making the sector weights closer to their market-cap weights instead of equally weighting them.
The sector rotation MAC strategy provides meaningful protection during down markets, but not as much as an S&P 500 MAC strategy. Nevertheless, the sector rotation MAC strategy outperformed both the S&P 500 MAC strategy and the buy-and-hold strategy over the full experience period shown.
Remember, if you have a question or comment, send it to firstname.lastname@example.org.