Blog - Investing Notes

February 27, 2007 - An Investor's Asset Class Momentum Trading Strategy

How well does momentum trading of asset classes work in the real world? In his January 2006 investing policy entitled "Class OutPerformance (COP) Strategy", reader and individual investor Mal Williams describes and justifies a dynamic investing strategy based on recent asset class outperformance. His historical justification is based on monthly return data (from Morningstar Principia) for 50 asset class proxies (exchange-traded funds and mutual funds) over a 17-year period. His implementation strategy selects the 10-15 asset class proxies with the highest momentum over the past 12 months, rebalanced monthly. He finds that:

• Investing in the top 40 of 50 classes in terms of momentum matches the return of Standard & Poor's Depositary Receipts (SPY) at much lower volatility.
• Investing in the top 15 of 50 classes in terms of momentum beats the buy-and-hold return of SPY at somewhat lower volatility.
• Investing in the top 8 of 50 classes in terms of momentum trounces the buy-and-hold return of SPY at about the same volatility.
• The cumulative return from investing in the top 15 asset classes is about 1200% over 15 years, compared to about 500% for SPY. The strategy produces only one down year over this period.
• During 2000-2002, the strategy dropped a little harder than SPY in the first few months, but quickly recovered to produce positive returns.
• When all asset classes weaken, a change in asset class leadership may be in progress. In such situations, sticking with the strategy is probably still the best bet.
• Recent actual results (drawing from the best of 80, rather than 50, asset class proxies) beat SPY by a cumulative 30% over four years.

The following chart, excerpted from the paper, compares the results of applying the strategy moderately (Top 15) and aggressively (Top 8) with those for buying and holding SPY over the period 1/1/91-1/1/06. Portfolio rebalancing occurs monthly immediately after recalculation of asset class momentums. Results suggest that the strategy favorably shifts the risk-reward ratio.

Our comments on the strategy paper and the active implementation are as follows:

• The recent research listed at Blog Synthesis: Momentum Investing/Trading indicates that a momentum investing approach is probably sound.
• Using 80 asset classes seems like overkill, with some pairs among the 80 likely having very high return correlations.
• Even though the strategy tracks and rebalances monthly, use of 12-month trailing returns to measure momentum and annual returns to measure results means that the number of independent observations from a 17-year sample is just 17, which is small for statistical inference.
• Returns for trend-following behavior may have weakened across the duration of the sample as more investors have studied and used it.
• As discussed extensively in David Aronson's Evidence-Based Technical Analysis: Applying the Scientific Method and Statistical Inference to Trading Signals (see our blog entry of 12/11/06), selecting the best from among many combinations of parameter values generally results in inferences that overstate future results.
• Since the strategy likely generates many short-term capital gains, a closer look at taxes and trading costs is in order. Obviously, trading in a tax-deferred account could have substantial benefits.

Mr. Williams responds that asset class momentum effects strengthened, rather than weakened, for 1999-2005.

In summary, an asset class momentum trading strategy may favorably tilt the risk-reward playing field for investors who systematically apply it.

For additional information regarding the strategy, including the appendices cited in the strategy paper, see Mr. Williams' full web site. Mr. Williams invites others to email him with questions, comments and suggestions.

For related research, see Blog Synthesis: Momentum Investing/Trading.