Gallery:Step 1|Step 1: Active Investors

Sell in May? Forget about It!

Gallery:Step 1|Step 1: Active Investors

It never fails that at the beginning of May we receive a query about the “Sell in May and go away” market-timing strategy. We have always dismissed it out of hand for three primary reasons:

1)   In an efficient market, there is no reason for some months to consistently outperform other months. Even if outperformance of certain months is observed, then it is not worth pursuing if it lacks a sound explanation.

2)   It would be costly to implement, especially for taxable investors who would be subject to short-term gains.

3)   The documented past performance of market-timers1 is dismal.

This year, however, we decided to dig a little deeper and see what the data reveals. Specifically, we compiled monthly returns going back to January 1926 for the U.S. stock market and 1-month Treasury Bills, a proxy for cash.  We calculated the long-term returns for five different strategies: 1) a 100% stocks strategy; 2) a 100% cash strategy (which we don't recommend); 3) the sell-in-May strategy (sell stocks on May 1st and buy them back on November 1st); 4) the opposite strategy that we will call “buy-in-May"; and 5) a 65% stocks/35% cash portfolio that is rebalanced monthly (this allocation has the same standard deviation (risk) as sell-in-May). Here is what we found:

1/1/1926 to 12/31/2012 (87 years)

 

100% Stocks

100% Cash

Sell-in-May

Buy-in-May

65/35

Annualized Return

9.6%

3.5%

8.7%

4.4%

7.9%

Annualized Standard Deviation

18.7%

0.9%

12.1%

14.3%

12.1%

T-Statistic Relative to Sell-in-May

1.2

-4.3

N/A

-1.8

-0.8

 

Contrary to popular belief, Sell-in-May did not beat the market. The constant 100% equity investor would have made an average of a 0.9% higher return per year before costs, and the cost of implementing the Sell-in-May strategy would be on the order of 2% per year2 due to trading costs and taxes on short-term gains. While Sell-in-May had a higher return than Buy-in-May, as well as the 65/35 portfolios, the differences were not found to be statistically significant at a 95% confidence level (t-statistics were less than 2). This means that the observed level of difference is attributable to chance or randomness. In other words, a completely arbitrary sorting of returns such as stocks that begin with the letter “m” versus stocks that begin with the letter “s” could have produced similar results.

 So the next time somebody tells you to "Sell-in-May and go away," you can feel quite safe in ignoring that advice, as well as all other market-timing suggestions. Happy May Day!


1Graham, John R. and Harvey, Campbell R., Market Timing Ability and Volatility Implied in Investment Newsletters' Asset Allocation Recommendations (February 1995). Available at SSRN: http://ssrn.com/abstract=6006

2Sharpe, William F., Likely Gains from Market Timing. Financial Analysts Journal, March/April 1975, Vol. 31, No. 2: 60-69.