Blogs & Comment

Sell in May and go away?

A book I read recently, Buy Dont Holdby Leslie Masonson, presents (among other things) an argument for selling stocks in May and buying them back in November. Shortly after reading Masonson, I came across some posts by passive-investing guru Larry Swedroe. They argue that the Sell-in May strategy is bogus. Who is right?
Lets review Masonson first. He cites findings from several studies:
in The Stock Traders Almanac 2010, Yale Hirsch writes that since 1950, holding the Dow Jones Industrial Average during the 6 months from Nov. 1 generated 7.3% per year compared to a loss of 0.1% per year during the other 6 months
Sam Stovall, chief investment strategist for S&P Equity Research claims the S&P 500 Index has advanced 6.6% annually between November and April since 1945 while the other six months averaged only 1.4% annually (less than holding cash, i.e. treasury bills)
Sy Harding, publisher of the Street Smart Report, tweaks Hirschs data with moving-average signals to allow flexibility in entering and exiting stocks around the endpoints of the six-month periods — and finds the approximate period from November to April yielded 9.2% annually versus a loss of 1.8% annually for the other approximate six months.
Seems rather impressive, especially after the May tumble we had in stocks this year. But then Swedroe crunches the data over a different time period, from January, 1926 to April, 2010. His Sell-in-May portfolio swapped between the S&P 500 and 30-day U.S. Treasury bills and recorded average annual returns of 8.18% compared to 9.86% from simply buying and holding the S&P 500.
What could account for the difference between his simulation and those of the Sell-in-May crowd? The endpoints of the time periods. Swedroe reran his calculations over different time periods to show that Sell-in-May could outperform at times, depending on the endpoints.
Sell-in-May, therefore, is a random phenomenon that can be made to look systematic by cherry-picking endpoints, says Swedroe. An investment strategy is of not much use if you cant tell whether or not it will reliably finish ahead of buy and hold. In addition, when transaction costs are included, there will be fewer periods when Sell-in-May outperforms — and still fewer if the strategy is pursued in a taxable account.