For the past week, pundits have confidently mouthed one of the oldest market bromides around: Sell in May and go away. If the seven-week market rally is a bear-market rally, which it probably is, selling now and going away probably won’t turn out to be a terrible idea. But just don’t do it because it’s May.
Remember the “January effect”? At the end of every year, commentators and analysts start slobbering about this marvellous money-making strategy. In years when stocks actually do go higher in January, it’s a great trade. In years when they don’t? Well then it’s not such a good trade. Such as this year, when stocks dropped about 10% for the month.
Like other market bromides, in other words, the “January effect” and “Sell in May and go away” work in some years and don’t work in other years. Which is to say they work about as well as most other market-timing strategies. In the meantime, they rack up brokerage fees and taxes, and they trick people into thinking they know something that other people don’t. Such is the joy of data mining.
(What is “data mining”? Seeing patterns in data that are merely normal variation–like concluding that one person is a “better” coin-flip predictor than another because he or she has a better average over 100 flips. Go ahead and try that one with a friend. One of you will almost certainly have a better record–and you’ll both begin to conclude that he or she is “better”.)
In any event, John Mauldin and Prieur du Plessis took a close look at “Sell in May and go away” this week. For the past 60 years, the pattern is certainly visible, especially in bear-market years. If these charts make you comfortable enough to bet your money on the sell-in-May trade, be our guest. But if it doesn’t work, don’t come crying to us.
(And, sorry, but if the market crashes this summer, we’re not going to chalk it up to “sell in May and go away.” We’re going to conclude that the March-May boom was yet another bear-market rally, which we still think it probably is.)
My friend and South African business partner Prieur du Plessis recently updated a chart on monthly stock market returns since 1950. It clearly shows that the November through April periods have on average been superior to the May through October half of the year. (To read his very interesting blog you can go to http://www.investmentpostcards.com/)
And the difference is quite significant. As Prieur notes, the “good” six-month period shows an average return of 7.9%, while the “bad” six-month period only shows a return of 2.5%. Of course, selling creates taxable events, which can hurt your returns.
Plus, you never know when the markets are going to go down and when they will be up. There can be a lot of variance from year to year. For instance, in 2007 the markets were up during the summer by 4.52% and down during the “good” period by -9.62%, which is opposite the average pattern. Of course, the markets did go down by 30% after May 1 last year and down another 5% since then. That is what bears markets can do.
Which caused me to wonder. The last 59 years have seen two significant secular bull markets (roughly 1950-1966 and 1982-1999) and two secular bear markets (1966-1982 and 2000-??? — the one we are in now). I wondered if the pattern changed during the bear cycles, so I shot a late-night note off to Prieur and came in the next morning and had my answer.
It made a significant difference. May through October in secular bear cycles has been ugly. Look at this graph:
And just for fun, let’s look at the monthly numbers since the present secular bear market began in 2000. So far, this has been a lot worse than the 1966-82 cycle, although we have not yet had the recovery phase from the current doldrums, which will likely make the overall numbers look better in 4-5 years.
Sign up for John Mauldin’s weekly e-letter here >