After two straight years of summer swoons in the stock market, “Sell in May and Go Away” has become one of the most popular sayings on Wall Street. And with markets down between 5% and 9% for the month so far, the Sell in May crowd is growing larger than ever.
But while the last month and the past two summers might be fresh in investors’ memories, that’s not nearly enough history to support a claim of market “seasonality.”
In addition, this year is different from the last two years in an important way: it’s a Presidential Election year. So in addition to the forces of seasonality, the market will also be influenced by the effects of the Presidential cycle.
In today’s Investment of the Week, I feature two Digest contributing analysts who have crunched the data for seasonality from a much longer time period, and taken into account the historical effect of Presidential Election years: InvesTech Research’s James Stack and Capital Growth Letter’s John Bollinger.
[In addition, I would be remiss not to mention The Stock Trader’s Almanac, the newsletter that did the initial calculations to support Sell in May—or, as they call it, the “Worst Six Months” strategy. You can read what they have to say about seasonality in election years in this blog post.]
Perspective #1: James Stack of InvesTech Research
InvesTech Research’s James Stack shows strong evidence that if you go away in May this Presidential Election year, you’re likely to miss a rising market. You should also note that the InvesTech Research portfolio is currently ranked by Hulbert Financial Digest as the seventh-best performer, adjusted for risk, over both the past five- and ten-year periods.
Here’s what Stack wrote about Seasonality in Presidential Election years under the headline “Sell PANIC in May, and Walk Away RUN!”
“After hitting a new bull market high on May 1, the Dow Jones Industrial Average immediately tumbled almost 450 points in six straight losing sessions. After two consecutive years of steep summer sell-offs, this ‘Sell in May and walk away’ maxim has become deeply ingrained in investors’ psyche—or perhaps we should say in their fears. … While it’s prudent that investors have become knowledgeable about seasonality, there is a good chance that they’ve carried the belief a bit too far—particularly in an Election year.
“While it is true that the overwhelming majority of stock market gains over the past 80+ years have come from the opposing November-April time period, it is fallacy to believe that every summer holds an ominous correction for the stock market.
“In fact, most investors following this theory would be shocked to learn that a sizable majority (63%) of May-October periods since 1928 have ended with positive gains—even excluding dividends. And of those positive gains, almost two-thirds of them have been greater than 5%.
“There’s still more insightful information to be gleaned from past Election years. Over three-quarters of Presidential Election years since 1928 have experienced positive gains in this seasonally unfavorable period. So before battening down the hatches based on a single Wall Street truism, let’s take a broader view of the historical probabilities—and update the latest evidence.
The Seasonal and Not-So-Seasonal Facts
“In years past, we have been one of the loudest voices about the six-month seasonality of the stock market. It is a truism based on fact, and one that shouldn’t be ignored—particularly in bear markets. Over the last 83 years, there have been 13 double-digit losses in the May-October period—however, all of them have been in bear markets. (Last year’s decline from May 1 to November 1—while an emotional roller coaster ride—experienced only a -8.1% loss.)
“So here are some of the most important facts: Historically, there is a strong seasonality in the stock market where the majority of stock market gains have been made in the November-April time period—and this effect has become more pronounced in the past 50 years.
“Shown below is a graphic ‘Tale of 2 Seasonal Investors’ in which two investors both start out in 1960 with $10,000, and buy a hypothetical S&P 500 Index fund at two opposing six-month periods of the year. …
“The difference in cumulative return over this 52-year period is surprising, if not downright shocking. Investor B does indeed double his money; however, Investor A multiplies his initial investment by over 40-fold. In total, Investor A captured over 96% of the available stock market profits!
“So why not sell in May and walk—or even run—away? The answer lies in the historical data and weight of the current evidence, which is not tilted into the bearish camp. Here are other relevant facts about this six-month seasonality that surrounds the ‘Sell in May’ axiom: Whether looking at the past 52 years (since 1960) or 83 years (since 1928), this maxim proves correct less than 37% of the time. Simply stated, the stock market has risen in the May-October period in over 63% of past years.
“Although all double-digit losses during this May-October period over the past 83 years have occurred since 1960, they have all accompanied bear markets. In other words, there were other strong negative influences and warning flags present. Of the May-October declines that did not occur during a bear market, the average loss was a relatively mild -3.8%.
“So while the ‘average’ returns after May 1 appear rather anemic, there is no reason to abandon a prudent strategy and become a seasonal trader. In addition, there’s another possibly more important ‘seasonal’ pattern at work right now—and that’s the fact that it’s a Presidential Election year.
“Let’s take a closer look at the month-by-month seasonal pattern that creates the ‘Sell in May’ phenomenon. Shown in the bar chart at left are the average monthly gains/losses in the S&P 500 Index since 1928.
“One can easily see that on average the favorable November-April returns are dominated by December, January and April gains, while the unfavorable May-October returns are weighed down heavily by September losses. Over 54% of Septembers since 1928 have seen losses in the S&P 500 Index. But let’s take a second look at these same average monthly gains/losses only for Presidential Election years since 1928. The picture is a very different one, with strong average monthly gains in June, July and August. All of this suggests that the ‘Sell in May and walk away’ truism isn’t quite so valid or true in a Presidential Election year.
“There are two other ways of looking at the Election cycle and Presidential Election years that we feel are important right now. If one looks at the quarter-by-quarter average gains or losses in the S&P 500 Index throughout the 4-year Election cycle, this is how it appears. This is now the 4th year (Election year) in the cycle, and we’re currently in the second quarter—which on average is a losing quarter. Note, however, that the third quarter of the Presidential Election year is historically one of the strongest in the 4-year cycle. The fourth quarter is also historically positive, on average.
“Perhaps the most straightforward means of looking at past Presidential Election years is to list every one of them (since 1928) along with their respective gain or loss from early May until Election Day. This is done in the table shown at left, and here are the most relevant observations: Over 80% of Presidential Election years have seen gains in the S&P 500 Index between May 1 and Election Day. Only one of the four losses has exceeded 10%, and that was during the Financial Crisis of 2008 (-27.4%).
“While a clear seasonal pattern exists between the opposing 6-month periods of May-October and November-April, the disparity in returns does not tell the whole story. The stock market still rises in the May-October time period almost two-thirds of the time. And the returns become more consistent and profitable in Presidential Election years.”—James Stack, InvesTech Research Market Analyst, May 11, 2012.
Perspective #2: John Bollinger of Capital Growth Letter
Our second perspective on the intersection of seasonality and the Presidential Election comes from the well-known John Bollinger, editor of Capital Growth Letter. Bollinger’s market timing system relies on a wide variety of ‘cycles,’ which can last from a few days to decades. So Seasonality and the Presidential Election cycle are right in his wheel house. Here’s what Bollinger wrote about the interplay between the two on May 11:
“When seasonals are working they are good guides, but when non-seasonal forces overwhelm them, doggedly sticking to the seasonal patterns can result in disaster. An important factor to consider is that some seasonals can influence other seasonal patterns. For example, the annual seasonal is strongly influenced by the four-year or presidential cycle, and that is the subject of this section.
“We present four charts, the first two depict the annual seasonal pattern with its familiar take away with a twist, buy in October and sell in June. The old wisdom used to be September and May, but as you can see the record suggests that the best time of the year to own stocks has shifted forward by a month over the years. Remarkably, the front- weighted version (the second, lower chart) highlights the importance of these inflection points while emphasizing the riskiness of being long during September and October. The data is the S&P 500 from 1968 through 2011, 44 market years and 11 four-year Presidential cycles. When employed, the weighting scheme is linear with each successive year’s weight increased by one. Charts [below], unweighted on the left, weighted on the right, are only part of the story.
“[Below] we present the four-year, election or Presidential cycle. There are a total of 11 cycles included with the election occurring in year one and the inauguration in the beginning of year two. Here there is more of a difference between the unweighted and the weighted, but let’s consider the unweighted first. The ‘sell in May’ idea is really only evident in the last two years of the cycle and most pronounced in the third year. The fall volatility that so scares investors is evident in every cycle. The strongest period by far is the October to June period of the third and fourth years; we are in an election year, year one, so that would be last year and the year before. The weighted cycle emphasizes the downside risk in the second half of an election year, but that is driven largely by the sharp decline in 2008, a rare event, like the crash of ’87, which ought not to significantly influence our thinking.
“So, the seasonal road map that I see calls for a small correction just past mid-year, a peak before the elections, and then a nice rally after the fall lows interrupted by a swoon a month or so after the inauguration. Once again, this is not carved in stone, just a sort of bias for prices.”—John Bollinger, Capital Growth Letter, May 11, 2012.
The bottom line: if you’re going to use a historical tendency to inform your investing, you must look back further than the last two years, and make sure you have the numbers right. Of course, even then, as fund managers say, “Past performance is not a guarantee of future returns.”
Wishing you success in your investing and beyond,
Editor of Investment of the Week