You Can Sell In May and Go Away

..but you might have to go very far.

Every May, the wires explode with articles about “Sell in May” regardless of what the stock market is doing. Last year we explored this aphorism to see what evidence there was for it historically and in particular 2017. Our conclusion was “You might get away with selling in May but don’t go away or you might miss the round trip of a generation.” So far the round trip has paused after an aggressive move higher. This May we have re-applied the same approach to reach what effectively is the same conclusion.

Mark Twain summed it when he said“October. This is one of the peculiarly dangerous months to speculate in stocks. The others are July, January, September, April, November, May, March, June, December, August and February.”Each and every month can and has been associated with rallies and declines. But they are prompted by a particular cause in a particular phase of the cycle. We often use averages that can produce conclusions such as ‘Sell in May’ but they gloss over some of nuances and in particular, other influences that may be driving the market more. In a sense if you use averages, you get average results. But you can take the average to the superior by factoring in the prevailing conditions.

At Matrix we believe strongly in evidencing all our analysis and trades. The weight of evidence suggests some countries, instruments or sectors can underperform in the May-October period compared to the rest of the year but not always. It is the particular position of the markets now within an historic Trumpflationary bubble that particularly skews this year’s Sell in May.

The phenomenon of falling stocks in the May-September period can be traced back to the UK as early as 1694 when money would flow out of equities into fixed-income loans to farmers to pay summer labor. When the harvest was gathered and sold, the money would flow back into equities. There is therefore an historic grain of truth to the Sell in May.

The earliest known direct reference to Sell in May aphorism is also British:

Sell in May
And go away
Come back on St Leger’s Day

The St. Leger Stakes, in September, is the last horse race in a society season that evolved in 17th century Britain and now starts with the 2000 Guineas in early May and comprises such events as the Chelsea Flower Show, the Derby, Henley Regatta and Cowes Week. These events were well attended by stockbrokers and investors who were almost exclusively drawn from the aristocracy. As they did not have access to live market information or dealing they would sell their stocks and go to cash for the summer.

The self-fulfilling nature of this ‘Sell in May’ social phenomenon could still conceptually apply as Northern Hemisphere vacation season does account for generally lower volumes, lower volatility and therefore technical pullbacks. But they would theoretically also limit the downside. In a financial world where the Southern Hemisphere is increasingly important, whose access to prices is instant, whose time horizons are much shorter and driven as much by non-seasonal events, the reasons for depressed stocks in the summer months have clearly declined.

The evidence is clear. US markets on average perform less well in the 5-6 months from May than they do the rest of the year. This underperformance is more noticeable outside of the US. But that doesn’t mean they go down or that the price movement in the northern hemisphere summer is either consistent or uniform. It varies considerably depending on where the market is within both the economic/political cycle and the overall trend.

May-October Relative US Under-Performance:
Over the last 20 years, the average move upwards in the SPX has been 8.7% in the winter half of the year and 2.8% in the summer half. The DJIA is even more pronounced. According to the Stock Trader’s Almanac, the Dow Jones Industrial Average has had an average gain of 7.5% during the November through April period and a gain of only 0.3% over the May through October period, going back to 1950.

Global Under-Performance: Bouman and Jacobsen[1] also found that the effect applied in just about every market, as shown in this chart by Jun Wei Yeo[2], even in Australia, where May to October is of course the winter period.

This under performance is stronger in the original source of the Sell in May phenomenon, the UK and highlights an important nuance. The sell off, if any, is most noticeable at the start of the period and may be due more to a shortlived self -fulfilling sentiment than any longer lasting fundamental cause.

On average there is a 3% decline in the first two months, which then fully recovers, and fades about 2% into mid-September. The UK Stock Almanac points out that May and September are statistically the worst months in the market.

The German DAX also followed this pattern of an early decline during the bull market of the 1980s but has since shifted to a stronger August sell off.

Seasonality within Seasonality:
A cursory glance at SPX performance between May and October 31st for the last 90 years and, separately, the more pronounced uptrends such as we have experienced since 2009 reveals a clear pattern. Lacklustre range trading from May to mid even late June (net -1% on average) is then followed by a rally in August-September that frequently exceeds the scale of the initial range (net +4% average).

But these are averages and, by definition, ignore cyclicality and the context of the overall trend. More specifically the chart below highlights the period May 1 to Sep 30 for the last ten years, mainly in uptrend. Only six showed negative returns for the first four weeks (May itself), but in eight of ten, a sharp pullback occurred after six weeks. Only in 2007 and 2014 did the market carry on rallying until late July. In the three years from 2010, the market literally sold off on May 1st.

This highlights the importance of specific conditions prevailing at the time and the underlying trend. The average May performance during the 1920s and 1980s bull market is not particularly impressive but nor is it particularly bad.

But it is also important where the market is within that trend. Clearly we think it is in the latter stages and, we also suspect, it is slightly in advance of the 1929 and 1987 seasonality.

The reasons for the Sell in May phenomenon are historical and now largely seem outdated. Although markets do underperform for the 6 months after May compared to the rest of the year, we suspect causality may be more self-fulfilling than anything else. The more people write articles like this the more likely Sell in May will feature in the market’s psyche and act as a deterrent to longs in, what we believe, could still produce a final flourish.

Here’s to making the very best.

Good Luck

Ed Matts
Matrix Trade

[1] The Halloween Indicator, Bouman and Jacobsen 1997

To receive our weekly articles by email in advance of publication, please sign up for our Sunday newsletter.