Don't Sell in May

By in
Don't Sell in May

By Jessica Rabe

At DataTrek, we’re not afraid to keep things simple when the data supports a clear message. Since February of this year we’ve been highlighting that January 2019 was a huge month for the S&P 500, and in that outsized return sits an important signal about market direction for the future. Consider the following points:

  • The S&P 500 gained 7.9% in January, one standard deviation above this month’s average return of 1.2% since 1958 (first full year of data). 

    There have only been eight other Januaries that have also returned +1 standard deviation above the average, or just 15% of the time. These years are: 1961 (+6.3%), 1967 (+7.8%), 1975 (+12.3%), 1976 (+11.8%), 1985 (+7.4%), 1987 (+13.2%), 1989 (+7.1%), and 1997 (6.1%).
  • During February and March of those 8 years after an abnormally big January, the S&P finished higher 75% of the time in both months. The average return in February was +1.4% for those 8 years versus +3.0% this year. The average return in March for said years was +1.5% compared to +1.8% during that month this year.
  • Overall, the S&P 500 posted a positive return in Q1 during all eight years when January returned over one standard deviation above the average. The average Q1 return for those years was 12.1% versus 13.1% in Q1 2019, slightly better than the average but almost on the money.

So what does our “Strong January” indicator say about Q2? Here’s the data:

  • During the eight years prior to 2019 when the S&P returned over one standard deviation above the average in January, the S&P gained an average of 2.2% in April and was positive in +60% of those years. With today marking the final trading session of April, the S&P locked in a +3.9% return.
  • As for the rest of this quarter: the S&P increased an average of 1.9% and was positive 75% of the time in May of said 8 years. The S&P was also up an average of 2.1% and positive 75% of the time in June.
  • In terms of Q2 as a whole, the S&P advanced 6.3% on average. It was only lower once out of those eight years in Q2 1961, down 65 basis points. The best performance was +16.9% in 1997. With a gain of 3.9% in April, there’s plenty of room to run these next two months to get closer to the quarterly average.

As we look towards the back half of the year, however, the data gets a little choppier. Here are the average returns for July through December during the eight years when the S&P returned +1 standard deviation above the average (save 2019) in January:

  • July: The average return was +2.7%. The lowest was -6.8% in 1975 and the highest was +8.8% in 1989.
  • August: The average return was minus 0.5%. The worst was -5.7% in 1997 and the best was +3.5% in 1987.
  • September: The average return was minus 0.1%. The weakest was -3.5% in 1985 and the strongest was +5.3% in 1997.
  • October: The average return was minus 2.5%. The lowest was -21.8% in 1987 and the highest was +6.2% in 1975.
  • November: The average return was +1.3%. The worst was -8.5% in 1987 and the best was +6.5% in 1985.
  • December: The average return was +2.8%. The weakest was -1.2% in 1975 and the strongest was +7.3% in 1987.

Whereas the S&P posted higher returns on average every month of the first half of years that started off with abnormally large January returns, just half did on average in the second half. If you’re partial to the “Sell in May and go away” mantra, this data points to selling at the end of July and buying again towards the end of October before enjoying a rip higher into the New Year.

That said, and without getting into the weeds on timing given that history doesn’t always repeat itself, here’s the outlook for the year based on this analysis:

  • Every year except 1987 generated double digit total returns for the S&P in the eight years when the index returned +1 standard deviation above the average (save 2019) during the first month of the year.
  • The total returns for each year were as follows: 1961 (+26.6%), 1967 (+23.8%), 1975 (+37.0%), 1976 (+23.8%), 1985 (+31.2%), 1987 (+5.8%), 1989 (+31.5%), 1997 (+33.1%).
  • The average total return for these years was 26.6%. This year, the S&P is up 17.5% YTD, so again plenty of potential gains still left ahead.
  • To further illustrate that last point, the earliest month when the S&P reached its peak for the year was in mid-July 1975. The next earliest was in late August 1987.

    The balance of years peaked in the fourth quarter. The S&P reached its highest level in September during 1967 and 1976. The S&P also topped out in early October during 1989, and peaked in December in 1961, 1985, and 1997.

Bottom line: we continue to hammer home the point that the 8 years with abnormally strong January returns mostly occurred in mid-to-late cycle, not unlike where we are now. Our playbook has worked through April, and history says there should be further upside during the balance of Q2 and the year. We’ll keep monitoring the data, but even with recent highs for the S&P this analysis shows the peak for the year doesn’t typically happen until the back half and mostly in Q4.