Two End-of-Month Seasonal Patterns to Remember

Heading into the final three days of July the S&P500 was up 8%. I initially ran some tests under the hypothesis that such unusual month-to-date strength would lead to further strength into month-end. The initial results were much more negative than I expected, and I moved on to other work, which was a mistake in retrospect. I was looking for research that would back up an existing opinion rather than letting the results speak for themselves. It wasn’t all for naught, as I revisited that research today and discovered that end-of-month negative tendency holds up in backtesting and actually represents a significant edge over random. I also uncovered a second, bullish seasonal pattern that’s one of the strongest I’ve come across (more on this in a moment).

It may be too late to act on this, but for reference the next time we’re in the midst of a big up month, the last few days of that month are more likely to see profit taking than further buying. Here’s a look at the last 30 occurrences in which the S&P was up over 3% in the three weeks prior to the last three days of the month…

SPX +3% Over Three Weeks Heading into Final Three Days of Month
07/27/10… S&P500 ??? three sessions later
11/24/09… S&P500 -0.9% three sessions later
09/25/09… S&P500 +1.2% three sessions later
07/28/09… S&P500 +0.8% three sessions later
03/26/09… S&P500 -4.2% three sessions later
12/26/08… S&P500 +3.5% three sessions later
04/25/07… S&P500 -0.9% three sessions later
11/25/05… S&P500 -1.5% three sessions later
11/24/04… S&P500 -0.7% three sessions later
08/26/03… S&P500 +1.1% three sessions later
03/26/03… S&P500 -2.5% three sessions later
10/28/02… S&P500 -0.5% three sessions later
08/27/02… S&P500 -2.0% three sessions later
11/27/01… S&P500 -0.9% three sessions later
10/26/01… S&P500 -4.1% three sessions later
04/25/01… S&P500 +1.7% three sessions later
03/28/00… S&P500 -0.6% three sessions later
11/24/99… S&P500 -2.0% three sessions later
08/26/99… S&P500 -3.1% three sessions later
04/27/99… S&P500 -2.0% three sessions later
12/28/98… S&P500 +0.3% three sessions later
11/24/98… S&P500 -1.6% three sessions later
10/27/98… S&P500 +3.1% three sessions later
09/25/98… S&P500 -2.7% three sessions later
06/25/98… S&P500 +0.4% three sessions later
03/26/98… S&P500 +0.1% three sessions later
06/25/97… S&P500 -0.4% three sessions later
02/25/97… S&P500 -2.6% three sessions later
11/25/96… S&P500 -0.0% three sessions later
09/25/96… S&P500 +0.2% three sessions later
05/28/96… S&P500 -0.5% three sessions later

Note that in 20 out of 30 cases, or 67% of the time, the S&P closed lower three days later (on the final day of the month), significantly greater than the 45% at-any-time odds for a lower S&P three days later in the same time frame. That’s a worthwhile bearish seasonal pattern to keep in mind the next time the market’s sporting a big MTD gain heading into month-end.

As I looked over the results of the study above, I couldn’t help but notice that my original hypothesis was usually proved correct, only not when I expected. Rather than the last few days of the strong month, it seemed investors waited until the beginning of the next month before resuming their buying. The tendency was so strong I ran a second test, utilizing the criteria outlined in the test above (SPX +3% over previous three weeks heading into last three days of month) but this time actually waiting until the end of the month. The results were astoundingly positive…

SPX +3% Over Three Weeks Heading into Final Three Days of Month,
Enter at Close on Last Day of Month
07/30/10… S&P500 ??? one week later
11/30/09… S&P500 +0.7% one week later
09/30/09… S&P500 +0.1% one week later
07/31/09… S&P500 +2.3% one week later
03/31/09… S&P500 +2.2% one week later
12/31/08… S&P500 +0.7% one week later
04/30/07… S&P500 +1.8% one week later
11/30/05… S&P500 +0.6% one week later
11/30/04… S&P500 +0.3% one week later
08/29/03… S&P500 +2.3% one week later
03/31/03… S&P500 +3.7% one week later
10/31/02… S&P500 +1.9% one week later
08/30/02… S&P500 -1.4% one week later
11/30/01… S&P500 +1.7% one week later
10/31/01… S&P500 +5.3% one week later
04/30/01… S&P500 +1.1% one week later
03/31/00… S&P500 +1.2% one week later
11/30/99… S&P500 +1.5% one week later
08/31/99… S&P500 +1.8% one week later
04/30/99… S&P500 +0.7% one week later
12/31/98… S&P500 +3.7% one week later
11/30/98… S&P500 +2.1% one week later
10/30/98… S&P500 +3.9% one week later
09/30/98… S&P500 -4.6% one week later
06/30/98… S&P500 +2.9% one week later
03/31/98… S&P500 +0.7% one week later
06/30/97… S&P500 +3.8% one week later
02/28/97… S&P500 +1.8% one week later
11/29/96… S&P500 -2.3% one week later
09/30/96… S&P500 +2.3% one week later
05/31/96… S&P500 +0.6% one week later

In 21 out of the last 22 occurrences since 1998, and in 27 out of the last 30 occurrences the S&P500 was higher one week later, averaging a gain of just under 2%. The random chance for a higher SPX one week later in the same time frame is only 54%, so this 90% win rate represents a huge edge over random.

Summing up these two solid end-of-month patterns… When the market is up big (3%+) in the three weeks leading up to the last three days of the month, expect some profit taking into month-end, then position yourself for a fresh influx of buyers during the first week of the following month. This bullish seasonal window goes into effect at Friday’s close and lasts until the following Friday’s close (August 6th).

Categories : Archived EOD Commentary, End-of-Day Commentary

 Print This Post