Active-Trader - Evidence of Autumn

Welcome back, Active Traders and Wealth Builders.

"History recalls how great the fall can be, while everybody's sleeping, the boats put out to sea. Borne on the wings of time, It seemed the answers were so easy to find. Too late, the prophets cry, The island's sinking, let's take to the sky." - Supertramp - Fool's Overture

Of course I am waxing poetic here, but I can't help myself. There is no escape from the cycles of the seasons, and I have taken enough spins around the sun to see this one coming. What the heck am I talking about?

I'm talking about the sell-off in the stock markets that almost inevitable occurs in the late August to October time frame. Plus after the run-up we have had in the averages this year - nearly 20% in the SP-500 - we are more than due for some selling. And we started to see some signs this past week with the Dow Jones Industrial Average posting its first negative weekly return in almost 2 months.

There's plenty of anecdotal evidence of sell-offs in the stock market that come in the autumn months. Notable examples include 1929. 1937. 1966, 1974 and 1987. But every year is different and where we are talking about money nobody, wants to hear anecdotes. You want facts, hard evidence - you want rigor, right?

Well fear not blog readers since I went on a search for some answers and using the awesome data analysis tools at my disposal here's what I came up with. The goal was to determine - is there clear numerical evidence that the stock market sells off in the autumn months?

I started with TradeStation and pulled up a weekly chart of the Dow Jones Industrial Average. I chose the Dow because it has more history than any other index and has weekly data all the way back to 1920. Next, I created an indicator in EasyLanguage which plots the year-to-date returns of the average as an indicator on the chart.  This was basically a two-liner in TradeStation as follows:

{ Grab last year's close } 
dLastYearsClose = CloseY(1); 

{ Percentage Change }
dPctChange = ((Close - dLastYearsClose)//dLastYearsClose*100.0);

Next step was to bring up the data window and export the data to a text file where we had the date, the weekly close and the Year-To-Date percentage change for the date.

Next step was to bring the data up in Microsoft Excel and save it as a spreadsheet which makes it slightly easier to import in MS-Access where the real number crunching gets done.

Next steps is to load up the data in Excel and add a column which returns a week number of the data for that week - first week of January is 1, 2nd week is 2, etc.  This allows seasonality to be extracted on a broad basis without getting involved in what day of the week the month starts on etc.  I used this method with some success in my prior post Fun with Forex Historical.

Next and almost final step was to create a query in MS-Access which groups the data by week number of the year and returns the average historical year-to-date change for that week number of the year.  I took that data and put in back into Excel to create this chart.

Let's stop for a second to understand what this chart means.  This chart represents the typical stock market year by week of the year using 92 years worth of data which is pretty powerful and useful information.

 So there's the information, what are the key takeaways?

- The stock market goes up over time.  In other words, there was not a single week of the year with a negative historical return

- There was a noticeable peek at week 35 followed by a trough at week 42, then it was straight up from there through the end of the year.  Week 35 corresponds roughly to the first week of September.  Week 42 corresponds roughly to the last week of October, or first week of November.

- The sell-off which occurs between these dates - on average - just gives back some of the returns for the year but does not put the overall proposition into the negative.  To illustrate this point, take a look at the screen shot of the data window from Tradestation from 1987.  The market was up over 40% for the year at the peak on 8/21/1987.  And after the historical sell-off ended the year still up +3.59 for the year!

So what does all this mean for my trading and investing?  It provides firm historical data which tells me to hold off on making any large long-side commitments between now and the end of August.  Wait for the sell-off and ugliness which occurs into the fall.  Then pile back into the market and the end of October for the year-end rally.

Like I said above, every year is different and we can't predict the future.  But we have some pretty clear evidence of seasonality in the stock market above.

Enjoy your week ahead and keep your power dry - until the end of October.