Stats Show Investors Should Reconsider "Sell In May" Rule

With the month of May now in the history books with a minus sign, it seems those who ridiculed the old Wall Street adage of "Sell in May and Go Away" may wish to re-think their stance. On that note, I would like to share some further research I have done regarding this bit of Wall Street wisdom.

First some background information. The data I have analyzed covers the period from 1950 through 2010, inclusive. That is a total of 61 years of market activity, the results of which are based upon the S&P500 Index.

The total 61 year time-frame is broken down into 4 distinct periods: 2 "Secular Bull Markets" (1950 thru 1965 and 1982 thru 1999) and 2 "Secular Bear Markets" (1966 thru 1982 and 2000 thru 2010). Note that the bull market from March 2009 is considered to be part of an ongoing Secular Bear. With the benefit of hindsight a few years hence we may conclude that the Secular Bear Market ended back in March 2009, but for now I am working with the assumption that it is still in force.

As for the return calculations I will be referencing, I am assuming returns while invested equal that of the S&P500 Index. When not invested I assume no return was made, thus ignoring any interest that may be earned for a "real" investor. Also ignored is the earning of any dividends while funds are invested. I will also be ignoring all transaction cost, fees, etc. Our purpose here is to have a generalized discussion of concepts.

Second, let's look at the "Sell in May and Go Away" rules. The basic concept is that one should sell out of the market as April ends and May begins and that one should remain in a cash position until the Fall; re-entering the market at some point during September or October. There are numerous versions of the adage utilizing various specific start and end dates. For our purposes I am using one which has an investor sell out of the market at the close on the 3rd trading day in May and buy back into the market at the close on the last trading day of September.

Now let's look at general overall results for a theoretical account which begins with $10,000. If we were to follow the adage's advice, our "account" would only be invested from October's first trading day through the 3rd trading day of May. The results are rather impressive in that our $10,000 account would grow to roughly $817,628, for a gain of 8,076.28%.

Compare this result to that of a 100% buy and hold account which remains fully invested throughout the 61 year period. For such an account, the result based on the S&P500 Index is that the same $10,000 account would grow to $798,980, a gain of 7,889.80%.

While not hugely different, the significance of the above figures lies in what they imply about the May-Sept period. You see, if one invested in a manner exactly opposite that suggested by the adage ("Buy, not Sell, in May and Go Away", until October) a $10,000 account starting in 1950 would decline to only $9,019.90 by 2010. That is a 61 year total return of -9.8%.

Clearly the May through September periods have not been kind to investors over the past 6 decades.

Another way to look at the data is that ALL of the market’s gains over the past 61 years have occurred during the October through April periods.

Here are some additional data points to consider. For the entire 61 years, the market advances during the May-Sept. period 62.3% of the time. For the October-April period advances occur 72.1% of the time. I would note that this is not a substantial difference. In fact it is not the frequency of advances during the two periods which causes such vastly different 61 year total results. Rather, it is the average size of the advances that have such a wide variance.

The average gain achieved during the October-April periods over the past 61 years is 8.30%. For the May-Sept. period the average result is only a 0.31% gain. (Note - the average result is a positive, albeit small, number because it is calculated be simply adding up 61 yearly results and dividing by 61, compounding of returns is what results in an overall loss of 9.801% for the 61 year period).

Many of you are aware of the above statistics. So let's break things down further. First, here are the return figures for the two periods segregated into secular bull markets and secular bear markets. To wit:

During secular bull markets (a total of 33 years):

• May - Sept: +165.16%
  
• Oct. - April: +2,983.50%

During secular bear markets (a total of 28 years):

• May - Sept: -51.63%
  
• Oct. - April: +86.65%

Perhaps not surprisingly, the returns for both periods are much worse during secular bear markets. And they are quite distressing in the May - Sept. time frame.

As I believe we are still in a Secular Bear Market, we may be able to make effective use of the above knowledge. I intend to favor smaller positions (both long and short) and to be more inclined to shorting the market during this generally weak 5-month period.

Here are a few other tidbits which you may find interesting. During secular bear markets the average "Maximum Gain" during the May-Sept period is +5.50%. The average "Maximum Loss" during the same time frame is -8.83%. Thus we see that the average maximum upside attained at any time during the 5-months is only about 60% of the average largest loss.

It seems we could fairly say that fading the market during the May-Sept period of a given year on average carries generally less potential loss and greater potential gain. Of course many other factors should be considered before one actually takes any action.

The last piece of research I will pass on for now is, in my opinion, potentially the most intriguing. I will need to ponder this data point for a while longer before I am comfortable in applying it to my trading. I thus share it "as is".

For all of the 61 May-Sept periods, on only 3 occasions was the market's high-point (Maximum Gain) reached during the month of June. Further, for 33 of the 61 periods the high-point was reached on or after August 16th; more than 50%. In fact, on 28 occasions the Maximum Gain occurred in September.

I am struck by the implication here, at first glance, that a continuation of this May's decline into June could possibly set up a very attractive opportunity to become "net long" and ride a rally into the last 6 weeks of the May-September period.

Certainly many other factors could override this possibility. And there is in fact previous precedent for the market to show a maximum gain of