Wednesday, April 26, 2006

Stock Market Seasonal Tendencies

As I mentioned in my past post, the stock markets do show various "tendencies" over long periods of time. One of the most famous is the "Seasonal" tendency discovered by Yale Hirsch in 1986.

The basic "rule" is markets tend to have their best performance in the November-April period and their worst performance in the May-October period.

This "rule of thumb" was modified by Sy Harding in 1999 when he adopted a simple MACD indicator to move in and out of the market. During this time he revised the timing dates to Oct 16 to look to be long the market (based upon MACD confirmation) and April 20 to be short the market or in cash (again based upon MACD confirmation).

Here is some information from his website:

Note: graphs have been left out; if you want them please refer to his website.

Stock Market Seasonal Tendencies

The background:

It’s long been said that the best inventions and products come from taking a new look at existing processes, using information that’s right under our noses to improve on pioneering efforts that have gone before. That’s been true of everything from automobiles and aircraft, to heart surgery and search engines for the Internet.

Previous research on seasonality indicated that the market’s pattern of favorable seasonality begins October 1st, (based on the 1980s research of Ned Davis Research Inc.), or November 1st (based on the 1980s research of the Hirsch Organization), and ends May 1. However, the intention of both researchers was only to determine whether the market moves in recognizable seasonal patterns, and they apparently used month-end data to make that determination.

It was our intention to develop the market’s seasonality into a specific investment strategy, one that would work in both bull and bear markets. We began by back-testing 100 years of market data with the goal of determining the exact days of the year, rather than the month, that on average would produce the best entry and exit dates for investing according to the market’s seasonality.

We discovered that those best days on average are October 16 for the entry into the market for its favorable seasonal period, and April 20 for the exit from the market’s favorable season. However, those are just the best days as averaged over a very long time period. Obviously the market does not begin a rally on the same day each year, or begin to decline from a top on the same day each year. So we then concentrated on determining a means by which the entries and exits could be more accurately pinpointed as they change each year.

The result was what we called our Seasonal Timing Strategy, or STS.

Since the market does not begin or end its positive period on the same day each year, we combined the market’s general seasonal pattern with a technical indicator, the Moving Average Convergence Divergence indicator, or MACD. It is a short-term momentum-reversal indicator developed by Gerald Appel in the 1980s, designed to signal when the market has begun either a short-term rally, or a short-term correction.

The idea is that if a rally is underway when the October 16 calendar date for seasonal entry arrives, as indicated by the MACD indicator, we will enter at that time. However, if the MACD indicator is on a sell signal when the October 16 calendar date arrives, indicating a market decline is underway and will continue, it would not make sense to enter before that decline ends, even though the calendar says the best average entry date has arrived. Instead, our Seasonal Timing Strategy simply waits to enter until MACD gives its next buy signal, indicating that the decline has ended.

We use the same method to better pinpoint the end of the market’s favorable period in the spring. If MACD is on a sell signal when the calendar exit day of April 20 arrives, we exit at that point. However, if the technical indicator is on a buy signal, indicating the market is in a rally when April 20 arrives, it makes no sense to exit the market just because the calendar date has arrived. So our Seasonal Timing Strategy’s ‘exit rule’ is to simply remain in the market until MACD triggers a sell signal indicating the rally has ended.

Using this strategy we are able to take advantage of the fact that although the market’s favorable and unfavorable seasonal periods average approximately six months each, they actually vary significantly from year to year, sometimes being as brief as four months, other times lasting as long as eight months.

The chart shows the action of the DJIA from mid-1997 to mid-1999, which encompasses two of the market’s favorable seasonal periods. The lower window of the chart shows the DJIA itself, while the upper window shows the MACD indicator.

The vertical lines are the calendar entry days of October 16, and the calendar exit days of April 20 the following year.

Note at the left end of the chart that when October 16, 1997 arrived MACD was on a sell signal. The entry rule of STS is that we are not to enter until MACD triggers its next buy signal. As indicated by the up-arrow that did not take place until mid-November.

When April 20 of 1998 arrived, MACD was on a buy signal. The STS exit rule is that we therefore are not to exit until MACD triggers its next sell signal, which in this case was just a few days later, and actually at a lower price than had we used the calendar date.

Moving on to the entry in the fall of 1998, when October 16 arrived, the earliest entry date acceptable to STS, the MACD indicator was already on a buy signal, so we would enter at that point.

However, when the exit date of April 20 arrived the following spring MACD was on a buy signal, meaning the exit would be postponed until MACD triggered its next sell signal. That did not occur until mid-May, providing almost an extra month of higher prices before STS signaled that the market’s favorable seasonal period was over.

Note that MACD, like the calendar dates, does not get an investor in at the exact bottom in the fall, nor out at the exact top in the spring. No strategy could possibly do that. But MACD does most often provide a better entry and exit than simply using the calendar, and produces market-beating gains over the long-term by avoiding most serious market corrections and then getting back in at a lower level.

So the ‘seasonal investor’ would have made a total return of 60.2% from the entry on 11-17-97 to the exit 5-11-99. The DJIA closed at 7698 on 11-17-97 and at 11,026 on 5-11-99. So a buy and hold investor would have made only 43.2%. So the seasonal investor made 40% more over the 18 month period, and outperformed a buy and hold investor by the same amount.

Since risk management is an essential part of money management it’s also important to note that seasonal investing also significantly decreased market risk, as the follower of our STS strategy was in the market only 12 of the 18 months, and then only during the less risky favorable seasonal period.

The above chart and accompanying table also illustrates how the market’s seasonal periods vary from year to year. In the first favorable period, the seasonal investor was in the market for five months, while in the second favorable period he or she was in the market for seven months.

The next chart shows the entry and exit signals on the DJIA over the seven years from 1998 through 2004, which included the final two ‘bubble years’ (1998 and 1999) of the powerful 1990s bull market; the severe three-year bear market that followed (2000, 2001, 2002); and the new bull market that began in 2002.

Given the volatile economic and political changes that also took place during the seven years shown in the chart, it is an example of why we are able to say that the market’s seasonal patterns are consistent through all kinds of conditions, war and peace, economic boom times and recessions, no matter which political party is in power, whether interest rates or inflation are rising or falling.

Does the Seasonal Timing Strategy outperform the market every year?

Obviously not. It under-performed all of the major market indexes in 2003. But it does consistently outperform the market in most years, and particularly over the long-term, to a degree that we have never seen approached by any other strategy.

Sy introduced STS publicly in his 1999 book Riding the Bear - How to Prosper in the Coming Bear Market, as a strategy that would allow an investor to continue to profit in the final year of the 1991-2000 bull market and keep those profits, and then go on to prosper further in “the coming bear market”. In Riding the Bear he revealed the back-tested data going back to 1964, based on applying the strategy to the DJIA, reporting that it had tripled the performance of the DJIA over the period.

Yale and Jeff Hirsch reported in their newsletter;
"We applied Harding's system, which he developed based on the Dow's seasonal pattern, to the S&P 500. The results were astounding!" Smart Money, July, 1999

They also included the following in their year 2000 Stock Traders Almanac, and have continued to include it in each subsequent annual edition.

"Tested over the last 51 years, the strategy more than doubled the already outstanding performance of the basic 'Best Six Months' seasonal strategy."

Bloomberg Personal Finance Magazine reported;

"Remarkably simple but also remarkably profitable, at least in the hands of a disciplined practitioner like Sy Harding, editor of StreetSmart"

They say that imitation is the purest form of flattery. So it was interesting that in addition to praising our Seasonal Timing Strategy in The Stock Traders Almanac, Jeff and Yale Hirsch also adopted our STS as their own, replacing their ‘Best Six Months’ strategy in their newsletter, which approximately equaled the market’s performance, with our STS strategy, which has tripled the market's performance, calling it the “Best Six Month Strategy Plus MACD”. (That has caused me some embarrassing moments over the years, since Yale Hirsch is so well known that when reporters interview me in relation to the market’s seasonal patterns, many assume that my Seasonal Timing Strategy is actually the work of the Hirsch Organization).

What Creates the Seasonal Pattern?

Why would the market move in such consistent seasonal patterns regardless of the surrounding economic and political conditions?

The driving force is the same force that creates all sustained market moves, a change in the amount of money flowing into the market (an increase or decrease in the amount of money that is chasing a fixed amount of stock). Just as the extra money that flows into the market at the end of each month creates the short-term ‘monthly strength period’, so significant changes take place in the amount of money that flows into the market in pre-determined patterns in the fall and spring months, and produce the annual seasonal pattern.

Money Flow - The Driving Force of the Seasonal Timing Strategy.

As the market enters the fall season, investors begin receiving large chunks of extra cash. For instance, most mutual funds have fiscal years that end September 30 so they can get their books closed and make their capital gains and dividend distributions to their investors in November and December. Additionally, third and fourth quarter dividend distributions from corporations are paid to investors in the period between November and March. Investors begin receiving checks from their employer’s profit-sharing plans, from employers’ year-end contributions to their employees’ 401K and pension plans, Christmas bonuses, year-end bonuses, income tax refunds in the spring, etc. Highly paid hedge-fund managers collect their large year-end fees at the end of the year. Small business owners close their books at the end of each year, and by February or March their accountants let them know what their profits were, and they then distribute those profits to themselves.

Much of that extra cash finds its way into the stock market beginning in November, driving prices higher. Wall Street institutions, money-management firms, and knowledgeable investors, aware of the market's seasonal tendency, also begin buying more heavily, often in October, in anticipation that the market will make its usual impressive gains in the favorable season.

However, in the spring of the year that huge flow of extra money into the hands of investors dries up, with income tax refunds being the final act. That creates a sizable decrease in buying pressure, which allows whatever selling there is to have more influence on the direction of the market. It also deprives mutual funds and investors of the extra money needed to buy the dips, which might otherwise prevent a market decline from taking place.

In addition, Wall Street institutions, money-management firms, and knowledgeable investors, aware of the frequent effect of the market's unfavorable season on stock prices, tend to “Sell in May and Go Away”. Interest in the market also diminishes significantly in the summer months, as many investors and traders are off on vacations. That can be seen in the way trading volume dries up significantly during the ‘summer doldrums’.

Thus does the market tend to make most of its gains each year in the favorable seasonal period when many billions of dollars of extra money flow in, and suffer most of its losses in the unfavorable seasonal period when that extra fuel dries up, and is not there to offset any selling pressure that develops during the unfavorable season as a result of bad news or economic conditions.

Seasonality is also the answer to the age old question of why the stock market is sometimes able to ‘climb a wall of worry’ (shrug off bad news), while other times it seems unable to rally even on good news. It is simply that when large amounts of extra money are flowing in during the favorable season, the market is able to rally no matter that surrounding news may be negative. But when the flow of that extra ‘fuel’ dries up in the unfavorable season it is often difficult for the market to rally even if the surrounding news is positive.


After we introduced our Seasonal Timing Strategy to the general public in Riding the Bear in 1999, we received hundreds of letters with good questions that need to be addressed.

Is your Seasonal Timing Strategy valid only for the DJIA and S&P 500 indexes?

The Seasonal Timing Strategy was back-tested only against the DJIA and S&P 500 because the Dow data goes back to the late 1800s, while the S&P 500 goes back more than 50 years. So both provide enough data to be statistically meaningful and predictive.

And STS is used as one of our Street Smart Report newsletter portfolios in real-time utilizing only index mutual funds (or exchange-traded-funds (ETFs)), on the DJIA and S&P 500 for the same reason. We are not comfortable using some other index, for example the NASDAQ, or a sector index, to follow the signals since they were not specifically included in the research, and have not been around long enough to necessarily provide sufficient historical data to be statistically meaningful.

While other holdings might do as well or better, it is only by using a Dow or S&P 500 Index fund that we can say that if history is any guide, our Seasonal Timing System should continue to greatly outperform the Dow, S&P 500, and Nasdaq over the long term, while using only an index fund on either the Dow or S&P 500.

Further since the performance of the S&P 500 is the standard benchmark for the performance of money-managers and mutual funds, what better holding to use than the S&P 500 index itself if market-timing does indeed better the performance of a buy and hold strategy?

However, since when the market goes up to any degree it carries most stocks and sectors up with it, it’s reasonable to expect other indexes and sectors to have similar seasonal patterns.

Would the STS strategy be useful using managed mutual funds?

Probably. Again, since the stock market makes most of its gains in its favorable seasons, most stocks and mutual funds should also make most of their gains in the market's favorable seasons. However, individual managed mutual funds were not included in the research for very simple reasons. More than 75% of mutual funds were not in existence even 15 years ago. So there is no way to statistically correlate their past performance with how they might perform in the future. In addition, even if they have longer track records, a change in manager frequently changes the performance.

Have you considered (fill in the blank) as a possible improvement to STS?

The answer in general is that in the course of our research we tried many, many variations in our work to optimize the strategy, and what we have is about the best we could find. For instance, there are technical indicators other than MACD that have been a bit better in specific periods, but were not as consistent over the long haul.

What about brokerage firms and mutual fund managers that say there is no useful seasonal pattern?

Brokerage firms and mutual fund companies would have a tough time surviving if very many investors were aware of the market's seasonality and moved their money to cash for four to eight months every year. So they must go to whatever lengths they can to distort the information.

Their problem is that numbers don’t lie, they are what they are. So, invariably in trying to refute the very clear proof of the market's consistent seasonal patterns as best they can, these firms run their data from 1900, and even 1850, even though all of the research on seasonality shows the seasonal pattern did not begin to show up until 1950.

As noted before, the seasonal pattern is the result of the extra chunks of money that flow into investors' hands beginning in the fall, from distributions from mutual funds, from Christmas and year-end bonuses, from profit-sharing bonuses, from year-end contributions to 401 K, IRA, and Keough plans, from income tax refunds, and so forth.

However, there were no such extra chunks of money to create favorable seasons prior to 1950, because there were no mutual funds, no 401K plans, IRAs, Keough plans, very few money managers collecting monthly fees, no hedge-fund managers collecting 20% performance fees at the end of each year. The concept of companies sharing their profits with employees through profit-sharing plans had not yet appeared. Income taxes were non-existent (and when they were introduced were a tiny fraction of what they are today), so income tax refunds were not a factor. And so on.

No one who has engaged in research on the market’s seasonality has ever claimed there was a significant seasonal pattern prior to 1950. So when a brokerage firm tries to refute the market's seasonality by running their numbers from 1900 or 1850, and then claim that the advantage of the seasonal periods is too small to utilize, they have totally distorted the results they know they would have if they ran the numbers from 1950 when the market’s clear seasonal patterns began.

Further, they invariably do not include the interest on cash that a seasonal investor would receive in the unfavorable seasons. That may not seem like much in these times when interest rates are extremely low. However, over the long-term interest rates repeatedly cycle between being low and being high, with many periods when they have been in double-digits, when an investor would have added an additional 5% to 6% per year to their profits just in the six months they were out of the market. Leaving interest income out of back-testing historical data is a gross distortion of statistical analysis.

Additionally, none of the firms whose seasonal research developed into the ‘Sell in May and Go Away’ observation ever claimed that an investor would outperform the market by investing only in the market's favorable season of Nov. 1 to May 1. The research showed only that an investor so investing would have matched the performance of the S&P 500 over the last 50 years, while taking only 50% of market risk since they would only be in the market six months out of every twelve.

The brokerage firms acknowledge that when they say that “the advantage of the seasonal periods is too small to utilize, that one might as well remain invested on a buy and hold basis”. (In doing so they ignore the fact that risk management is a very important part of portfolio management).

However, most importantly, Wall Street’s attempts to refute the market's seasonality only refer to the calendar-based ‘Sell in May and Go Away’ maxim. They do not tackle our Seasonal Timing Strategy, which employs MACD to produce seasonal periods that vary from four to seven months in duration.

It is only the combination of the momentum reversal indicator with our more closely defined basic calendar dates that produces the back-tested and real-time performance, in which our Seasonal Timing Strategy approximately triples the performance of the S&P 500.

Does the fact that STS did not beat the market in 2003 mean seasonal timing no longer works?

Absolutely not! Historically there have been some individual years in which STS did not beat the market. Those years did not prevent the remarkable 50-year record, any more than the underperformance in 2003 prevented the record of the last 7 years.

What about the back-tested performance:

Going back further than just the last 7 years of real-time use, our STS also significantly beat the Dow and S&P 500 when back-tested over the last 10, 20, 40 and 50 year periods, with its normal two trades a year, using just an index fund.


Post a Comment

Subscribe to Post Comments [Atom]

<< Home

Your email address:

Powered by FeedBlitz