Investing based on the calendar has intrigued investors, market technicians, and other investment professionals for years. For the most part, the broadcast media have mentioned seasonal investing strategies more frequently in the past few years, but they have not given them the full attention they deserve. Therefore, most investors probably have heard about such strategies but have not taken them seriously or done any homework to determine their value. As it turns out, this lack of initiative on the part of most investors probably is a huge mistake because careful analysis of seasonal investing strategies reveals their superiority to the buyand- hold strategy over a long time period. Many academic studies have been performed on the seasonal influences on stock market returns. This chapter will present two profitable strategies that have worked not only in the past but also in the present, although not perfectly.
Investing during specific months each year is an example of a seasonality pattern that will be covered in detail in this chapter. According to the Wall Street Journal, Ned Davis Research found that since 1950, on average, stocks have gone up 8 percent from the beginning of October through the end of April but have increased only 1 percent from the beginning of May through the end of September.
That firm also found that from the years 1950 to 2000, investing $10,000 from the fourth trading day in May to the last trading day in September of the following year (or a continuous period of about six months) resulted in a miniscule total gain of $2,977 for that entire period. However, entering the market every October 1 and exiting the following May 3 resulted in a total gain of $585,909.
More recent data compiled by Standard & Poor’s (S&P) indicate that the best-performing months since January 1970 through September 2010 were March, April, November, and December (see Table 7-1). Therefore, the months of November through January continue to provide above-average performance. The average for all months was 0.64 percent.
Table 7-2 provides additional insight into the best-performing consecutive three-month periods. Note that the period November through January again offers the highest return compared with any of the other periods shown.
A research paper entitled, “The Halloween Indicator, Sell in May and Go Away: Another Puzzle” (September 1999), by Sven Bouman, ING Investment Management, Netherlands, and Ben Jacobsen, Faculty of Economics and Econometrics, University of Amsterdam, also weighed in on the best and worst months. Is the presence of the November–April period of stock market strength simply a manifestation of the U.S. economy, banking system, and markets? Not at all. Bouman and Jacobsen discovered that stock prices rose more sharply in the November–April time period than in the corresponding May–October period in 36 of 37 countries. This effect has occurred in the U.K. stock market since 1694. The authors found no evidence that this phenomenon can be explained
by the January effect, the stock market crash of 1987, dividend payment seasonality, or time-varying risk parameters. The main reason turns out to be the extent and timing of vacations. For example, summer vacations in Europe have a strong seasonal effect on financial markets.
A more recent paper entitled, “Are Monthly Seasonals Real? A Three Century Perspective” (October 25, 2010), which was authored by Ben Jacobsen (also coauthor of the study mentioned in the preceding paragraph) and Cherry Yi Zhang, both of Massey University’s Department of Economics and Finance in New Zealand, examined seasonal patterns over 300 years in the United Kingdom using monthly stock price data. They found that the “Sell in May and go away” (the Halloween strategy) wisdom that originated in the United Kingdom had a persistence of positive returns in all their 100- and 50-year subsamples and in 24 of 32 of the 10- year subsamples.
More than 80 percent of the time horizons over five years and more than 90 percent of the horizons over 10 years had returns that were, on average, three times higher than the market. The authors found that the average returns for the six-month period (called the winter period) are positive and higher than the summer returns for all sample periods. They found that the Halloween strategy consistently beat buy-and-hold over the entire sample period and in all 100- year and 50-year subsamples. Its only underperformance was in the 1941–1970 subsample period (30 years). Figure 7-1 shows the returns in the winter and summer months with the significant outperformance of the winter months. Figure 7-2 shows the trend and performance since 1693 compared with buy-and-hold that clearly demonstrates the consistency of investing in the winter period.
TESTS ON U.S. DATA
Let’s take a look at the individual monthly performance of the S&P 500 Index over 50 years. Mark Vakkur, M.D, a psychiatrist by training, as well as a stock and options trader, analyzed the historical data.3 His findings are shown in Table 7-3. This table provides the monthly performance of the S&P 500 Index (excluding reinvested dividends) from 1950 through March 2002 divided into two time frames. Clearly, there are specific strong and weak months. January, April, July, November, and December have been the bestperforming months from 1950 through 1995. Excluding July, the
continuous months of November through January are the best performing, with an average gain of 1.58 percent per month. By including February, March, and April, the performance falls to 1.19 percent, but it is still positive.
Now let’s look at the period from 1996 to March 2002 in the same table. The best-performing months were January, March, April, June, October, November, and December. March, June, and October have now emerged as strong months. Five months— March, April, June, October, and November—have provided higher performance in the most recent six-year period compared with the prior period. Except for June, these are all months in the October–April time frame.
The August 21, 2001, issue of the Formula Research newsletter, which is edited and published by Nelson Freeburg, a long-time trading systems developer who uses rigorous statistical testing on timing models, noted:
Since 1950 there have been 34 declines of at least 10 percent in the Dow Jones Industrials. In 31 of these cases, key portions of the pullback— often the brunt of the sell-off—occurred during the May–October period. Twelve declines took place entirely within the bearish period. The record since 1900 shows more “crashes,” “massacres,” and “panics” in October than in any other month. Of the 34 corrections cited above, fully 12 ended in October.
Freeburg goes on to say, “Over 90 percent of the net gain in the Dow since 1950 came in just 40 months, a mere 8 percent of the time span.” Furthermore, he continues, “To a remarkable extent, almost all the stock market’s advance since World War II came at a specific, recurring time of the year.”
History indicates that there is a consistent nonrandom period of strong and weak months. The risk-averse investor can take advantage of this strategy, especially in retirement accounts, where there are no capital gains implications for selling. As we’ll see throughout this chapter, specific strategies will be provided to capitalize on these strong and weak monthly patterns.
BEST SIX MONTHS STRATEGY
Astute investors should consider the preceding seasonal patterns because they have stood the test of time. The only seasonal strategy that I will cover in this chapter—the best six months strategy—has the following characteristics:
- There are only two signals a year—one buy and one sell.
- The buy signal is near November 1, and the sell signal is near April 30 (the beginning of the worst six months).
- The annual rate of return for the best six-month period beginning about November 1 each year exceeded the buyand- hold return of the same period. One reason for the strategy’s overall positive performance is that market crashes or corrections have tended to occur in the worst six-month period (e.g., September or October in 1929, 1974, 1987, 2001, 2002, and 2008).
- The strategy misses the brunt of bear markets because it is not invested in the weakest consecutive months of the year but instead is invested in the strongest consecutive months of the year.
- The strategy provides 50 percent less risk than buy-andhold because it is invested only half of the year, and the proceeds can be placed in a money-market fund during the other half.
- The time required to implement the strategy is minimal (about 10 minutes a year).
These attributes should whet most investors’ appetites because all the research data over six decades in the United States and over hundreds of years in the United Kingdom support it. The findings of well-respected researchers who have rigorously evaluated the U.S. stock market data are presented for your review. So let’s begin the journey of understanding how successful the best six months investing strategy has been and why you definitely should consider it for investing your hard-earned cash.
Very few simple timing strategies have such a long-term track record as this strategy, developed in 1986 by Yale Hirsch. (I’ll refer to it as the best six months strategy or the BSM strategy.) It was first published in the Hirsch Organization’s 1987 edition of its Stock Trader’s Almanac. It has been tweaked a bit over the years by Yale Hirsch’s son Jeffrey and is updated annually in each year’s edition of the almanac. Moreover, the BSM’s current buy and sell signals are provided in real time to subscribers of the monthly “Stock Trader’s Almanac blog,” as well as via automatic Internet updates.
The BSM strategy’s original buy and sell rules are simple: Invest in the stock market (e.g., an S&P 500 Index fund) on November 1 of each year, and then sell on April 30 of the following year (and go into cash equivalents—a money-market fund or T-bills) until November 1 of that year, when the next investment is made.
According to the Stock Trader’s Almanac 2011 (p. 48), from 1950 through 2009, an initial investment of $10,000 produced a total loss of $474.00 or a 0.4 percent average annual loss for all the May–October time periods.4 Compare this with a gain of $527,388 or 7.4 percent average annual gain during the November–April time periods. The strategy assumed that the funds were invested in the Dow Jones Industrial Average (DJIA).5 Without a doubt, the months selected for investing play a significant role in the total return over that extensive time period. On page 50 of the almanac, the same strategy is tested, but in conjunction with a technical indicator known as the moving-average convergence-divergence (MACD), developed by Gerald Appel based on the analysis of Sy Harding. This indicator was used to better identify the entry and exit dates in each of the periods. Using Hirsch’s enhanced MACD’s buy signal near November and sell signal near April, there was a gain of $1,472,790 or an average gain of 9.2 percent a year for the DJIA from November through May in the same period from 1950–2009, whereas the May through November period showed a loss of $6,542 or –1.2 percent a year. Clearly, this strategy almost tripled the previous strategy’s return. Sy Harding’s contribution to this strategy is covered in detail in the next section.
Keep in mind that for an investor who had been using the BSM strategy, the worst market meltdowns, as measured by the DJIA, all would have been avoided:
- The October 28 and 29, 1929, crash, in which the DJIA dropped 25.2 percent
- The October 19, 1987, stock market crash, in which the DJIA plunged over 508 points, dropping 22.6 percent
- The 555-point drop on October 27, 1997 (–7.2 percent)
- The 513-point drop on August 31, 1998 (–6.4 percent)
- The 357-point drop on August 27, 1998 (–4.2 percent)
- The 1,370-point drop the week of September 21, 2001 (–14.3 percent), after the terrorist attack on the World Trade Center and Washington, DC.
- The 1,651-point decline during the third-quarter 2002 (–17.9 percent)
- The 733.08-point drop on October 15, 2008 (–7.9 percent), the 678.91-point drop on October 9, 2008 (–7.3 percent), and 777.68-point drop on September 29, 2008 (–7.0 percent)
- The 1,874.19-point drop the week of October 10, 2008 (–18.2 percent)
Jeffrey Hirsch used a different monthly strategy employing the Nasdaq Composite Index instead of the DJIA to see how that index performed. In this case, he used the best eight months (November 1–June 30) and the worst four months (July 1–October 31) in conjunction with the MACD signal to identify the optimal dates. Since that index began in 1971, Hirsch ran the numbers from 1971 through 2009. He found that a $10,000 initial investment in 1971 mushroomed to $874,360 if invested during the best eightmonth period compared with a loss of $7,461 if it was invested during the other four months. Clearly, there is seasonality in the stock market, and it worked on this index as well, which further adds to the validity of the concept.
Sy Harding’s Street Smart Report’s Seasonal Timing Strategy
Sy Harding, founder and president of Asset Management Research Corporation, publishes the Sy Harding’s Street Smart Report newsletter on his Web site, www.StreetSmartReport.com. Harding is also the author of Riding the BEAR: How to Prosper in the Coming Bear Market (Adams Media, 1999), a paperback book that predicted the 2000–2002 bear market.
Harding was searching for a strategy that would allow an investor to make money in the bear market that he anticipated was about to occur. His research eventually led him to evaluating seasonality. Building on the pioneering work on seasonality of Yale Hirsch, Ned Davis, and Norman Fosback, Harding recognized that the market does indeed have a favorable season and an unfavorable season, but his additional research convinced him that the beginning of those seasons varies quite widely from year to year. Rather than being an even-keeled six-months-in, six-months-out situation, he found that the market’s favorable season can vary from as few as four months to as many as eight months.
His initial research first led him to use the next-to-last trading day of October as a strictly calendar-based entry and the fourth trading day of May as the optimal calendar-based exit. But his research also led him to believe that the calendar alone could not produce optimal results.
His seasonal timing strategy (STS) incorporates the MACD, a short-term momentum-reversal indicator, to better pinpoint the entries and exits. The indicator is used to determine if a rally has begun prior to the arrival of the calendar-based entry date. If so, the buy signal of the MACD indicator is used to provide an early entry rather than waiting for the actual calendar date. However, if the MACD indicator remains on a sell signal when the calendarbased entry date arrives, the rule is to delay the entry until the MACD indicator does trigger a buy signal.
If the MACD indicator triggers a sell signal prior to the arrival of the calendar date in the market’s favorable season, that signal is used as the exit signal rather than waiting for the calendar date. However, if the MACD is still on a buy signal when the calendar date arrives, the investor simply waits until it does trigger a sell signal before exiting.
Figure 7-3 shows Harding’s STS applied to the DJIA from 2000 through 2010, a period that encompassed very different market conditions— a strong bull market and two devastating bear markets.
The dotted vertical lines in the chart represent the typical calendar- based entries and exits. The arrows show the actual entries and exits of the strategy (as triggered by the MACD). Harding points out that the many short-term signals of the MACD throughout the year are ignored until within a week or two of the approaching calendar date. This means that the investor buys on the first MACD buy signal that takes place (or is already in place) on or after October 16 and exits on the first MACD sell signal that takes place (or is already in place) on or after April 20.
For comparisons of returns on an annual basis, Table 7-4 shows the conversion of the STS signals shown in Figure 7-3 to end-of-year numbers. Harding notes that the performance shown for the years 1999 forward is the actual performance of the actual Street Smart Report newsletter STS portfolio, which used various index funds in the favorable seasons. According to Harding, “There have been only seven years since 1970 in which the strategy was down for the year. The worst of those declines was the 3.6 percent decline in 2008, in the midst of one of the worst bear market years since the 1929 crash, and the 4.2 percent decline last year (2009). Prior to that, the worst decline for a year was 2.5 percent in 1977.”
Harding’s STS resulted in good performance during bull market years but, more important, not losing in the 2000–2002 period, and losing only 3.6 percent in the crushing year of 2008, but unfortunately, underperforming badly in the 2009 recovery year. Thus, as you can see, STS underperformed in the bull markets of 2003, 2005, 2006, and 2009, yet overall returns were better than buy-and-hold with 50 percent less risk—certainly a commendable live performance.
As Table 7-4 indicates, over the 11-year period the STS using the DJIA Index produced a return of 124.1 percent compared with 40.8 percent for the same index using the buy-and-hold approach. The Nasdaq Composite and S&P 500 indexes were able to produce only single-digit gains. The STS wins hands down during bear market years. For example, during the 2000–2002 bear market, where the averages sustained large losses, STS actually gained a cumulative 16.3 percent and did not have a losing year. Again, in the brutal 2008 bear market, where the major averages lost between 31.3 and 40.5 percent, STS lost a meager 3.6 percent. In the first half of 2010, the STS was up 6.7 percent, whereas the three major averages were down about the same amount. The key point is that bear markets devastate portfolios. Therefore, any strategy that can consistently avoid bear markets is welcome.
MARK VAKKUR’S CONTRIBUTION TO SEASONAL INVESTING
Mark Vakkur, M.D., mentioned in the beginning of this chapter, expanded on Hirsch’s BSM strategy. Vakkur compared six strategies with buy-and-hold and cash for the 1950–1995 period. His work was published in the June 1996 issue of the Technical Analysis of Stocks & Commodities magazine.8 Vakkur did not include reinvested dividends, which would have bolstered the buy-and-hold results.
Vakkur tested a number of different strategies, including:
- Avoid September entirely. Liquidate investments on the last trading day of August, and buy back on the first trading day in October. Then, don’t sell again until the last trading day of August of the next year. Do the same for all subsequent years.
- Invest using Hirsch’s BSM strategy using November entry and April exit dates.
- Invest during Hirsch’s worst months using a May entry and an October exit date.
- Adopt a switching strategy: Be 100 percent invested in the best six months of November, December, January, March, April, and July. Be 50 percent invested and have 50 percent in cash (T-bills) in the next best three months of February, August, and October. And be 100 percent in cash for the worst three months of May, June, and September.
- Have a 2:1 leverage, where 50 percent is invested with margin (2:1 leverage) for the best six months. As defined in strategy number 4, 100 percent is invested in the next best three months, and 100 percent is in cash for the worst three months.
- Invest in all months of the year with 2:1 leverage.
Results of Vakkur’s Analysis
Table 7-5 shows the performance results of Vakkur’s strategies. During this 45-year period, simply buying and holding the S&P 500 Index with a $10,000 stake in 1950 resulted in a total ending principal in 1995 of $372,388, which is equivalent to an average annual return of 8.4 percent. Investing solely in cash (T-bills) for the
entire 45 years rather than in the S&P 500 Index resulted in a total ending principal of $110,905. This provided a 5.5 percent annual return, with $261,483 less than buy-and-hold. Cash is not a viable investment, especially when adjusted for inflation.
Now let’s review the results of not investing during September. By eliminating this one month each year, the total proceeds were $624,135, or an annual average return of 9.6 percent, which is $251,747 better than buy-and-hold. Moreover, the standard deviation (measure of volatility) and maximum 12-month loss encountered were lower than buy-and-hold. This means that an investor was getting a higher return with less risk. This is the smart way to invest.
According to the Formula Research newsletter (August 21, 2001), September has been the worst month over the past 25, 50, and 100 years: “In a study of 20 global markets from 1970 to 1992, September was the only month with negative returns in all 20 cases.” The Hirsch BSM strategy of being invested from November to April did even better, totaling $703,935, or a 9.9 percent average annual return. This performance was $331,546 better than buy-andhold. This strategy also had a much lower standard deviation and lower maximum 12-month loss than buy-and-hold. The opposite strategy of buying and remaining invested during the six unfavorable months of May through October resulted in an ending value of only $58,670, or 4 percent annualized with a lower standard deviation and lower maximum 12-month loss than buy-and-hold. Clearly, the worst six months are consistently poor in all respects.
A switching strategy also was tested, where a 100 percent invested position was taken during the best six months (as mentioned previously). Additionally, a 50 percent invested position and 50 percent cash position were taken for the next-best three months, respectively, and a 100 percent cash position (T-bills) was taken for the worst three months. This strategy resulted in an ending value of $997,620, with a 10.8 percent annual return, translating into $625,232 more than buy-and-hold. In essence, although the switching strategy is less risky than buy-and-hold, it returned 2.7 times the principal of buy-and-hold. This is an outstanding compromise of risk versus return.
The next strategy tested used margin for more leverage. The strategy of 2:1 leveraging consisted of using 50 percent margin during the best six months, 100 percent invested with no margin for the next-best three months, and going into cash during the worst three months. This strategy had an ending value of $1,839,958 and provided a 12.3 percent annual return. This translates into $1,467,570 more than buy-and-hold, an astounding difference. Of course, this strategy was more risky than buy-and-hold because of the leverage, and it had a higher standard deviation and larger yearly maximum gains and losses. Remember, though, that risk and reward usually go hand in hand. In this case, the higher risk turned into an exceptionally high reward, with a 394 percent increase in value. With the advent of leveraged mutual funds and leveraged exchange-traded funds (ETFs), this type of strategy now can be used without the need for the cost of margin.
Performance from January 1996 to March 2002
Vakkur updated his 1996 data by publishing an article in the September 2002 issue of Active Trader magazine.9 He did not provide the risk-adjusted return and 12-month maximum gain for this period, as he had in the previous article, nor did he include reinvested dividends in his calculations. And he didn’t include the more aggressive strategies in his latest analysis. Table 7-6 provides the data for this latest time period.
Again, buy-and-hold is beaten by both the November–April strategy and by the September-avoidance strategy. In both instances, the annual returns were greater than buy-and-hold; the standard deviations and the maximum 12-month losses were lower as well. As expected, 100 percent cash fared poorly but better than investing during the weak monthly period of May to October. Moreover, the opposite strategy of buying in the weak period again showed the worst results. Thus, even during the last six years, the seasonal strategies continue to work as expected. This is consistent with Vakkur’s prior (1996) work. Since that study was published, Vakkur has not updated it or it would have been included here.
Additional Testing of September Avoidance Strategy
Formula Research (August 21, 2001) also tested Vakkur’s Septemberavoidance strategy from 1950 through 2001.10 With an initial $10,000 stake, this strategy would have returned an annualized total return of 13.4 percent, worth $5.6 million. During this same period, buy-and-hold would have generated a return of 12.5 percent and earned $3.9 million, whereas the Hirsch BSM strategy would have returned 12.1 percent a year, totaling $3.3 million. Clearly, the September-avoidance strategy has significantly outpaced the other two strategies. And remember that in the month of September in the years 2000, 2001, 2002, and 2008, the Nasdaq dropped –12.7, –17.0, –10.9, and –11.6 percent, respectively. Another four recent years of bad Septembers—the worst on record for the Nasdaq Composite.
Robert W. Colby also tested the September-avoidance strategy using the DJIA over a 101-year period from 1900 through 2000. In addition, instead of going to all-cash in September, Colby sold short on the last trading day in August and covered the short on the last trading day in September. Starting with $100 in 1900, this strategy would have resulted in a profit of $164,048, which was 644 percent greater than buy-and-hold (ending balance of $22,055).
Colby also tested a slight variation of this strategy—buying on October 27 (or the next closest trading day if the market was closed on that day) while selling on the next September 5 (or the next closest trading day if the market was closed on that day). The same short strategy was used on September 5 and covered on October 27. This slight adjustment to the original test provided a substantial increase in profits to $644,467 compared with $20,699 for buyand- hold. Thus this strategy beat buy-and-hold by 3,014 percent!
Of course, this strategy is not perfect. Not investing in September 2008 was an unexpected gift because the financial crisis was crushing Wall Street with a Nasadaq Composite loss of 11.6 percent. However, if you decided to sit out September 2010, you made a very big mistake because all the major averages had one of their best Septembers in decades, with the Nasdaq Composite up 12 percent, followed by the S&P 500 up 7.7 percent, and the DJIA up 6.7 percent.
Now that you are thoroughly enlightened by the extensive research about the BSM strategy and the month of September, you’re probably wondering how to go about using it to your advantage. You should consider the following alternatives:
- Use the MACD indicator with the BSM strategy. Use a free charting Web site such as www.bigcharts.com or www.StockCharts.com, or chart the index you are tracking (e.g., the DJIA or the S&P 500) for the past year using daily prices. Place the MACD indicator with settings of 12, 26, and 9 on the bottom of the chart. In the mid-April to May time frame, look for a MACD crossover sell signal to the downside (from above the 0 line). Likewise, in the mid- October to November time frame, look for a MACD crossover buy signal to the upside (from below its 0 line).
- Avoid the stock market entirely each September. If you’re not using the BSM strategy, then you can avoid the stock market in September. Simply sell on the last day of August and buy on the first day of October. For the more venturesome investor, follow Colby’s September strategy of shorting (using inverse mutual funds or ETFs) during September and covering and going long in October. And you may want to consider using his other strategy of shorting (using inverse mutual funds or ETFs) on September 5 and covering the short on October 27 and then going long.
- Subscribe to Sy Harding’s Street Smart Report monthly newsletter for his STS signals for $275 a year. Harding provides his newly revised methodology for the MACDbased signals via his hotline and on his Web site. This seasonal strategy is only one feature of this newsletter and hotline service. See Chapter 12 for complete information on this newsletter.
- Subscribe to the monthly Stock Trader’s Almanac for $295 a year. Hirsch tracks the BSM strategy and provides his MACD buy and sell signals when they occur via Web and phone access. This is only one feature of the many features of this newsletter. Be aware, however, that Harding and Hirsch use a slightly different methodology for obtaining their MACD signals, so you have to decide which one has the best track record. In 2003, for example, Hirsch’s MACD sell signal was triggered on April 10, whereas Harding’s was hit on May 19. Harding’s signal captured a higher return than Hirsch’s signal.
It’s your choice which alternative BSM to use based on your ability to work with the MACD and the charting software and on your understanding of the strategy. As far as actually using the September-avoidance strategy or the strategies with leverage illustrated by Vakkur, that choice is solely up to you based on your level of knowledge and comfort with the risks these strategies entail. As far as which investment vehicles to use, you can select from any of those mentioned in previous chapters, such as index funds, sector funds, leveraged funds, and the many ETFs. Be sure to use stoplimit orders on ETFs in case the buy strategy does not work in certain years, which is always a possibility. You would reenter the market on the next year’s buy signal.
Finally, to visually observe the BSM (without using the MACD) performance over 60 years, look at Figure 7-4, which compares the BSM with the worst six months and the S&P 500 Index from 1950 to 2010 October. Clearly, with 50 percent less risk than
buy-and-hold, a higher average annualized return, and the simplicity of executing this strategy, it is definitely one to consider.