Market Timing using Harding’s STS

Can an investor effectively time the market? Time and time again, financial planners and articles on financial websites, magazines, and newspapers, emphatically declare "No!", sometimes using a concrete statistical analysis indicating the the stock market's direction can be more accurately described using random walk, a mathematical formalization of a trajectory (i.e. it cannot be predicted).

They argue (occasionally citing various economic journals) that it is nearly impossible for an individual to continually enter and exit the market at the proper inflection points. In fact, market timing critics, which include the majority of the financial services industry, suggest that the emotions involved in individual investing actually leads to a significant under performance of market averages. That is, when investing on emotion, it has been shown that individuals have the propensity to buy high and sell low, completely antithetical to the mantra of "buy low, sell high." Add excess trading fees accrued by the increased number of trades to those incorrect choices in the first place, and market timers who make their judgments based on emotion routinely under perform - unless they just happen to be lucky.

However, even those who swear by market timing typically concede to the fact that market timing based on emotion is a fool's game. Instead, they argue, that there are systematic approaches that take emotion out of the investing game and work for the long-term. We are here to investigate one such strategy: Sy Harding's Street Smart Report Seasonal Timing Strategy (STS).

Seasonal Timing Strategy Basics

Harding claims on his website that the traditional STS was down only 3.6% in 2008 compared to -36.1% for the S&P 500. Here's the chart comparing the STS to S&P 500 for 3-, 5-, and 10-year returns, taken from the above linked website (as of 6/2009):

Total Return / S&P 500 / STS
1-Year / -36.1% / -3.6%
3-Year / -22.3% / +22.4%
5-Year / - 9.6% / +47.2%
10-Year / -13.2% / +132.5%

Taking the last ten years on an individual basis, Harding's STS on the DJIA has outperformed the DJIA in seven out of ten years (according to his analysis; falling slightly short of the market average in 2003, 2005, and 2006). It was curious to me as to why he chose to do the analysis of the STS on the DJIA instead of the S&P 500. Was he trying to make his returns look better?

We're here to find out if these market timing claims are valid, what type of investors might consider market timing (if any), and the difference in performance throughout bull and bear markets.

Also, it's important to note that just because this strategy has or has not been effective in the past (we'll soon find out), that certainly is no indication of how it may play out in the future. As they say, past performance does not guarantee future results. In fact, in regards to mutual funds, I'd argue (and studies would back me up) that hot funds typically turn cold. This is a result of various factors, but most significantly the huge increase in the amount of institutional and individual monies pouring into the fund, which then inhibits the fund manager's ability to maintain his or her intended objective. It's a lot easier to take on risks and invest in small cap stocks, for example, without having a huge effect on the stock price when there is little money invested in your fund; when large amounts start pouring in, the strategy of the fund must adapt. Although mutual funds may exhibit this behavior, investing strategies don't necessarily have the same inherent restrictions and thus a correlation between the past and future may be more easily argued.

I am conducting this analysis also based on the information provided Les Masonson's text All About Market Timing, which outlines a wide variety of market timing strategy and will be reviewed in a future post.

Essentially, Harding's bare-bones version of the STS boils down to this:
  • Buy into the market on the next-to-last trading day of October
  • Exit the market on the fourth trading day of May
This simple strategy echoes the popular adage, "Sell in May and go away." Harding, however, takes it one step further, using the Moving Average Convergence Divergence (MACD) indicator to fine-tune the entry and exit points. The short-term signals of the MACD that occur throughout the year are completely ignored until these certain pre-determined calendar dates. This cuts down on trading fees, simplifies the strategy, and significantly reduces the amount of time required of the investor to make such decisions (the investor only has to look at the MACD indicator during a two week or so window twice a year).

If a buy signal from the MACD indicator happens before the calendar date, the investor is advised to undertake a position in equities earlier. On the other hand, if the MACD is stuck on the sell signal when the date approaches, the investor is supposed to wait until the MACD finally once again indicates a buy signal. The MACD is used the same way in May, except in the opposite fashion.

To summarize the STS using the MACD in its simplest terms requires just two points:
  • Buy on the first MACD buy signal on or after October 16
  • Sell on the first MACD sell signal on or after April 20
That's easy, right? Does something so simple actually work?

Historical Results and Analysis

Let's backtest it and see what the results show us. Note that in the following analysis, I don't take the t-bill or cash rate into account when using the STS strategy. Thus, the STS is a significant handicap when compared to buy-and-hold since the investment is guaranteed to return 0% in the months that it's not invested in the market. So, it's not truly a fair comparison, but I wanted to see how the STS would fare under conditions that perhaps aren't equal footing (and I didn't want to have to go through the effort of looking up the t-bill rates of all these months). So, if my final results indicate that the STS outperformed buy-and-hold, this outperforming figure actually understates the degree to which STS outperformed buy-and-hold, assuming a sensible investor had taken his or her money to t-bills or cash equivalents during the non-invested period. On the other hand, if STS underperformed under certain time ranges based on my analysis, the degree to which the STS underperformed the market averages is exaggerated. The power of compounding just a couple percentage points over the course of ten years is fairly significant, so clearly, I'm not giving the STS any breaks. Let's see how his strategy does given those caveats!

First, let's look at how the STS performs using the S&P 500 (a low cost index fund such as VFINX or an ETF such as SPY is most appropriate) during one of the largest bull markets in American history: 1991-1999. Remember, the STS recommends buying on or after October 16, and then selling on or after April 20, depending on the MACD indicator.

Huge Bull - Seasonal Timing Strategy
Buy Date Buy Price Shares Cost Basis Sell Date
10/17/1991 391.92 25.52 $10,000 5/14/1992
10/16/1992 411.73 25.60 $10,541 4/20/1993
10/18/1993 468.45 24.33 $11,396 5/11/1993
10/17/1994 468.96 22.90 $10,740 4/20/1995
10/16/1995 583.03 19.85 $11,572 4/20/1996
10/16/1996 704.41 18.26 $12,859 5/29/1997
11/6/1997 938.03 16.43 $15,409 4/20/1998
10/16/1998 1056.42 17.47 $18,458 4/20/1999

Sell Price Sell Value Difference % Net Change Cumulative
413.14 $10,541 $541 5.4% 5.4%
445.1 $11,396 $854 8.1% 14.0%
441.49 $10,740 -$656 -5.8% 7.4%
505.29 $11,572 $832 7.7% 15.7%
647.89 $12,859 $1,287 11.1% 28.6%
844.08 $15,409 $2,550 19.8% 54.1%
1123.65 $18,458 $3,049 19.8% 84.6%
1306.17 $22,822 $4,364 23.6% 128.2%
 Cumulative + 128.22%


Huge Bull - Buy and hold
         Year 31-Oct 31-Oct Difference % Net Change Cumulative
1992 392.45 418.68 $26.23 6.68% 6.68%
1993 418.68 467.83 $49.15 11.74% 19.21%
1994 467.83 472.35 $4.52 0.97% 20.36%
1995 472.35 581.5 $109.15 23.11% 48.17%
1996 581.5 705.27 $123.77 21.28% 79.71%
1997 705.27 914.62 $209.35 29.68% 133.05%
1998 914.62 1,098.67 $184.05 20.12% 179.95%
1999 1,098.67 1362.93 $264.26 24.05% 247.29%
Cumulative + 247.29%

So, the STS significantly underperforms buy and hold over this period. This is not surprising since this was a period in which nearly every month the market was up, so being out of the market 50% of the time would certainly have impacts on overall return.

Ok, let's take a look at a relative bear market then: 1999 - December 2009.

Note: While this post was originally written in June 2009, I have updated it with results through December 2009.

Relative Bear - Seasonal Timing Strategy
Buy Date Buy Price Shares Cost Basis Sell Date
10/22/1999 1301.65 7.68 $10,000.00 4/20/2000
10/23/2000 1452.3 7.59 $11,020.93 5/11/2001
10/16/2001 1097.54 8.61 $9,452.90 4/22/2002
10/16/2002 860.02 11.41 $9,814.90 5/19/2003
10/16/2003 1050.06 10.01 $10,508.20 4/23/2004
10/27/2004 1125.4 10.14 $11,414.26 4/18/2005
10/24/2005 1199.38 9.69 $11,622.99 4/17/2006
10/16/2006 1369.06 9.10 $12,455.91 5/10/2007
11/28/2007 1469.02 9.24 $13,569.62 5/8/2008
10/28/2008 940.51 13.73 $12,910.64 4/22/2009
10/16/2009 1087.68 10.65 $11,579.64 Now


Sell Price Sell Value Difference % Net Change Cumulative
1434.54 $11,021 $1,021 10.2% 10.2%
1245.67 $9,453 -$1,568 -14.2% -5.5%
1139.57 $9,815 $362 3.8% -1.9%
920.77 $10,508 $693 7.1% 5.1%
1140.6 $11,414 $906 8.6% 14.1%
1145.98 $11,623 $209 1.8% 16.2%
1285.33 $12,456 $833 7.2% 24.6%
1491.47 $13,570 $1,114 8.9% 35.7%
1397.68 $12,911 -$659 -4.9% 29.1%
843.55 $11,580 -$1,331 -10.3% 15.8%
1102.47 $11,737 $157 1.4% 17.4%
Cumulative + 17.37%

Relative Bear - Buy and hold
Year 31-Oct 31-Oct Difference % Net Change Cumulative
2000 1362.93 1429.4 $66.47 4.88% 4.88%
2001 1429.4 1,059.78 -$369.62 -25.86% -22.24%
2002 1,059.78 885.76 -$174.02 -16.42% -35.01%
2003 885.76 1,050.71 $164.95 18.62% -22.91%
2004 1,050.71 1,130.20 $79.49 7.57% -17.08%
2005 1,130.20 1,207.01 $76.81 6.80% -11.44%
2006 1,207.01 1,377.94 $170.93 14.16% 1.10%
2007 1,377.94 1,549.38 $171.44 12.44% 13.68%
2008 1,549.38 968.75 -$580.63 -37.47% -28.92%
2009 968.75 1066.11 $97.36 10.05% -21.78%
2010 1,066.11 1,102.47 $36.36 3.41% -19.11%
Cumulative -19.11%

During a ten-year period in which the S&P 500 was down nearly 20%, the STS was up 17%, an outperformance of an impressive 37% considering this strategy is so simple. Again, that 37% understates the outperformance if the investor had chosen to put his or her assets in t-bills, cash equivalents, or even bond holdings during the periods in which he or she was out of the market.  As of June 2009, STS was outperforming buy and hold by 56% for the ten-year period, but had a dreadful 2009 while the market was bullish throughout.

How about combining these two markets? Taking the period of 1991 - December 2009, a period beginning with one of the largest bull markets in history and ending with one of the largest bears:

Huge Bull followed by relative Bear - STS
Buy Date Buy Price Shares Cost Basis Sell Date
10/17/1991 391.92 25.52 $10,000 5/14/1992
10/16/1992 411.73 25.60 $10,541 4/20/1993
10/18/1993 468.45 24.33 $11,396 5/11/1993
10/17/1994 468.96 22.90 $10,740 4/20/1995
10/16/1995 583.03 19.85 $11,572 4/20/1996
10/16/1996 704.41 18.26 $12,859 5/29/1997
11/6/1997 938.03 16.43 $15,409 4/20/1998
10/16/1998 1056.42 17.47 $18,458 4/20/1999
10/22/1999 1301.65 17.53 $22,822 4/20/2000
10/23/2000 1452.3 17.32 $25,152 5/11/2001
10/16/2001 1097.54 19.66 $21,573 4/22/2002
10/16/2002 860.02 26.05 $22,400 5/19/2003
10/16/2003 1050.06 22.84 $23,982 4/23/2004
10/27/2004 1125.4 23.15 $26,050 4/18/2005
10/24/2005 1199.38 22.12 $26,526 4/17/2006
10/16/2006 1369.06 20.76 $28,427 5/10/2007
11/28/2007 1469.02 21.08 $30,969 5/8/2008
10/28/2008 940.51 31.33 $29,465 4/22/2009
10/16/2009 1087.68 24.24 $26,363 Now

 
Sell Price Sell Value Difference % Net Change Cumulative
413.14 $10,541 $541 5.4% 5.4%
445.1 $11,396 $854 8.1% 14.0%
441.49 $10,740 -$656 -5.8% 7.4%
505.29 $11,572 $832 7.7% 15.7%
647.89 $12,859 $1,287 11.1% 28.6%
844.08 $15,409 $2,550 19.8% 54.1%
1123.65 $18,458 $3,049 19.8% 84.6%
1306.17 $22,822 $4,364 23.6% 128.2%
1434.54 $25,152 $2,330 10.2% 151.5%
1245.67 $21,573 -$3,579 -14.2% 115.7%
1139.57 $22,400 $826 3.8% 124.0%
920.77 $23,982 $1,582 7.1% 139.8%
1140.6 $26,050 $2,068 8.6% 160.5%
1145.98 $26,526 $476 1.8% 165.3%
1285.33 $28,427 $1,901 7.2% 184.3%
1491.47 $30,969 $2,542 8.9% 209.7%
1397.68 $29,465 -$1,504 -4.9% 194.6%
841.5 $26,363 -$3,102 -10.5% 163.6%
1102.47 $26,721 $358 1.4% 167.2%

Cumulative + 167.21%


Huge Bull followed by relative Bear - Buy and hold
Year 31-Oct 31-Oct Difference % Net Change Cumulative
1992 392.45 418.68 $26.23 6.68% 6.68%
1993 418.68 467.83 $49.15 11.74% 19.21%
1994 467.83 472.35 $4.52 0.97% 20.36%
1995 472.35 581.5 $109.15 23.11% 48.17%
1996 581.5 705.27 $123.77 21.28% 79.71%
1997 705.27 914.62 $209.35 29.68% 133.05%
1998 914.62 1,098.67 $184.05 20.12% 179.95%
1999 1,098.67 1362.93 $264.26 24.05% 247.29%
2000 1362.93 1429.4 $66.47 4.88% 264.22%
2001 1429.4 1,059.78 -$369.62 -25.86% 170.04%
2002 1,059.78 885.76 -$174.02 -16.42% 125.70%
2003 885.76 1,050.71 $164.95 18.62% 167.73%
2004 1,050.71 1,130.20 $79.49 7.57% 187.99%
2005 1,130.20 1,207.01 $76.81 6.80% 207.56%
2006 1,207.01 1,377.94 $170.93 14.16% 251.11%
2007 1,377.94 1,549.38 $171.44 12.44% 294.80%
2008 1,549.38 968.75 -$580.63 -37.47% 146.85%
2009 968.75 1066.11 $97.36 10.05% 171.65%
2010 1,066.11 1,102.47 $36.36 3.41% 180.92%
Cumulative + 180.92%

The STS and buy and hold performed approximately the same over this period (again, the STS probably would have been about equal or even slightly above the S&P 500 as opposed to -13% if t-bill rates were considered). Also note that one might also consider the tax ramifications of buying and selling twice a year as opposed to holding positions over a significant period of time. This analysis does not take the tax benefits (writing off losses) or downfalls (short-term income tax as opposed to capital gains tax) into account.

I also did this analysis with an even more simple market timing strategy - simply avoiding the market each September and October, the two months that historically have performed the worst and have been the period in which significant pullbacks occurred most frequently. Following is a graphical representation of the cumulative performance of buy and hold, STS, and avoiding September and October.

Source: Self

Blue = STS; Green = Sept and Oct avoidance; Red = Buy and hold

As you can see, buy and hold is more volatile than either timing strategy, but all three end up at about the same place. Market timing is actually a conservative strategy. This is logical since you are out of the market for a significant portion of the year. Buy and hold was significantly better during the bull market of 1991-1999, but gave up almost all its gains in the following years. It you can't stomach the ups and downs, a market timing strategy might actually make more sense for you. You just have to be prepared for missing some gains if a huge bull is just around the corner. It's also interesting to note in the above graph that the STS and September and October avoidance strategically are quite similar.

Source: Self

Looking at a relative bear, you can see that STS outperforms both other strategies fairly significantly. Again, this is understated due to the caveat explained above. It's quite remarkable really thatthe STS is that even remotely reliable at capturing the majority of the market gains, while avoiding the market before it turns sour.

Finally, here is a year-by-year analyis of the three strategies.

Source: Self

Again, this re-affirms the conclusion that buy and hold is more volatile, while the timing strategies moderate both the yearly gains and losses.

Here's a chart summarizing the performance of the three strategies:


Strategy 1991 - 2009 1999 - 2009
STS +167.2% +17.4%
Buy and hold +180.9% -19.1%
S&O Avoid +188.2% -6.5%

In 2009 (as of 12/21/09 close), STS is down about 8% (perhaps the worst on record), while buy-and-hold isup a whopping 18.3%! Avoiding September and October is also up 15.8%. You can see Sy Harding's defense of the STS despite its 2009 performance here. He argues that the seasonality phenomenon still typically occurs; however, cases in which excess liquidity floods the market, like what happened in 2003 and 2008, have even greater consequences than the seasonality. Harding cites a study that explored the results from 1926 - 2006 and concluded that the seasonality patterns are robust over time and not likely data mining. In any event, it's certainly interesting to hear his perspective and respond to the criticism of this year's performance. It's clear that he has purposefully neglected to update his own website with the 2009 results in an effort to not broadcast them.

Conclusion

Finally, here are some summary points based on this analysis:
  1. During continuous bull markets, market timing will always lag buy and hold. While STS has shown to be able to capture most of the upside of the market, being out of the market around 50% of the year subjects your money to missing significant gains during continuous bull markets. This isn't unique to STS and all market timing strategies will underperform during huge bull markets - this is just a fact.
  2. Market timing is a conservative strategy. If you want to reduce risk, perhaps timing the market (with sound, systematic strategies and not emotion) is for you. Since exposure to the market is cut by about half, during bear markets these strategies offer significant protection against the downside.
  3. STS has historically been able to avoid the worse downturns. On the flip side, of course, is that while it captures the majority of the gains, it certainly misses out on some of them as well. Since 1999, STS is up 17%, while the S&P is down 19%, for an impressive 36% outperformance based on a simple strategy.
I'll admit that I was a skeptic of market timing before this procedure.  After reading various articles, texts, and performing my own individual analysis, I still think that the vast majority of people are best served by dollar-cost averaging into a low-cost index fund over time. However, my perspective has changed a bit, and I think that a simple market timing strategy (such as the STS) can be a sensible approach for a small segment of the population that hope to use a risk-reducing strategy. As they say, past performance is no guarantee of future results and those employing this strategy should not expect outperformance over the long-term similar to what occurred during the past 10-year bear market.  But it appears this seasonality affect does moderate volatility a bit.  An even simpler (and typically preferable) method to reduce risk would be to simply increase one's bond allocation.

1 comment:

Related Posts Plugin for WordPress, Blogger...