Buy-and-Hold Versus Market Timing
The market collapse of 2008–2009 has led some investors to question the merits of “buy-and-hold” investing. This is not surprising, seeing that this was the second time in this decade the market has fallen by more than 45%. It has also led to renewed interest in ways of sidestepping such market meltdowns—namely, market timing. For proponents of Burton Malkiel’s “random walk” theory, market timing is a fool’s errand. They argue that you cannot use past market activity to predict future movements.
Print this article
In this article
Share this article
Poking Holes in Buy & Hold
In a series of articles for Advisor Perspectives, Inc. (www.advisorperspectives.com) between May and July 2009, Theodore Wong set out to defend the veracity of market timing using one of the simplest mechanical trading systems—the moving average crossover ( . Originally an MIT-educated electrical engineer, Wong shifted his focus to investing and earned an MBA. He now combines his engineering analytics with econometric modeling to create quantitative investing strategies.
His goal wasn’t to present the MAC as the best means of timing the market. Instead, his aim was to see whether it was possible to use active investment management to outperform a buy-and-hold approach over a long period of time.
The “Missing Out” Argument
The first of Wong’s articles tackles a common argument used to discredit market-timing—the notion of “missing out.” Missing out refers to the impact on investment returns if someone were to mistime the market and “miss out” on some of the best investing days. However, Wong points out that missing out is a two-way street; when considering the impact of missing the best days, you should also consider what happens if you miss the worst days.
For his research, Wong analyzed stock market price data maintained by Yale economics professor Robert Shiller over the period starting at the beginning of 1871 and going through the end of April 2009. The data, which is updated daily, is available for free online at (www.econ.yale.edu/~shiller/data/ie_data.xls). In the spreadsheet, the price data is referred to as the S&P Composite Stock Price Index. However, Standard and Poor’s did not introduce its first stock market index until 1923, and the widely followed S&P 500 index did not arrive in its current form until March 4, 1957. According to “Market Volatility” (The MIT Press, 1992), a book Shiller wrote that also makes use of the data, the data dating back from the start of 1871 was compiled using Standard & Poor’s Statistical Service “Security Price Index Record,” from tables entitled Monthly Stock Price Indexes—Long Term.
Analyzing this data set, Wong made some interesting discoveries:
- A buy-and-hold investor would have earned a compound annual growth rate of 8.6% over the period;
- Excluding the best 24 months over the period would have lowered the buy-and-hold benchmark return to 6.4%;
- Excluding the worst 24 months would have increased the return to 11.5%.
Examining daily stock market data from January 1942 through April 2009 yielded even more striking results:
- The buy-and-hold benchmark annual return for the period was 10.0%;
- Excluding the best 50 days lowers the annual return to 6.1%;
- Eliminating the worst 50 days increases the annual return to 15.2%.
Wong also discovered that excluding the best and worst periods over both timeframes did not give the buy-and-hold investor any appreciable advantage. In fact, looking at daily data from 1942 to 2009, missing the best and worst days would have boosted your annual return relative to the buy-and-hold return. As Wong concludes: “Who would mind getting similar returns to buy-and-hold without the volatile extremes?”
Drawdown: Measuring “Emotional Pain”
Wong also addresses another market phenomenon that seems to benefit buy-and-hold investors—the upward bias of the market. This means that, in theory, the longer you hold an equity investment, the more likely you are of earning a positive return.
To analyze the breakeven holding period, Wong calculated the “drawdown” of the stock market total return index over the 137-year period from 1871 through April 2009. Drawdown is the percentage decline from the last equity peak. Wong describes drawdown as the “emotional pain” investors endure while holding stocks that have fallen in value. For example, in the last market downturn, many investors experienced too much emotional pain and moved their investments to cash. As a result, they missed out on the significant rebound in the market between March 2009 and April 2010.
Wong discovered that the market was below breakeven a staggering 92% of the time from 1871 through April 2009 (meaning it was at or above the previous peak only 8% of the time). Even during the supposed “bullish” 1990s, the market was underwater over 50% of the time. Keep in mind that this means the market could have been down any amount, no matter how slight. However, Wong also found that, between 1871 and 2009, the market was down from its previous peak by more than 40% fully one-third of the time. This seemingly dispels the notion that such severe market declines are “once-in-a-lifetime” events, as the last decade confirms. No matter how non-risk-averse some investors claim to be, Wong doubts that many would be able to stay fully invested in the market faced with such frequent and pronounced market downturns.
In addition, Wong’s analysis indicated that it can take a long time to recover from large drawdowns. Using an extreme example, it took 26 years for an investor to recover from the losses of the 1929 stock market crash and Great Depression.
What Diversification Benefit?
Most investors will tell you that the best way to lower the overall risk of a portfolio is with diversification. By constructing a portfolio of low or negatively correlated assets, we lower the risk inherent to individual assets (unsystematic risk). (Correlation is the degree to which assets move together.)
However, no amount of diversification can eliminate all risk from a portfolio. During bear markets, as the most recent one made abundantly clear, most stocks will decline. However, the economic shocks that hit the market in 2008–2009 caused almost all asset classes to fall in value—U.S. and foreign equities, real estate, and commodities. In this case, Wong feels that the benefits of diversification “failed” investors at exactly the time they needed it the most. As it turned out, only cash and U.S. Treasuries bucked the stock market’s downward trend in 2008.
Unfortunately, many investors viewed diversification as holding different types of equities, without sufficient holdings of cash or bonds. But, diversification did seemingly do its job, according to Sam Stovall of S&P Equity Research. In his March 2010 AAII Journal article [“Diversification: A Failure of Fact or Expectation?”], he showed that a portfolio invested 60% in large-cap U.S. equities and 40% in long-term government bonds experienced a loss of around 13% in 2008, versus a loss of 37% in the S&P 500 total return index.
It is Wong’s opinion that allocating your portfolio to a greater percentage of cash or bonds as the market declines is not market timing, it is merely prudent diversification.
The MAC System
After providing some compelling arguments as to why market timing deserves a closer look, Wong shifts his attention to illustrating a way in which investors can use market timing to lower risk and preserve capital during market downturns. He is quick to point out that he is not advocating an optimal strategy, nor does he claim to have found investing’s “holy grail.” He is only attempting to disprove the myths often associated with market timing.
To illustrate his point, Wong uses one of the simplest trading systems used by investors—the moving average crossover, or MAC. In articles for Advisor Perspectives, market timers Brian Schreiner and Mebane Faber wrote about using a 10-month MAC to avoid the bear markets of 2000 and 2008. [You can find these articles, as well as those written by Theodore Wong, by going to the Advisor Perspective’s website and using the search function.]
With a moving average crossover system, you buy when the price crosses over (rises above) the moving average and you sell when the price drops below the moving average.
Exponential Moving Average (EMA)
Wong prefers using an exponential moving average (EMA), which reduces the lag between the actual price and the smoothed average by applying more weight to recent prices. The weighting applied to the most recent price depends on the number of periods in the moving average. There are three steps to calculating an exponential moving average:
- Calculate the simple moving average (SMA). An EMA has to start somewhere, so a simple moving average is used as the previous period’s EMA in the first calculation;
- Calculate the weighting multiplier (discussed momentarily);
- Calculate the exponential moving average.
Table 1 shows the formula for a 10-period EMA.
|1. SMA||=||10-period sum ÷ 10|
|2. Multiplier||=||[2 ÷ (Time periods + 1)]|
|=||[2 ÷ (10 + 1)]|
|3. EMA||=||[Close – EMA (previous day)] × multiplier + EMA (previous day)|
A 10-period exponential moving average applies an 18.18% weighting to the most recent price. Notice that the weighting for the shorter time period is more than the weighting for the longer time period. In fact, the weighting drops by half every time the moving average period doubles. This reduces the lag between the actual price curve and the smoothed moving average curve. Table 2 illustrates the calculation of a 10-day EMA using closing price data for the S&P 500 total return index from April 30, 2010, through May 28, 2010.
2/(10 + 1)
|9||5/12/2010||1171.67||1161.63||<< 9-day SMA|
To learn more about moving averages and using a spreadsheet to calculate them, see the Spreadsheet Corner column in this issue.
Testing the MAC System
When testing the MAC, Wong outlines three explicit assumptions:
- All proceeds from selling are held in non-interest-bearing cash;
- There are no transaction costs (commissions, bid-ask spreads, etc.); and
- There are no tax implications.
In addition, he assumes that he is buying and selling in the same month the signal is generated. He says it is possible to do this buy trading in the closing moments of the last trading day in the month when it is apparent what the signal for that month will be.
As mentioned earlier, a buy-and-hold investor would have earned around 8.6% a year between 1871 and April 2009 by investing in the total market. Using exponential moving averages (while he doesn’t explicitly state this, we assume he uses exponential moving averages since he states his preference for them) ranging in length from two to 23 months, Wong found that moving averages of 10 months or less consistently outperformed the buy-and-hold benchmark return. Furthermore, on a risk-adjusted basis (using the ratio of the compounded annual growth rate to the annualized standard deviation of monthly standard deviation), the MAC beat buy-and-hold over all moving average lengths.
Wong feels that standard deviation as a risk measure has one significant drawback—it treats volatility to the upside the same as downside volatility. In his opinion, however, investors who have long positions in stock do not have a problem with above-average gains. They are concerned only with downside risk. Therefore, Wong looks at drawdown to consider only the downside risk. To reiterate, drawdown is the percentage decline from the latest equity peak.
Wong’s analysis found that buy-and-hold investors experienced a maximum drawdown of –85% from January 1871 through April 2009. This took place between the pre-crash peak of 1929 through the trough of 1932. The average drawdown over the entire period from 1871 to 2009 was –26%. Using exponential moving averages from two months to 23 months in length, Wong found a maximum drawdown of –15% and an average drawdown of –4%. This indicates that the MAC offers significant downside protection during market declines relative to buy-and-hold.
Table 3 summarizes Wong’s analysis, presenting data for the buy-and-hold benchmark as well as the six-month MAC, which yielded the best results, and the 23-month MAC, which did the worst.
While transaction costs (commissions) were not considered when calculating the performance of the MAC system, Wong was aware of their potential negative impact. However, he found that the MAC would have generated, on average, less than 0.4 round-trip trades (buy and sell) a year across all moving average lengths. Over 137 years, this translates into one round-trip trade every 2.6 years. Even using a two-month moving average, which is most susceptible to price changes, would have generated 0.9 round-trip trades a year or one round-trip trade every 1.1 years. Not only does this show that transaction costs would have had only a negligible impact on returns, it also shows that the MAC system, on average, would generate only long-term capital gains (which are taxed at a lower rate than ordinary income). However, Wong does acknowledge that over shorter periods, it is possible for the MAC to generate false signals, also known as whipsaw, that would have investors entering and exiting trades on a more frequent basis.
Wong had seemingly found a market timing system that outperforms buy-and-hold on both an absolute and risk-adjusted basis over the last 137 years. He proceeded to examine the MAC returns by decades across the entire period. Specifically, he wanted to see how the MAC performed relative to buy-and-hold during bear markets.
Between January 1871 and June 2009, $1 invested in the S&P 500 total return index using the six-month MAC system would have grown to $332,000. In stark contrast, the same investment in the index using buy-and-hold would have grown to $105,000. However, Wong was concerned that the 1929 stock market crash may have skewed the results in the MAC’s favor. So he looked at the period from January 1941 to June 2009 and found that the six-month MAC and buy-and-hold yielded almost identical results—roughly a 1,000% cumulative gain. Over these seven decades, however, buy-and-hold outperforms MAC most of the time. It was not until the most recent bear market that the returns equalized.
Wong then broke the 137-year test period into decades. Over the 14 decades between 1871 and 2009, buy-and-hold outperforms MAC in only six of them, but outperformance in five of those decades was only by slight margins. In all of the decades where buy-and-hold outperformed MAC, the margin was never more than 4% over 10 years, or less than 0.4% a year.
In contrast, in eight of the 14 decades in which MAC outperformed buy-and-hold, the margin was never less than 4%.
Since 2001, MAC has outperformed buy-and-hold by over 8% (although the decade isn’t quite yet over). In addition, the current decade thus far is the only one where buy-and-hold has experienced a loss over the last 137 years.
Over the entire 137-year test period, buy-and-hold outperformed the MAC over half of the time. However, Wong believes that buy-and-hold has benefited from the strong “secular” bull markets of the last six decades as the U.S. economy and its stock market exploded following World War II. In a secular bull market, strong investor sentiment drives prices higher, as there are more net buyers than sellers. Furthermore, he feels that the research that supports buy-and-hold is based on the secular bull markets over this period. As a result, the research has concluded that no one can outperform the market.
However, the last decade has painted a very different picture as investor sentiment has dipped in the face of more frequent economic shocks. As a result, investors have not had the benefit of these same bull markets to recover from the 2000 market bubble. Furthermore, they were hit by another market collapse only eight years
|Buy & Hold||8.6||84,660||23.8||-25.9||-81.8|
Source: Theodore Wong/Robert Shiller.
later. Wong’s research has led him to conclude that buy-and-hold works during periods of economic expansion, but underperforms during economic slowdowns and contractions. Once again, in his opinion, this opens the door for market timing.
If you had invested $1 in the S&P 500 total return index at the beginning of 2000, this investment would have fallen to less than $0.80 through mid-2009. By contrast, $1 invested in the index using the six-month MAC would have grown to more than $1.40 over the same period (this mimics the investment return if you had bought and held the S&P 600 SmallCap total return index over the same period).
In the end, Wong contends that the MAC doesn’t predict markets; it merely follows market trends. The MAC doesn’t sell at market peaks or buy at market bottoms. What it does do, as his research seems to indicate, is preserve wealth in bear markets and accumulate wealth in good times. The question he poses is whether we believe a strong secular bull market will return once this recession is over. Only time will tell.