In our last article, we discussed the concept of long term trend-following and how it might be applicable to the S&P 500 index and the SPDR S&P500 ETF (NYSEARCA:SPY). We then evaluated what is perhaps the most simple trend-following idea possible:
- Buy if the market closes this month above the close "X" months ago
- Sell if the market closes below the close of "X" months ago
Note that this signal is like a switch that is either on or off. If you currently have a position (due to a prior signal) you simply hold until you have an exit signal. If you have no position, then you stay in cash until the next buy signal.
In this article, we will advance the concept a bit further with a moderately more complex trading signal. Rather than looking at the close this month relative to the close "X" months ago, we will use a moving average approach. A moving average is simply an average of the "X" most recent data points in a series that moves forward with time. In this case, we will be using a moving average of monthly closing prices. Here is our new trading rule:
- Buy if the close this month is greater than the "X" period moving average
- Sell if the close this month is less than the "X" period moving average
Like with our original system, once you are long you stay long until there is an exit signal. If you are in cash, you stay in cash until the next buy signal. Historic short term treasury rates have been used to simulate returns on cash holdings.
Let's Review the results using a 10 month moving average.
Maximum drawdown as measured by month-end closing prices has been reduced from about 52% to about 24%. Maximum drawdown is calculated as the largest percentage loss from a peak equity high.
One important component of evaluating a trading idea is to check for parameter sensitivity. In this case, our parameter is the length of time we are using to calculate the moving average. All else equal, it is better to see that a broad range of parameters are historically effective. An approach that is "easily broken" by parameter or subtle rule changes is less likely to be a valid in the future.
If you are interested in reading more about the NAS Trading approach to evaluating technical trading ideas, click here.
Here are the results using an 8 to 16 period moving average, stepping two months at a time:
Note that I have ordered the legend so that the results appear in the same order as on the chart. For example, cash is at the bottom of the legend, and is also the bottom line on the chart.
What we see is that all of the parameter combinations we are looking at between 8 and 16 months handily outperform buy and hold. The above studies have been created using S&P 500 index data, which is not an investable security. We used the index because it gave us a longer period over which to evaluate the idea.
If you would like to see how this system worked using the SPDR S&P500 ETF during its complete trading history, I have created and posted a supplemental study here.
In order to get a better idea of how and why the MA system pulled so far ahead, let's evaluate one of the periods where this approach added significant value. Between the fall of 1972 and late 1974, the market endured a significant selloff. At its worst point, the S&P500 was down about -46%. (The exact dates of the bear market depend on how you are measuring this - I am simply eyeballing the equity curve).
All of the moving average systems exited the market between January and April 1973, and only re-entered in early 1975, which allowed them to skip the vast majority of the decline.
Yet missing the 1973-1974 bear market was only part of the story. Interest rates during this period fluctuated around seven percent. This means that during a period where the market was getting decimated, the MA systems were able to stay in cash for about two years and earn this yield.
When the market finally turned around, the buy and hold portfolio had significant ground to recover, while the MA systems jumped on the next uptrend with equity already at peak highs.
Analysis of this one period of outperformance provides significant insight into the important role interest earned on cash plays in the total long term return for a market-timing system. This long term effect also brings to light how confiscatory our current interest rate policy (near zero percent interest rates) is towards savers. It is effectively a 100% tax on income from savings, yet it is never framed this way by the media.
The question arises: Is this study simply "looking at history?" Will the results contain practical insights for the future? This is a judgment call, just like everything else in the world of practical investing.
I first learned about this investment concept (applying long term trend systems to stock indexes) about 14 years ago while reading the book Market Wizards, written by Jack Schwager. Bruce Kovner (one of the interview subjects) stated that very long-term trend decision models work in the stock market. This book was published in 1988, and since that time the concept of using very long term trend models on the stock market has continued to be effective.
One other common criticism of trend-based models is that it is better to simply tell people to "stick it out" and not consider timing approaches. I would say that by definition, every investor must use a timing approach. After all, our investments usually serve a finite purpose over a defined period of time. In other words, not having a timing approach is itself a timing approach.
The real question is: Do you want a timing mechanism you have thought about and converted into a deliberate strategy, or do you want one that is guided by circumstance and even emotions?
One of the (unfortunately) most popular timing approaches used by investors is to panic after a large decline, and then to stay out of the market until it has rallied for a long period of time (If the investor gets back in at all). For those interested, this problem is the second point I address in this article.
My viewpoint is that any timing model that helps informed investors avoid this fate is likely to be useful - and a long term trend approach is certainly not the only approach to doing this.
Like with all investment strategies, mainstream acceptance of this approach will not be good for its continued success. Those of us who believe the results of long term trend systems are practical and significant can only hope that the majority will continue to disagree.