A Trend Following Approach

At RQA, we spend a significant amount of time exploring and researching sound investment strategies that not only provide attractive risk-adjusted returns over time but are also differentiated and unique, especially in comparison to traditional assets classes and Traditional Portfolios.  The goal of our strategy research is to develop and implement complementary return streams that can ultimately enhance investor portfolios, particularly those that are heavily dependent on more common approaches, such as long-term buy-and-hold positions in stocks and bonds.  By adding these unique, uncorrelated return streams to more common, base-level portfolios, we can increase the probability of realizing better absolute and risk-adjusted returns over time.  

A strategy that fits this description well and has a long history in both academia and real world implementation is trend following.  In what follows, we will illustrate how and why this strategy works and the benefits it can provide.  Additionally, we will review the historical application of a simple trend following methodology on the Traditional Portfolio, creating a “Trend Portfolio”, which we can then analyze and compare to more traditional, long-term investment approaches.


Trend following strategies are commonplace in the alternative investment landscape and have been executed largely within the futures markets for decades. The basic thesis around trend following is that asset price returns have been observed to be autocorrelated over time - meaning prices that are rising typically continue to rise, and prices that are falling typically continue to fall. Trend following seeks to capture and profit from the behavioral aspects of a marketplace where herd mentality takes over and large trends unfold for extended periods of time.

Using rules-based or systematic approaches to capture these trends has been proven by many investors and traders over the years to successfully generate attractive risk-adjusted returns on a persistent basis. To further support this claim, numerous studies (AQR, CFM) have researched in great depth the persistent nature of trends in global markets and the various techniques used to capture them. Additionally, as a very interesting data point, the CFM whitepaper cited above analyzed over 200 years of price data across various markets and concluded that trend following was an extremely robust and statistically significant strategy for generating superior risk-adjusted returns over time. More specifically (and more surprisingly), the paper noted that the probability of trend following strategies producing their long history of superior performance by random chance was equal to about 1-in-1023 (i.e. a hundred thousand times more than the number of grains of sand on Earth).

Capturing a Trend

So, how do we effectively capture a trend in financial markets?  A basic way to isolate whether a certain asset class is rising or falling is to simply apply a moving average filter to the historical prices of that asset.  By plotting a moving average alongside the price of the underlying asset, we can quickly determine if the asset is trending positively or negatively.  For reference, see the chart in Figure 1, which shows a monthly price chart of the Vanguard Total U.S. Stock Market Index (“U.S. Stocks”) along with its respective 8-month simple moving average (“SMA”) - calculated by averaging the prior 8 months of the U.S. Stock Index monthly closing values.  Here we can see that when the U.S. Stock Index (blue line) is above its 8-month SMA (black line), U.S. Stocks are visibly in an uptrend, and vice versa.  

To further expand upon this fairly simple methodology for more a diversified application, we will next use the 8-month SMA as our trend filter and apply it to each of the asset classes within the Traditional Portfolio, effectively deriving our consolidated Trend Portfolio.  More specifically, for this Trend Portfolio, we will buy and hold each asset when its price is above its respective 8-month SMA, and sell (i.e. rotate to cash) when its price is below this SMA.  It’s important to note that the SMA look-back time period is largely irrelevant, as a wide range of SMA look-back periods (3-months up to 18-months) produce fairly consistent results.  Further, the SMA itself is a rather elementary way to identify and capture a trend, and there are a variety of more sophisticated techniques that can be utilized.   It is also important to note that our Trend Portfolio analysis will only consider being long or flat each asset class (i.e. owning an asset class or returning to cash) at any given time, as opposed to going both long and short.

Figure 1:  U.S. Stocks & 8-Month Simple Moving Average of U.S. Stocks (1995-2017)

F1 - revised.png

Trend Filter Application to the Traditional Portfolio

To better illustrate the Trend Portfolio approach, we can review a basic example.  The first step is to start with a base portfolio, which can be any asset class or portfolio of choice, such as the U.S. Stock Index, U.S. Treasuries, or a globally diversified portfolio.  For this example, the base portfolio will be the same Traditional Portfolio noted above and detailed in Taking Diversification A Step Further which is a simple, equal-weight portfolio consisting of U.S. stocks, international and emerging market stocks, real estate, long-term treasuries, gold, and commodities.

Figure 2:  Traditional Portfolio and Component Asset Class Performance (1995-2017)


 Figure 3:  Traditional Portfolio and Component Asset Class Equity Curves (1995-2017)


As depicted in Figures 2 and 3 above, between 1995 and 2017, the Traditional Portfolio generated an annual return of roughly 7.7% and experienced annualized volatility of 9.7%, resulting in a Return/Risk ratio of 0.79.  Overall, this diversified portfolio's risk-adjusted return profile is fairly attractive and exceeds that of any of its individual component asset classes; however, by applying an 8-month SMA trend filter, we can explore the potential to further enhance the portfolio's performance.

By applying the 8-month SMA filter to each of the Traditional Portfolio’s individual components, we are able to successfully construct our Trend Portfolio.  More specifically, we will buy and hold each asset class only when its respective price trades above its 8-month SMA, allowing us to systematically deploy capital into upward trending markets and seeking to profit from the positive price trends continuing.  However, if and when the price of each asset class falls below its respective moving average, we will divest our position (i.e. rotate to cash) in order to avoid more prolonged downward trends.  Lastly, as a logistical note, all trades are assumed to be made on a monthly basis, executing at the closing price on the last day of each month.  The performance comparison of the Trend Portfolio against the Traditional Portfolio is depicted in Figures 4 and 5 below.

Figure 4:  Traditional Portfolio v. Trend Portfolio Performance (1995-2017)


 Figure 5:  Traditional Portfolio v. Trend Portfolio Equity Curves (1995-2017)


We can note the Trend Portfolio has a compounded annual return of 7.4% (vs. the Traditional Portfolio’s 7.6%) between 1995 and 2017.  More importantly, however, we should take note of the material improvements achieved in terms of portfolio risk, as annual portfolio volatility and maximum drawdown improve substantially within the Trend Portfolio.  Specifically, annual portfolio volatility declines from 9.7% to 6.2%, while maximum drawdown meaningfully improves from -30.8% to -6.1%.  As a result of the significant reduction in volatility within the Trend Portfolio, the overall Return/Risk Ratio increases from 0.79 to 1.19, marking a substantial improvement over the Traditional Portfolio.  Further, we can see that the Trend Portfolio’s correlation to U.S. Stocks of 0.43 is far lower than the Traditional Portfolio’s 0.75, and therefore, it is likely that the Trend Portfolio would act as a more robust diversification tool over time.  (It’s important to note that this historical exercise assumes 0% interest earned on positions that rotate into cash and does not account for any individual investor's tax implications, which can, as with any active strategy, affect overall returns.)

Trend Portfolio with Modest Leverage

Due to the Trend Portfolio’s ability to materially de-risk and move to cash during times of elevated volatility and market downturns, the overall portfolio benefits most significantly from a risk-reduction standpoint, as evidenced above.  To further explore how we might increase absolute returns, however, we can analyze the effects of applying a moderate amount of leverage to the Trend Portfolio, while being sure to remain conservative from a portfolio risk perspective.   Figures 6 and 7 show the statistics and portfolio return profiles of a moderately levered Trend Portfolio, which assumes the utilization of 25% leverage across any open positions, and the unlevered Traditional Portfolio.

Figure 6:  Traditional Portfolio v. Trend Portfolio (w/25% leverage) Performance (1995-2017)


Figure 7:  Traditional Portfolio v. Trend Portfolio (w/25% leverage) Equity Curves (1995-2017)


The addition of a conservative amount of leverage to the Trend Portfolio increased annual returns from 7.4% to 9.1%, while only moderately increasing the portfolio’s risk metrics.   As a result, the Trend Portfolio w/ 25% leverage has the capability to outperform the Traditional Portfolio from a risk-adjusted and absolute return basis.  Furthermore, it is important to note the correlation to U.S. Stocks of the levered Trend Portfolio remains low at 0.43, maintaining consistent effectiveness as a portfolio diversifier. 

In conclusion, it is RQA's view that Traditional Portfolios can achieve greater returns without materially increasing downside risk through the conservative addition of alternative, uncorrelated strategies such as the Trend Portfolio illustrated above.  In addition, by restructuring a base portfolio with the goal of multi-dimensional diversification in mind, investors can significantly improve the risk profiles of their portfolios, reducing downside risk exposure and alleviating the reliance on any dominant return factors (e.g. economic growth and interest rates), asset classes, and individual strategies.  We would note, however, that the Trend Portfolio illustrated above is a relatively simple example, and there are many more lowly-correlated, high-quality strategies or return streams that can be added to improve expected portfolio results even further.

Disclaimer: These materials have been prepared solely for informational purposes and do not constitute a recommendation to make or dispose of any investment or engage in any particular investment strategy.  These materials include general information and have not been tailored for any specific recipient or recipients.  Information or data shown or used in these materials were obtained from sources believed to be reliable, but accuracy is not guaranteed.  Furthermore, past results are not necessarily indicative of future results. The analyses presented are based on simulated or hypothetical performance that has certain inherent limitations.  Simulated or hypothetical trading programs in general are also subject to the fact that they are designed with the benefit of hindsight.