Uncorrelated Assets And Strategies – Benefits And Advantages (Non Correlated Examples And Backtests)

Last Updated on June 19, 2022 by Quantified Trading

Uncorrelated or non-correlated assets and strategies are a traders’ goldmine. Why? Because it reduces risk and (might) increase returns. However, constructing a basket of stocks, assets, or strategies that are uncorrelated or non-correlated is probably the most difficult task in trading and investing. In this article, we show you examples and backtests of the power of making a diversified and uncorrelated portfolio of both different assets and strategies.

The benefits and advantages of having uncorrelated assets and strategies are both smaller drawdowns and higher returns. We provide you with examples and backtests to show you why.

You kill two birds with one stone. If you include assets that are uncorrelated with stocks, you can increase total returns (or at least get the same return) even if the included assets have lower total returns. The reason why is due to correlation – or lack thereof.

If you have a portfolio of trading strategies that are only mean-revertive, for example, you obtain very little by adding similar strategies. Quite the contrary, you might end up with lower returns and bigger drawdowns. You don’t want to put all your eggs in one basket, whether it be assets or similar types of strategies. Why? Because you want to have the smallest drawdown you can get and to have a portfolio that can withstand shocks and tail risks. You want an antifragile portfolio. We have touched upon this in a previous article:

Let’s start by explaining the basics of correlation:

What is correlation in trading and investing?

Correlation is probably the most important factor if you want to have a robust, diversified, and antifragile portfolio. In case you are not sure of what correlation is, we’ll give you a short primer:

Correlation is a mathematical term used to describe the covariance of two time series data.

For example, the weight of people is dependent on height. In a spreadsheet, you can put the height of a group of people in column A, and in column B you put their weight. If you have a sample with many observations, there will be a clear relationship between height and weight. This is correlation. Tall people normally weigh more than small people (we generalize).

Likewise, most tasks yield a result that corresponds to the effort you put in.

Hence, correlation is a statistical measure that shows how two variables are related, but it doesn’t give any causation. Two variables can correlate purely out of randomness.

Mathematically, a perfect correlation has a value of 1, and the opposite, when two variables move completely different ways, the correlation coefficient is -1. A zero correlation coefficient means that the asset prices are uncorrelated.

Nevertheless, a negative correlation doesn’t necessarily mean when Asset A provides a negative return, Asset B provides a positive return—even if it’s a perfect negative correlation (-1).

We have previously written more extensively about correlation in trading.

What are uncorrelated assets?

Now that you know what correlation is, you can probably guess that uncorrelated assets are assets that don’t move in tandem with each other.

If you are long stocks or real estate, like most investors are, you want something that can mitigate the risk by providing a “safe haven” if the stocks and real estate were to drop in value. That is, of course, difficult, because during a crisis most assets tend to correlate.

Mark Spitznagel, a disciple of Nassim Nicholas Taleb, has written a book about risk-mitigating and what role defensive assets play in a portfolio. He explains very well how to seek shelter until the storm has passed and even goes on to explain why we get better results simply because the risk is lowered. Spitznagel does a great job of explaining why including assets that have low returns can increase the return of the portfolio if they are uncorrelated to the other asset(s). This is called risk mitigation.

What will combining uncorrelated assets do?

You want to mitigate risk and reduce the drawdown in your portfolio. One of the reasons why is that most traders and investors are susceptible to behavioral mistakes if they suffer big drawdowns.

A good example of the benefits of having uncorrelated assets is Ray Dalio‘s All-Weather Portfolio:

The All Weather portfolio by Ray Dalio

A perfect example of how to build and construct a diversified and uncorrelated portfolio is Ray Dalio’s All-Weather Portfolio.

The aim of Ray Dalio and his team at his hedge fund Bridgewater was to create a portfolio that uses diversification as a tool to smooth uncorrelated equity returns and lower drawdowns. Ray Dalio’s Bridgewater Associates composed the portfolio like this:

  • 55% bonds
  • 30% US stocks (US stocks are about 50% of world market capitalization)
  • 15% hard assets and commodities

That portfolio contains a lot of bonds!

We made a backtest during Covid-19 based on the following ETFs:

  • 40% TLT (long-term Treasuries)
  • 30% SPY (US stocks, S&P 500)
  • 15% IEI (intermediate term U.S. Bonds)
  • 7.5% GLD (gold)
  • 7.5% DBC (commodities, commodity index tracking fund)

The chart below shows the performance in 2020:

Stocks dropped 30% (pink line) while the diversified portfolio only fell 6% (blue line). When the dust settled, the All-Weather portfolio continued compounding from a higher level.

Meb Faber’s Three-Way Momentum/Trend-Following Model Strategy

A few weeks back we wrote an article about Meb Faber’s Three-Way Strategy. Although Faber’s strategy is primarily about momentum and trend-following, it serves as an example of how you can successfully add different asset classes to create higher returns – both nominal and risk-adjusted.

Mean reversion and trend following: uncorrelated strategies

Let’s go on a look at how you can use uncorrelated strategies to your advantage:

The two main types of trading and investing are mean reversion and trend-following. They have very different characteristics:

Mean reversion has a high win rate, many trades, on average small gains, rare big losers, and is almost always negatively skewed. Fat tails are problematic – mean reversion suffers from tail-risk.

Opposite, trend-following has a low win rate, with many small losers, but a few very big winners. Tail risk is very low but normally to your advantage.

Because the two types of strategies have opposing characteristics you should combine them. They are normally uncorrelated asset classes – exactly what you want in a portfolio!

Trend-following adds non correlated equity returns

The start of 2022 has been rough. Just look at this:

  • S&P 500  -18.40%
  • Nasdaq -28.10%
  • Long-term Treasuries (TLT) -17.90%
  • Bitcoin -38.00%
  • Gold -2.20%

But what about trend following? Here are a few select funds:

  • Lynx +28.78%
  • Eurekahedge Trend +18.75%
  • Barclay CTA Index +8.10%
  • Paul Mulvaney Trend +87.80%

In 2022, all trend-following funds are performing very well. These funds can be hard to be invested in because returns often are “lump-sum” (meaning a few years make much of the difference). Despite this, trend following works.

Backtest of uncorrelated asset classes

Let’s make a backtest and show how a 100% stock position can benefit from adding a trend-following fund.

Why do we add a trend-following fund?

We add a trend-following fund because they employ strategies that can be profitable in both rising and falling markets, just as we described above. Moreover, trend-following strategies have a nonlinear payoff to their underlying market universe. Trend following often performs very well when stocks are having a tough time.

This is exactly what you are looking for in terms of diversification! As long as you have large moves either up or down you perform well with any trend-following strategies. 

To give you a better understanding of the annual returns of trend-following and commodity hedge funds, we have assembled this table of annual returns for you:

Year Paul Mulvaney Trend
Eurekahedge Barclay CTA Index Lynx S&P 500
2000 24.51 23.2 7.86 13.97 -10.14
2001 6.69 16.56 0.84 15.33 -13.04
2002 19.37 23.07 12.36 18.68 -23.37
2003 29.28 25.7 8.69 31.58 26.38
2004 -0.1 8.85 3.3 12.61 8.99
2005 32.34 6.24 1.71 8.31 3
2006 21.94 10.16 3.54 9 13.62
2007 -23.14 20.64 7.64 15.3 3.53
2008 108.87 29.8 14.09 38.24 -38.49
2009 -5.9 5.4 -0.1 -9.43 23.45
2010 34.9 10.91 7.05 19.18 12.78
2011 -5.26 1 -3.09 -2.24 0
2012 -33.72 -1.89 -1.7 -6.73 13.41
2013 43.12 1.19 -1.42 11.11 29.6
2014 67.36 12.91 7.61 27.03 11.39
2015 -0.77 -1.82 -1.5 -8.73 -0.73
2016 -1.82 -0.76 -1.23 -3.29 9.54
2017 1.57 0.13 0.7 -4.05 19.42
2018 -4.33 -6.23 -3.17 0.35 -6.24
2019 -21.28 6.74 5.17 18.36 28.88
2020 18.53 11.29 5.43 7.78 16.26
2021 32.93 5.92 5.04 1.29 26.89
2022 87.8 18.75 8.1 28.78 -18.4

 

The second column is the hedge fund of Paul Mulvaney who has one of the best track records there is in the commodity business. The third column is an index made by Eurekahedge that tracks the performance of trend-following funds. The fourth column is Barclay’s CTA Index, and the fifth column is the Swedish hedge fund Lynx (a systematic trend following fund).

What is the correlation like? All funds have a correlation between -0.3 to -0.5 to S&P 500. Perfect!

What is most interesting with this table is to look at the performance in the alternative funds when the S&P 500 performs poorly, especially in 2000-2003 and 2008. For example, look at Lynx in 2008 when it was up 38.24% while S&P 500 was down 38%! This pattern is repeated so far in 2022 (which has been a poor year for stocks thus far).

A backtest of uncorrelated assets

Let’s make a simple backtest as an example to show you the power of mixing different types of strategies.

Below we show how adding uncorrelated assets reduce volatility in your stock portfolio and also increases the total returns.

Let’s assume you buy the S&P 500 in January 2000. However, you also add the Swedish trend-following fund called Lynx (look at its performance in the table above). We select Lynx because we indirectly have units in the fund and we have Lynx’s data available at our disposal. Keep in mind that Lynx has performed well, and we might be liable to survivorship bias in our selection.

How much should you allocate to each fund? In this example, we allocate 75% to S&P 500 and 25%. We start our backtest in 2000 and rebalanced at the start of every year thereafter (this is the USD returns of Lynx). The accumulated returns would have been like this:

The red line is S&P 500 and the blue line is our diversified portfolio. CAGR is 7.25% for the combined portfolio while it’s only 5.6% for S&P 500 (dividend reinvested is not included). This is a huge improvement compared to only owning stocks. Drawdowns are also substantially lower.

We might argue we are lucky with our pick with Lynx. What if we only managed the return of the Barclays CTA Index? That index had an annual return of only 3.7% over the period while Lynx had 9.9%.

But even if we allocate 25% to Barclays CTA Index and rebalance every year, we would get the same return as holding the S&P500 but with lower drawdowns:

We get the same return with less volatility! The bottoms are higher and your start compounding at a higher level when the dust settles. This is why you want to mitigate volatility!

Many argue volatility doesn’t matter, but we disagree. Due to behavioral mistakes in trading, you risk selling at the worst time the more volatility you have in your portfolio. All empirical evidence points toward that.

Adding similar and correlated trading strategies

Let’s assume you have an existing portfolio of trading strategies with this equity curve:

The average gain per trade is 0.3% over more than 1 500 trades. This is an overnight trading strategy.

Then you come across a promising mean-reverting strategy that has the following equity curve (and you consider adding it to the other strategies above):

The average gain is 0.33% and the win rate is 65%. It’s a pretty good strategy, we would say.

What happens when you add the extra strategy together with the other strategies? Something surprising happens – the trading and performance metrics deteriorate:

When you add this seemingly good strategy, the overall performance drops! The overall returns drop and drawdown increases.

Why?

This happens because you add a similar strategy that is highly correlated with your existing strategies. Your trades overlap, and you mainly add the losers from the new strategy.

This is why you can never add trading strategies without looking at your strategies as a portfolio of strategies.

Uncorrelated Assets And Strategies – What Are The Benefits And Advantages

This article has hopefully shown you the benefits and advantages of uncorrelated assets and strategies. Total returns of an asset are not necessarily the main criteria to look for when you want to add assets or strategies to your portfolio. Even assets or strategies that have lower total returns can be much more beneficial than including assets with higher returns. The simple reason is due to non-correlation. This is why the composition of assets and strategies is not an easy task.

Similar Posts