Diversifying Outside of Stocks: The Most Correlated and Non-Correlated Assets

Research and data analysis by Ashley McKillips.

We all want rewards without risk. While this is impossible in investing, the goal of diversification is to move in that direction. A balanced investment portfolio, in theory, offers increased growth while minimizing the potential for loss. One of the best ways to do this is by examining the correlation between various assets and allocating your portfolio accordingly, incorporating uncorrelated—and even negatively correlated—assets.

A traditional portfolio does this by balancing stocks with bonds. While bonds don't usually offer impressive returns, they're lower risk than stocks and tend to move in opposite directions. This can help minimize loss in the event of a market downturn.

Billionaires, big institutional investors and hedge funds have been using this tactic for centuries, but now just about anybody can start taking advantage of these benefits with new platforms that make alternatives more accessible to retail investors.

One of the first steps to creating a balanced portfolio with alternative assets is figuring out how they correlate with the stock market and each other. In this study, we determined the correlation between 17 popular alternative assets and the stock market, as measured by the S&P 500. We outlined the assets with a strong correlation and moderate correlation as well as those that are uncorrelated and negatively correlated.

• Assets with the strongest correlation to the S&P 500 were robo advisors, real estate, sports cards, farmland, silver, and most popular cryptocurrencies, including Bitcoin and Ethereum. 
• Wine, gold, crude oil, and platinum all were moderately correlated with the S&P 500.
• Bonds and fine art were shown to be negatively correlated to the S&P 500.
• Investment-grade wine and art move in opposite directions.
• Correlation between major cryptos was strong, except Ripple (XRP), which was only moderately correlated to the rest of the coins included in this study.

Correlation between two assets is measured using a mathematical formula that produces what is called the Pearson correlation coefficient, represented by R. This correlation coefficient can range from -1.0 to 1.0.

A positive number represents a positive correlation, meaning the assets tend to move in the same direction—a 1.0 is a perfect positive correlation. A negative number represents a negative correlation, meaning the assets are inversely correlated or tend to move in opposite directions—a -1.0 is a perfect negative correlation. Finally, a correlation coefficient of zero represents no correlation. This means that the two assets move independently and have no relationship with each other.

In reality, a correlation coefficient of exactly 1.0, -1,0, or zero is extremely rare, and a correlation coefficient that is close to zero still means there's little to no correlation. For example, a correlation coefficient of 0.1 is low enough to suggest that any slight correlation found is likely coincidental and not an indicator of any real correlation. For the purpose of this study, here is how we define correlation using the correlation coefficient:

• Strong positive correlation: 0.75 to 1.0
• Moderate positive correlation: 0.50 to 0.75
• Low positive correlation: 0.25 to 0.50
• No correlation: -0.25 to 0.25
• Low negative correlation: -0.25 to -0.50
• Moderate negative correlation: -0.50 to -0.75
• Strong negative correlation: -0.75 to -1.0

The Pearson correlation coefficient in this study is calculated on a monthly basis from 2018 to 2021 with the exception of farmland and art, which are calculated on a quarterly basis. Most of the assets we studied had a positive correlation to the S&P 500 over this time period. However, there were two with a negative correlation.

• Crypto (Bitcoin, Ethereum, Dogecoin, Cardano, and Solana)
• Silver
• Robo advisors
• Sports cards
• Real estate
• Farmland

Most investment assets are correlated with the stock market. Strongly correlated assets—those with a correlation coefficient of 0.75 or greater—tend to follow stock market patterns. While that doesn't necessarily mean they aren't great diversification tools for other reasons, it does mean that they might not offer as much recession protection as other asset classes.

Robo advisors

Of those, robo advisors predictably had one of the highest correlation coefficients at 0.9345. This is predictable, considering robo advisors sell funds and indices that aren't too dissimilar from the S&P 500. Most investors using robo advisors are doing so to invest in the stock market. Data for robo advisor performance was pulled from our own proprietary robo advisor index. Our platforms include the most well-known robo advisors, such as Betterment, Wealthfront, Acorns, SoFi, Personal Capital, Vanguard's Digital Advisor, Schwab's Intelligent Portfolios, and Wells Fargo's Intuitive Investor.

Real estate

Real estate's correlation coefficient was an even higher 0.9636, coming in as the most correlated alternative asset. This also makes sense, as real estate performance data came from Vanguard Real Estate Index Fund Admiral Shares (VGSLX), a publicly-traded index that invests in real estate investment trusts (REITs). This supports the hypothesis that, unlike direct real estate investment, REITs move closely with stock market patterns and thus aren't always the best way to diversify outside of stocks.

Because these REITs are publicly traded on the same markets as equities, they tend to be subject to the same factors that cause stocks to move up and down. However, private and non-traded REITs can be largely uncorrelated to stocks. For example, the eREIT offered by Fundrise, which is a non-traded REIT, is significantly less correlated to the S&P 500 than its publicly-traded counterparts such as Vanguard's REITs.

Most crypto

Crypto was measured by the performance of Bitcoin, Ethereum, Dogecoin, Ripple, Cardano, and Solana. Bitcoin was just behind robo advisors and real estate in terms of stock market correlation, with a very high correlation coefficient of 0.9175. Ethereum was rather high as well, while smaller altcoins like Dogecoin (0.8338)  and Cardano (0.8849) had a slightly lower correlation coefficient. 

Ripple (XRP) and Solana were notably less correlated with stocks than the other cryptocurrencies we looked at, with an almost moderate correlation coefficient of 0.7738 and 0.7709 respectively. It's worth noting that Ripple was delisted from many centralized exchanges making it a bit more difficult to trade than other cryptocurrencies included in this study.

At the time of this study, all of these cryptocurrencies are in the top 12 crypto coins by market cap and are generally available on centralized exchanges. Emerging coins with much smaller market caps and those that are more difficult to buy and sell on centralized exchanges may not be as highly correlated with stocks.

Silver, sports cards, and farmland

Finally, silver, sports cards, and farmland all have positive correlation coefficients of over 0.75. This may seem surprising for alternative assets—particularly farmland, which is often seen as a safe-haven asset during times of economic crisis. While it does hold a fairly high correlation to stocks, farmland has also historically been far less volatile than the stock market.

• Gold
• Wine
• Crude Oil
• Platinum

Assets with a moderate correlation to the stock market—a correlation coefficient between 0.50 and 0.75—still show a notable degree of correlation. While they might not move in tandem with stocks quite as closely as the assets above, they still tend to follow similar patterns.

Gold and platinum

Precious metals have long been used as a hedge against inflation and a store of value. Based on the past three years, gold and platinum seem to be viable options. 

Gold's correlation with stocks is complicated. While there have been times when the price of gold has held strong during bear markets, the asset's price often declines sharply at the start of a stock market decline. This happened at the beginning of the 2020 coronavirus pandemic, during an equities downturn at the end of 2018, during the financial crisis of 2008, and during the dot-com bubble in 2001. However, according to chief gold strategist for the SPDR ETF George Milling-Stanley in an interview with ETF.com, gold bounced back within a week each time (much more quickly than equities), largely due to it being cashed in by investors who bought equities on margin to cover their losses. "…rather than selling their equities in order to meet the calls for additional margin," says Milling-Stanley, "they sell something that has held its value, which at that point is gold. …as soon as equities stabilize, usually at lower levels, they will tend to buy their gold back."

Platinum, on the other hand, is an under-discussed asset that has become increasingly interesting for investors in recent years. Used mainly in cars and jewelry, platinum distinguishes itself from gold due to the fact that it is both more useful and rare than gold. Gold and platinum also have a low positive correlation to each other of 0.5432, showing that while the two assets move similarly with regard to the S&P 500, they don’t move as closely together.

Crude oil

Crude oil and the S&P 500 have a correlation coefficient of 0.6406, making oil a moderately correlated asset as well. In fact, commodities in general tend to be less correlated with the stock market and great diversification tools for a stock-heavy portfolio. That said, they don't generally trend upward over the long term the way stocks do, so they're often used for short-term growth or long-term wealth preservation.

Assets with a negative, or inverse, correlation to the S&P 500 move in opposite directions from the stock market. While uncorrelated assets can help preserve wealth, the point of inversely correlated assets is to help your portfolio continue to grow when stocks are down.

Bonds

Bonds have long been a useful tool for balancing a portfolio because they're considered low-risk and can help reduce losses. Our study found that with an inverse correlation to stocks (at a correlation coefficient of -0.3380, bonds can not only help reduce the risk of loss but may actually help recoup some losses in a market downturn. 

This effect is magnified when investors who are wary of an impending recession run to Treasury bonds as a safe haven, therefore increasing bond prices. As with most inversely correlated assets, however, bonds don't offer impressive returns, especially when the market is doing well. So while holding some percentage of assets in bonds can help pad a stock-heavy portfolio, shifting too many assets into bonds can diminish long-term returns.

Art

Art came out as the most negatively correlated asset, with a correlation coefficient of -0.6350. Our study used a combination of Artprice Global Indices that cover investment-grade fine art from various mediums and time periods. So while blue-chip art may offer increased returns when stocks are falling, this doesn't mean that all fine art will.

There were also interesting findings in regards to the correlation between various alternatives.

Wine and art move in completely opposite directions, which is surprising considering both derive their value primarily from the same quality: reputation. What's more, one might expect the market for investment-grade art and investment-grade fine wine to have a lot of overlap.

For investors who primarily invest in stocks via robo advisors, it's worth noting the difference in correlation between robo advisors and alternative assets versus the S&P 500 and alternative assets. For example, crude oil is slightly less correlated with robo advisors than it is with the S&P 500. Also, robo advisors and art hold a moderate positive correlation even though art was shown to be negatively correlated with the S&P 500.

Finally, the correlation across various cryptos might be important to crypto investors looking to diversify their coin holdings. Bitcoin and Ethereum, perhaps unsurprisingly as the two biggest cryptos, are almost perfectly correlated. Bitcoin is also strongly correlated with Cardano, Dogecoin, and Ripple, with Ripple being the least correlated of the three. The only coin that didn't hold as strong a correlation to Bitcoin was Solana, although at a correlation coefficient of 0.766, the two are still positively correlated.

Our study pulled historical returns on 17 different popular alternative assets using price data pulled from Yahoo! Finance and various indices that track the performance of a specific asset. Some of these indices are created by external sources (such as Artprice for art and Live-ex 100 for wine) while others were created using proprietary MoneyMade data (such as the robo advisor index). 

It's worth noting that some of these indices might not provide an accurate indication of the asset class's overall correlation to the stock market. For example, for real estate, we looked at the correlation between the S&P 500 and Vanguard real estate index fund VGSLX. While this may help REIT investors determine correlation information, it will be less accurate for those directly invested in real estate.

As stated in the beginning of the article, assets were correlated on a monthly basis from October 2018 to December 2021 using Pearson correlation. The exceptions are art and farmland, which were correlated on a quarterly basis from Q4 2018 to Q4 2021. (Quarterly correlation was calculated between all assets. The difference between quarterly and monthly correlation was negligible, therefore it is reasonable to compare the two categories of correlation coefficients.) For assets that have daily releases, such as the cryptocurrencies and the S&P, data was translated to monthly by pulling the price on the 1st of every month. Similarly, for the monthly and daily releases, the quarterly price was denoted as the price on the 1st day of each quarter.

We chose to calculate correlation over this three-year period primarily to maximize the number of comparable assets, but there was also another reason. Over longer time frames the asset correlation is far more diverse, but the long time frame isn’t always relevant when you're making near-term investment decisions. We wanted to consider a smaller time frame in order to consider how various alternative portfolios could move in the short term.

1. Artprice. (2021). Artprice Global Index [An index of the top blue-chip artworks].
2. Liv-ex. (2021). Liv-ex 100 [An index of the top 100 investment-grade wines].
3. MoneyMade. (2021). MoneyMade robo advisor index [A proprietary index of robo advisors].
4. PWCC Marketplace. (2021). PWCC 500 [An index of the top 500 blue-chip trading cards].
5. U.S. Department of the Treasury. (n.d.). Treasury.Gov. Retrieved December 1, 2021, from https://Treasury.gov
6. Yahoo! Finance. (n.d.). Yahoo! Finance. Retrieved December 1, 2021, from https://finance.yahoo.com/