Updated data: Dec 2019
After experimenting one year with my CryptoCurious portfolios, I learned that if you want to add major diversification effects to your portfolio of cryptocurrencies, it won’t come from adding more cryptocurrencies.
Nevertheless having a well diversified portfolio of different cryptocurrencies is important as it will help you diminish the overall risk of your portfolio. In The Intelligent Investor, Benjamin Graham argued that a portfolio of 10 to 30 stocks provides adequate diversification.
Indeed by the time the portfolio contains close to 20 equal-sized and well-diversified issues, the total risk (standard deviation of returns) of the portfolio is reduced by 70 percent. Further increase in the number of holdings does not produce any significant further risk reduction (Figure 1).
S&P 500 and cryptocurrencies are de-correlated
This is all good when correlations between different cryptocurrencies vary significantly but it has not been the case unfortunately over the past 4 months. Most cryptocurrencies have a high degree of correlation, which means that they all move basically in lockstep with each other (figure 2).
Why does correlation matter in portfolio construction ?
Choosing assets with low correlation with each other can help reduce the risk of a portfolio and increase risk adjusted returns.
If we have a look at the correlations of the CryptoCurious portfolios composed of portfolios ranging from 4 to 20 crypto assets, they are highly correlated (figure 3). For more information about their composition, visit The CryptoCurious.
What I did then is to check the correlation of US equities (figure 4).
The S&P 500 has been correlated 98% over the last 4 months with the Nasdaq as major tech companies belong to both indices. We are in a world and at a time where global assets are highly correlated.
If you combine both the cryptocurrency market with the US equity market, you get near to 0 correlation which is great news for diversification.
I found a study dating from June 2017 which also looked at the correlations. They found the same results and also studied the correlation of Bitcoin with other asset classes such as real estate, commodities, forex, bonds etc …. All of these asset classes seem to be de-correlated with Bitcoin (figure 5).
For diversification purposes, low correlation is preferable to high negative correlation. Indeed when you have two perfectly negatively correlated assets, if one asset gains 5% the other should lose 5%, which leads to returns cancelling each other. When correlation is low, you get un-correlated returns on both side, the effects of an asset on the other should be as low as possible.
One problem is that during a major crisis, most markets whether equity, bond, commodities or real estate will move in the same direction, that is, fall in price; and there is not much you can do except hold on tight. A portfolio that has been properly constructed should eventually recover.
Creating a hybrid portfolio
In order to benefit from the low correlation, I created a hybrid portfolio combined of 50% the cryptocurrency market and 50% of the US equity market.
How did I decide to represent both markets ? I used the S&P 500 for the US equity market, since it is a well diversified stock market index based on the market capitalizations of 500 US large companies. As for the cryptocurrency market, I created my own diversified index composed of the Top 20 cryptocurrencies ranked by market capitalization, equally weighted (visit The CryptoCurious).
Figure 6 displays its price index evolution since I started the study. Half of the performance of this hybrid portfolio comes from the performance of the S&P 500 and the other half from the Top 20 crypto Index. It will never be able to beat both markets but overall risk drops and it diversifies the performance (Figure 7). The maximum drawdown is significantly lower, it goes from -53% with a 100% crypto portfolio to -30% with the hybrid version but it is nevertheless double the maximum drawdown on the S&P 500 (-15.7%).
To study the power of de-correlated assets, we must rebalance the portfolios on a regular basis. Portfolio rebalancing helps reduce downside investment risk. A study on the benefits of algorithmic rebalancing of cryptocurrencies has shown that rebalancing is profitable when rebalancing periods are no longer than 60 days, profit thereafter seems to fall off. I have chosen to rebalance monthly.
Figure 8 displays the price evolution of the rebalanced hybrid portfolio. The results are remarkable: the monthly rebalanced portfolio captured higher returns than the hybrid holding portfolio (31% vs. 22% figure 9), and did so with less risk (39% vs. 44% annualized).
Given that cryptocurrencies are uncorrelated with traditional markets, rebalancing managed to beat both of the performance of the S&P 500 by 13.5% and the Top 20 Crypto Index by 2.4%.
If you rebalanced your funds regularly each month between the S&P500 and the Top 20 basket of crypto assets, you would have outperformed the long-only passive portfolio by about 9%.
To benefit from the un-correlated assets rebalancing is necessary. On periods of low volatility for both markets, the rebalanced portfolio did better, it probably can be attributed to the compounding effects of rebalancing with near to zero correlation.
1- Having a well diversified portfolio of different cryptocurrencies is important as it will help you diminish the overall risk of your portfolio.
2 - Choosing assets with low correlation will help reduce the risk of a portfolio and increase risk adjusted returns.
3 - If you combine both the cryptocurrency market with the US equity market, you get near to 0 correlation, great for diversification.
4 - It is possible to diminish risk and generate extra return by rebalancing regularly your Cryptocurrencies / US Equities portfolio.
5 - Correlations evolve with time, so these results might hold for now but as the market matures they might not anymore.
Nothing in this article is intended to serve as financial advice. Do your own research. Past performance is not indicative of future performance.
This is ongoing work, any suggestions are welcomed.
Thank you for reading.