Demystifying Cryptocurrency Investments Part I: Modern Portfolio Theory
To help investors navigate the crypto landscape, Dcrypt will be posting a three-part series on investing in the crypto economy. Each segment will shed light on a different challenge involved in making sound crypto investments.
Modern Portfolio Theory & The Cryptocurrency Market
As investors, our primary goal is to generate alpha in a risk averse way, yet despite this universality, our individual thresholds for risk vary greatly. Harry Markowitz confronted this in the 1950s with his Modern Portfolio Theory (MPT). MPT begins by asking how the assets in one’s portfolio move in relation to one another, a variable we quantify mathematically as the correlation coefficient. With these coefficients, we explore the space of possible portfolios, each with their respective expected return and volatility. Lastly, we construct the “efficient frontier”, which defines our greatest expected return for each level of volatility.
Champions of MPT include Ray Dalio, whose all-weather fund utilizes uncorrelated assets to consistently outperform the mean, and Blackrock, where asset managers constantly rebalance portfolios to reduce risk and maintain a desired level of return.
The overarching result of MPT is to drive efficiency within the capital markets, but a new type of market has recently developed that currently lacks this efficiency, the cryptocurrency market. The advent of blockchain technology and ICO funding models have driven an unprecedented amount of capital to the space, upwards of $200 billion. However, as money has flowed to the space, risk analysis tools have not followed suit. Consequently, we’ve yet to apply the radically successful tools of MPT to understand the interplay between risk and reward both in the market and our portfolios.
The Crypto Market Correlations in 2016
Using Poloniex pricing data, we first investigate the correlation coefficients between the major coins in the latter half of 2016. The correlation matrix displays a wide dispersion of both positively and negatively correlated assets, thereby indicating a portfolio of these coins could reduce investor risk compared to holding one asset alone. An investor during this time period would be wise to hold a portfolio that would protect against the downside, while also keeping exposure to the dramatic upswings we oftentimes see in the crypto market. For example, one may want to invest in both bitcoin and Ethereum Classic, as it appears that when BTC goes up, ETC may not necessarily follow.
The Crypto Market Correlations in 2017
Moving to Poloniex data available in 2017, the correlation matrix tells a much different story. The sea of red shows how correlated the market has become, so if the tide rises, all of the ships rise with it. As we will see, high correlations make constructing a portfolio on the efficient frontier difficult, but not impossible.
The Efficient Frontier Methodology
Now that we’ve explored the relationship between assets in the crypto market, we can leverage the efficient frontier to construct the optimal portfolio given our desired level of risk.
Using Python, we randomize the weights assigned to different assets within a basket. With these weightings and their asset’s corresponding historic returns, we calculate the expected return, volatility and sharpe ratio of the portfolio. We repeat this process until we have explored the space sufficiently enough to draw an efficient frontier curve.
After calculating our efficient frontier, we select the portfolio with the maximum Sharpe ratio, which is a portfolio’s return divided by its volatility. It’s important to note that the max Sharpe ratio is for the more aggressive investors who want the highest return based on risk ratio. The low volatility portfolio can be for investors trying to stay within their risk tolerance or as a hedge against being margin long volatility.
Note, for our purposes we negated the risk-free rate in determining our Sharpe ratio.
2016 Efficient Frontier Portfolio
- Max Sharpe Portfolio: Green Diamond, Sharpe: 4.19, Volatility: 0.82
- Low Volatility Portfolio: Blue Diamond, Sharpe: 3.85, Volatility: 0.78
This portfolio consists of a set of assets that could have been purchased as of July 2016. By analyzing our efficient frontier, we see that the portfolio with the highest Sharpe ratio (4.1) produced returns of 340% annually.
2017 Efficient Frontier Portfolio
- Max Sharpe Portfolio: Green Diamond, Sharpe: 5.19, Volatility: 1.03
- Low Volatility Portfolio: Blue Diamond, Sharpe: 4.47, Volatility: 0.93
We now proceed to a portfolio consisting of assets purchasable as of January 2017. We’ve expanded the set of assets in the hopes to diversify away the inherent risk of increased market correlation.
However, as we investigate the maximum Sharpe portfolio on the efficient frontier (green diamond), we see our volatility has actually increased (1.03), despite the 540% return. Our Sharpe ratio has increased as a result of the expected return of adding new assets, despite the risk these assets have introduced to our optimal portfolio.
Investment Analysis
As we’ve seen, assets within the cryptocurrency market have become increasingly correlated while the market cap has continued to increase. Our attempts to confront this increased correlation through diversification resulted in a portfolio with higher expected returns at a cost of higher risk. Consequently, we have encountered both the power and limitations of MPT.
The risk, return, and correlation coefficients are built using historical values, so the weightings selected by the efficient frontier are only as representative of future returns as they’ve been of past returns. Given the sudden, dramatic fluctuations in the crypto market, this “overfitting” of historical data is a potent argument against its application. Additionally, MPT also fails to account for transaction costs, which could eat into the expected returns of a portfolio over time.
As a result, approaching the crypto market from a pure MPT point of view doesn't paint the whole picture. Our investment strategies must take other aspects into consideration. The next segment in Dcrypt’s three-part series on crypto investment strategies will explore the value of management and analyze various ICO token models.
