Cryptocurrency Portfolio Diversification Using Network Community Detection

Abstract

As of the end of 2013 till now we are witnessing huge volatility and risk in the cryptocurrency market compared to flat currency or stock market. Thus, in this market the portfolio diversification is of big importance in order to reduce volatility and keep the optimal return for the investors. A usual approach for portfolio construction is to keep a balance between returns and volatility, based on their interdependence and individual returns. One way of diversification is employing clustering or community detection algorithms to select a more diverse set of assets. We study the utilization of the Louvain algorithm and affinity propagation for community detection, based on correlation and mutual information between cryptocurrencies, for potential application in portfolio diversification.