CMStatistics 2018: Start Registration
View Submission - CFE
Title: Market risk of cryptocurrencies Authors:  Annalisa Molino - University of Rome Tor Vergata (Italy) [presenting]
Abstract: Due to the increasing popularity of cryptocurrencies, understanding the risk features of this market is important for both investors and regulators. An analysis of the risk of holding hypothetical portfolios made of cryptocurrencies is carried out by using Monte Carlo simulations. The key decision for a Monte Carlo simulation is the choice of the probability distribution that better approximates the return distribution. The latter has to be modeled with a particular focus on the tails, from which the extreme quantiles are extracted. The univariate distribution of returns is modeled by a combination of extreme value theory and kernel estimation, and copulas are used to model their dependence. The findings point to the fact that cryptocurrencies returns have peculiar characteristics that make them a very risky investment: the portfolios experience indeed extraordinary large gains and losses. Due to the zero or low correlations with fiat currencies, allocating a percentage of investment in cryptocurrencies reduces the risk of holding a hypothetical portfolio of fiat currencies for one month. Besides the empirical study, the contribution is twofold. First, it suggests a semiparametric way to model the distribution of cryptocurrencies. Second, it suggests a flexible way to estimate the Value-at-Risk and the Expected Shortfall of cryptocurrencies.