VaR Estimation Using Extreme Value Mixture Models for Cryptocurrencies

Abstract

Cryptocurrencies have obtained a crucial position in the international financial landscape. The cryptocurrency market has been perceived as a highly volatile market since the inception of Bitcoin. This study investigates the relevant performance of extreme value models (EVM) in estimating the Value-at-Risk (VaR) of Bitcoin and Ethereum returns. The extreme value mixture models, GPD-Normal-GPD (GNG) and GPD-KDE-GPD models are fitted to the returns of Bitcoin and Ethereum and the Kupiec likelihood backtesting procedure is performed on the VaR estimates to assess the fits. Both models’ results showed that the fits were a much more decent representation of the observed data when compared to the Normal distribution. The backtesting results showed that the GPD-KDE-GPD model’s fit was superior to that of the GPD-Normal-GPD for both sets of returns at all VaR risk levels except at the 99% level. The results of this study may assist with understanding the dynamics and risks associated with cryptocurrencies and can serve as a beneficial tool for decision-making and risk management to investors, traders, financial institutions and many other participants in the cryptocurrency ecosystem.

Analyzing Portfolio Optimization in Cryptocurrency Markets: A Comparative Study of Short-Term Investment Strategies Using High-Frequency Data

“…Cryptocurrencies are reshaping the financial landscape, poised to revolutionize the global financial system (Rajharia, 2023). These digital assets garner attention from policymakers, investors, and academics as a burgeoning alternative asset class (Jana, 2024), exhibiting unprecedented growth and establishing a significant international foothold (Subramoney et al, 2023). Consequently, interest in their impact on traditional financial markets is burgeoning (Krishna, 2023).…”