Forecasting Value-at-Risk and Expected Shortfall of Cryptocurrencies using Combinations based on Jump-Robust and Regime-Switching Models

Abstract

Several procedures to estimate daily risk measures in cryptocurrency markets have been recently proposed in the literature. Among them, procedures taking into account the presence of extreme observations, as well as procedures that include more than a single regime, have performed substantially better than standard methods in terms of volatility and Value-at-Risk forecasting. Three of those procedures are revisited in this paper, and their Value-at-Risk forecasting performance is evaluated using recent cryptocurrency data that includes periods of turbulence. Those procedures are also extended to estimate the Expected Shortfall, and a comprehensive backtesting exercise based on both calibration tests and scoring functions is performed. In order to mitigate the influence of model misspecification and enhance the forecasting performance obtained by individual models, we evaluate the use of forecast combinations strategies. In our empirical application, procedures that are robust to outliers performed slightly better than regime-switching models. We found some evidence that combining strategies can improve the forecasting of Value-at-Risk and Expected Shortfall, particularly for the 1% risk levels, making them an interesting alternative to be used by practitioners.