Title: Value-at-risk prediction in R with the GAS package
Authors: Leopoldo Catania - Aarhus BBS (Denmark) [presenting]
David Ardia - HEC Montréal (Canada)
Kris Boudt - Vrije Universiteit Brussel and VU Amsterdam (Belgium)
Abstract: Generalized Autoregressive Score (GAS) models have been recently proposed as valuable tools for signal extraction and prediction of time series processes with time-varying parameters. For financial risk managers, GAS models are useful as they take the non-normal shape of the conditional distribution into account in the specification of the volatility process. Moreover, they lead to a completely specified conditional distribution and thus to a straightforward calculation of the one-step ahead predictive Value-at-Risk (VaR). It is shown how the novel GAS package for R can be used for Value-at-Risk (VaR) prediction and provides illustration using the series of log-returns of the Dow Jones Industrial Average constituents. Details and code snippets for prediction, comparison and backtesting with GAS models are presented.