Comparing Methods to Forecasting Var: Fhs Versus Evt Approaches

Abstract

Financial institutions have for many years sought measures, which identify and approach market risks in portfolios of financial instruments. Forecasting VaR has attracted a great deal of attention in the financial econometrics literature, the so-called filtered historical simulation (FHS) and Extreme Value Theory approaches. The paper provides a methodological contribution to the one-step-ahead VaR forecasting through some FHS and EVT approaches, which combine some conditional models for variance. Accounting asymmetrically distributed portfolio returns within the conditional variance process, accurate 99 per cent VaR forecasts are calculated for one-day-ahead horizon under Basel accords. After calculating VaR, we evaluate these methods for presenting the most appropriate methodology.