Do Not Torture the Data, Just…Listen to Them: An Alternative Value at Risk Approach

Abstract

Even though econometric and technological advances have contributed to the vast risk modelling literature, practitioners in most cases use the simplest and most conventional approaches in order to avoid algorithmic complexity and increased costs. In this study we present a new approach which adopts a volatility-based filtering data approach, and uses only the most representative data in order to achieve more accurate VaR estimations than the conventional historical VaR models do. In simple words, the Filtering Historical VaR method, listens to the market’s previously documented behavior and presents more accurate than the conventional approaches VaR estimations.

Keywords: Value-at-Risk; Filtering Models; Accuracy; Emerging Markets; Empirical Risk Estimations

JEL Classification: G01; G15; G32; D81