Simulação Histórica Filtrada: Incorporação da Volatilidade ao Modelo Histórico de Cálculo de Risco para Ativos Não-Lineares

Historical models are being quite used in the value at risk (VaR) estimation due to the fact that many returns of financial assets cannot be described by a theoretical distribution. In these models, each observation of the past can be a possible scenery and for each scenery there is a price for the asset, what represents an implicit assumption that the returns are identically and independently distributed. The violation of this assumption is the main criticism to these models, evidenced by the existence of volatility clusters in the financial series that can cause an inconsistency in the value at risk estimates. This work applies a solution for this problem incorporating the volatility to the historical model for the value at risk estimate of stock options in the Brazilian market. The obtained results show that, during the studied period, the methodology presents good performance for a VaR estimation of 99% probability. For the 98% and 95% probabilities, an overestimation of the VaR is verified.