This study proposes a comparative analysis of the use of ten volatility models to calculate the Value-at-Risk (VaR) for Ibovespa considering the presence of long memory in time series for daily returns. Data from January 02, 2000 to January 02, 2008 were used. The results showed that the long memory models, in especial the FICARCH (1,d1) model, are better to calculate the Value-at-Risk if compared to traditional models, such as the RiskMetrics model.
Value-at-Risk; Stock market; Conditional volatility; Ibovespa
