Forecasting volatility using range data: analysis for emerging equity markets in Latin America

The article suggests a simple but effective approach for estimating value-at-risk thresholds using range data, working with the filtered historical simulation. For this purpose, we consider asymmetric heterogeneous Autoregressive Moving Average (ARMA) model for log-range, which captures the leverage effects and the effects from daily, weekly and monthly horizons. The empirical analysis on stock market indices on the US, Mexico, Brazil and Argentina shows that 1% and 5% Value at Risk (VaR) thresholds based on one-step-ahead forecasts of log-range are satisfactory for the period includes the global financial crisis.