Multiple-threshold asymmetric volatility models for financial time series

This article is concerned with asymmetric volatility models for financial time series. A generalization of standard single-threshold volatility model is discussed via multiple-threshold in which we specialize to two -threshold case for ease of presentation. An empirical illustration is made by analyzing S&P500 data from NYSE (New York Stock Exchange). For comparison measures between competing models, parametric bootstrap method is used to generate forecast distributions from which summary statistics of CP (Coverage Probability) and PE (Prediction Error) are obtained. It is demonstrated that our suggestion is useful in the field of asymmetric volatility analysis.