Estimation of volatility of CSI 300 index based on regime switching PTTGARCH model

Considering the volatility of stock market presenting regime switching as well as nonlinear and asymmetric characteristics which can be described using the power-transformed and threshold (PTT) GARCH model, the paper proposes regime switching PTTGARCH (RS-PTTGARCH) model. The daily log return series of CSI 300 Index are taken as the study example and regimes of volatility of the series are classified into three states: the falling, the rising and the consolidation. The in-sample period is selected from July 1 2013 to Dec. 17 2015 and the out-of-sample period from Dec. 18 2015 to Jan. 8 2016. Estimation and forecasting of the volatility of CSI 300 Index have been performed using GARCH, EGARCH, APGARCH, PTTGARCH, and corresponding models with Markov regime switching, respectively. Performance of these models has been evaluated using MSE₁, MSE₂, MAE₁, MAE₂, in which the actual volatilities of series are estimated by the realized volatility using high frequency data of the series. Besides, the model confidence set (MCS) test has been used to evaluate the performance of these models. It has been concluded that the PTTGARCH models both with single regime and Markov regime switching outperform the other models in estimation and prediction of the volatilities of the return series within the sample and out-of-sample.