A Comparison between Parametric and Nonparametric Volatility Forecasting of Stock Index Futures in China. Issue 9 (15th July 2022)

• Record Type: Journal Article
• Authors: Jiang, Rui; Wen, Conghua
• Is Part Of: Emerging markets finance & trade. Volume 58:Issue 9(2022)
• Page Start: 2522
• Page End: 2537
• Publication Date: 2022-07-15
• Subjects: Stock index futures -- volatility forecasting -- intraday data -- HAR model -- GARCH model; G15 -- G17; Balkan Peninsula -- Commerce -- Periodicals; Balkan Peninsula -- Foreign economic relations -- Periodicals; Europe, Central -- Commerce -- Periodicals; Europe, Central -- Foreign economic relations -- Periodicals; Europe, Eastern -- Commerce -- Periodicals; Europe, Eastern -- Foreign economic relations -- Periodicals; Turkey -- Commerce -- Periodicals; Turkey -- Foreign economic relations -- Periodicals; Asia -- Commerce -- Periodicals; Asia -- Foreign economic relations -- Periodicals; Eurasia -- Commerce -- Periodicals; Eurasia -- Foreign economic relations -- Periodicals; 382.05
• DOI: 10.1080/1540496X.2021.2002142
• Languages: English
• ISSNs: 1540-496X

Abstract

ABSTRACT: In this study, the volatilities of CSI300 index futures, SSE50 index futures, and CSI500 index futures are modeled and predicted based on both parametric and nonparametric modeling approaches. Four ARMA-GARCH-type models and four HAR-type models are taken as the framework of volatility prediction. The last one-third of transaction data are used as the testing sample and the rolling window approach is adopted for prediction. The best predictive models for these three stock index futures vary with the properties of the futures, while volatility prediction based on the HAR-type models always has a higher accuracy than the ARMA-GARCH-type models. Moreover, we find that the property of target assets influences the performance of models, and the choice of extended models in prediction is suggested to be based on the peculiarity in the sample.