Data analytics enhanced component volatility model

作者：

Highlights：

• We propose a neural network enhanced hybrid model to forecast realized volatility.

• Hybrid model is constructed by HP filter, neural network and autoregressive model.

• Application of hybrid model on one-hour and one-day realized volatility.

• Hybrid model significantly outperforms selected traditional volatility models.

摘要

•We propose a neural network enhanced hybrid model to forecast realized volatility.•Hybrid model is constructed by HP filter, neural network and autoregressive model.•Application of hybrid model on one-hour and one-day realized volatility.•Hybrid model significantly outperforms selected traditional volatility models.

论文关键词：Autoregressive neural network,Hybrid model,Two-component,Volatility model

论文评审过程：Received 11 January 2017, Revised 17 April 2017, Accepted 9 May 2017, Available online 10 May 2017, Version of Record 15 May 2017.

论文官网地址：https://doi.org/10.1016/j.eswa.2017.05.025