On improving GARCH volatility forecasts for Bitcoin via a meta-learning approach

作者：

Highlights：

• 110 GARCH-type models are constructed for Bitcoin volatility.

• A new meta-learning strategy is proposed to improve forecast accuracy.

• The proposed method leads to better performance than all the GARCH-type models.

摘要

•110 GARCH-type models are constructed for Bitcoin volatility.•A new meta-learning strategy is proposed to improve forecast accuracy.•The proposed method leads to better performance than all the GARCH-type models.

论文关键词：Volatility,Bitcoin,Model Confidence Set,Combining forecasts,GARCH

论文评审过程：Received 7 May 2021, Revised 21 July 2021, Accepted 10 August 2021, Available online 12 August 2021, Version of Record 19 August 2021.

论文官网地址：https://doi.org/10.1016/j.knosys.2021.107393