基于SVR-GARCH的货币波动率预测研究——以加密货币和传统货币为例  

Research on the Prediction of Currency Volatility Based on SVR-GARCH:A Case Study of Cryptocurrencies and Traditional Currencies

作　　者：刘晨旸 LIU Chen-yang

机构地区：[1]山西财经大学金融学院

出　　处：《国际经贸探索》2023年第1期52-67,共16页International Economics and Trade Research

基　　金：国家社会科学基金项目(18BJY231)。

摘　　要：货币价格波动率的建模和预测一直以来都是国家和企业层面重要的研究内容。文章使用加密货币和传统货币的分频汇率数据构建SVR-GARCH模型,为波动率的预测引入非线性交互,提高了模型的预测精度和泛化能力。此外,进一步构建了DCC-SVR-GARCH模型探讨加密货币与传统货币之间的关联性。实证结果表明:首先,在预测方面,机器学习算法的引入显著增强了模型预测性能,SVR-GARCH模型在低频和高频数据上均有稳健的预测表现。其次,加密货币波动率的特征较传统货币更为明显,两者间具有某种同步性,并进一步验证了其存在。最后,两者波动率的关联性表现为:在高不确定性时期呈现正相关关系,且具有明显的事件驱动性;在低不确定性时期往往是负相关关系。The modeling and prediction of currency volatility has always been the important research content at the national and enterprise levels. This paper constructs SVR-GARCH model based on three different frequencies data to introduce nonlinear interaction for the prediction of volatility, which improves the prediction accuracy and generalization ability of the model. In addition, the paper further constructs DCC-SVR-GARCH model to explore the correlation between cryptocurrencies and traditional currencies. The empirical results show that: firstly, in the aspect of prediction, the introduction of machine learning algorithm significantly enhances the prediction performance of the model and SVR-GARCH model presents robust prediction performance in both low-frequency and high-frequency data;secondly, the volatility of cryptocurrencies is more significant than that of traditional currencies and there is some synchronization between them, which is verified by further analysis;finally, in the period of high uncertainty, there is a positive correlation between the volatility of cryptocurrencies and the volatility of traditional currencies which is always event-driven, and in the period of low uncertainty, the correlation is often negative.

关 键 词：GARCH SVR-GARCH 加密货币 波动率 

分 类 号：F823[经济管理—财政学] F827