论文部分内容阅读
波动率可以衡量市场风险,对其准确预测在衍生品定价、风险管理和资产配置等方面有重要意义,是政府、资本市场及投资者共同关心的热点问题。本文在介绍了现有典型的各种波动率预测模型的基础上,将表示隐含波动率的市场波动率指数(CVX)作为影响因子引入高频数据HAR模型,构成HAR-CVX模型,该模型既利用了股票交易的高频数据,又利用了期权模拟交易的信息,最大程度地综合了可以利用的信息,所以预测效果更佳。以沪深300指数为研究对象,将几种常用波动率预测模型(GARCH模型、SV模型和HAR模型)与所提出的HARCVX模型进行滚动时间窗口样本外预测,并采用4种损失函数和SPA检验,对这几种模型的预测效果进行了评估,发现基于高频数据的HAR模型表现优于基于日收益率数据的GARCH模型和SV模型,并且加入了隐含波动率的HAR-CVX模型的预测效果更好。
Volatility can measure market risk. Accurate forecasting is of great significance in derivatives pricing, risk management and asset allocation. It is a hot issue of common concern for governments, capital markets and investors. Based on the existing typical volatility forecasting models, this paper introduces the market volatility index (CVX), which represents the implied volatility, as the influencing factor into the high-frequency data HAR model to form the HAR-CVX model. This model It not only takes advantage of the high-frequency data of stock trading, but also uses the information of the option to simulate the transaction to maximize the available information so that the forecasting effect is better. Taking the Shanghai-Shenzhen 300 Index as the research object, several popular volatility prediction models (GARCH model, SV model and HAR model) are combined with the proposed HARCVX model for out-of-sample rolling time window prediction. Four loss functions and SPA tests , The prediction results of these models were evaluated and found that the HAR model based on high-frequency data performed better than the GARCH model and SV model based on daily rate of return data and the prediction of the HAR-CVX model with implied volatility Better results.