论文部分内容阅读
大维数据给传统的协方差阵估计方法带来了巨大的挑战,数据维度和噪声的影响使得协方差阵的估计较为困难.在文章的研究中,将高频数据和低频数据相结合,提出了基于混合频率数据的协方差阵的估计和预测模型——MFD模型,MFD模型在解决了维数诅咒的同时还考虑了过去市场信息对协方差阵的影响,动态地估计和预测了未来的协方差阵.通过实证研究发现:较基于低频数据和高频数据的协方差阵估计和预测模型而言,MFD模型明显提高了高维协方差阵的估计和预测效率;并且将其应用在投资组合时,投资者获得了更高的投资收益和经济福利.
Large dimension data pose great challenges to the traditional covariance matrix estimation method, and the influence of data dimension and noise makes it difficult to estimate the covariance matrix.In the research of this paper, the high frequency data and the low frequency data are combined to propose The MFD model, which is an estimation and prediction model of covariance matrix based on mixed frequency data, considers the influence of past market information on the covariance matrix while solving the curse of dimensionality, and dynamically estimates and predicts the future Covariance matrix.Experimental results show that compared with the covariance matrix estimation and prediction model based on low-frequency and high-frequency data, the MFD model significantly improves the estimation and prediction efficiency of high-dimensional covariance matrix and applies it to the portfolio , Investors get higher investment income and economic welfare.