金融资产收益的高频协方差矩阵研究：估计、预测与应用

本项目拟构建同时考虑非同步交易、市场微观结构噪声以及跳跃影响的金融资产收益的高频协方差矩阵估计量，以及同时考虑维数灾难问题、协方差矩阵的正定性和对称性以及金融资产波动性的长记忆性的动态预测模型来预测高维的金融高频协方差矩阵，并构建稳健的预测精度评价方法来实证评价和比较各类协方差矩阵预测模型的预测性能。此外，本项目还在此框架下进一步探讨高频协方差矩阵在投资组合构建、波动率溢出和动态套期保值方面的应用。. 期望本项目的研究成果能推动高频协方差矩阵的估计和预测研究的理论进展，完善和丰富其在金融市场中的应用研究，为金融监管部门提高金融市场的风险监管水平提供决策参考，以及为各类投资者优化资产组合的配置，度量、预测和防范金融资产价格波动的风险提供实践指导。

This study proposes a new estimator for high-frequency covariance matrices of financial asset returns which take into account the impact of non-synchronous trading, the market microstructure noises and jumping simultaneously, and develops a new dynamic forecasting model which accounts for the curse of dimensionality, the symmetry and positive definiteness of covariance matrices and the long memory in financial asset volatility simultaneously to forecast high-dimensional high-frequency covariance matrices, and proposes a robust prediction accuracy evaluation method to evaluate and compare the forecast performance of various forecasting models for covariance matrices. In addition, under this framework, this study also further explores the high-frequency covariance matrices applications in portfolio construction, volatility spillover and dynamic hedging.. The purpose of this study, we expect, is to promote theoretical progress in the estimation and prediction of high-frequency covariance matrices, to improve and enrich the application research on high-frequency covariance matrices, to provide references for the financial supervision department to improve the financial markets risk supervision level, and also, to provide practical guidance for all kinds of investors to optimize investment portfolio, measure, forecast and guard against the risk of financial asset price fluctuations.