基于贝叶斯理论的长、短数据资产组合选择

Portfolio Selection of Unequal Histories of Returns with Bayesian Tramework

本文利用贝叶斯理论对不同市场上长度不一致的股票数据进行了回归分析刻画,并在改进的收益和协方差矩阵基础上构建了投资组合选择模型。算例结果表明,在股票数据长度不一致时,基于贝叶斯理论改良的投资组合选择模型比截断数据的投资组合模型有更好的表现,说明贝叶斯理论能够挖掘长数据被截断部分的信息,对于优化投资组合有积极的意义。

Unequal histories of returns make trouble to the portfolio selection.In this paper,Bayesian methods are used to measure the dependence of the short series on the long series,and a portfolio selection model is constructed with the modified return and covariance matrices.The numerical example shows that when the historical data is of unequal lengths,the improved portfolio selection in Bayesian framework performs better than the traditional method with data truncated,which implies that the Bayesian theory helps dig the information of truncated data.