基于谱映射的非线性Sharpe模型及其在基金投资风格识别中的应用

发布时间：2018-02-26 15:34

本文关键词： 谱映射非线性Sharpe模型基金投资风格识别　出处：《系统管理学报》2017年04期 　论文类型：期刊论文

【摘要】：由于金融时间序列具有非线性的特点,经典的Sharpe模型难以全面、准确地描述经济变量之间的关系。为此,提出了基于谱映射的非线性Sharpe模型。谱映射的核心思想来源于经典的谱图理论。谱映射借助由数据集导出的一些特殊矩阵的部分特征向量,将该数据集投影至高维空间。在投影后的高维空间,数据集的有些性质得到改进,如原空间中线性不可分的数据集在高维空间线性可分。基于谱映射的非线性Sharpe模型,利用谱映射将基金收益率数据和风格指数收益率数据投影至高维空间,在高维空间利用经典Sharpe模型识别各基金的投资风格。利用基于谱映射的非线性Sharpe模型分析了部分开放式基金的投资风格。实验结果表明,本文提出的基于谱映射的非线性Sharpe模型的性能优于在原数据集上直接利用经典Sharpe模型得到的结果。
[Abstract]:Because of the nonlinear character of financial time series, the classical Sharpe model is difficult to describe the relationship between economic variables comprehensively and accurately. A nonlinear Sharpe model based on spectral mapping is proposed. The core idea of spectral mapping is derived from classical spectral map theory. Spectral mapping is based on some characteristic vectors of some special matrices derived from data sets. Some properties of the data set are improved in the projected high-dimensional space, such as the linear separable data set in the original space is linearly separable in the high-dimensional space. The nonlinear Sharpe model based on spectral mapping is proposed. Using spectral mapping, the data of return rate of funds and the data of return rate of style index are projected to high dimensional space. In high-dimensional space, the classical Sharpe model is used to identify the investment styles of various funds, and the nonlinear Sharpe model based on spectral mapping is used to analyze the investment styles of some open-end funds. The experimental results show that, The performance of the proposed nonlinear Sharpe model based on spectral mapping is better than that obtained by using the classical Sharpe model directly on the original data set.
【作者单位】：内蒙古大学经济管理学院;中国科学院数学与系统科学研究院;大连理工大学管理与经济学部;
【基金】：国家自然科学基金资助项目(71171030,61463039) 中国博士后基金资助项目(2015M581192) 内蒙古自然科学基金资助项目(2014BS0706)
【分类号】：F830.91

【相似文献】