Mean-Variance optimal portfolio selection integrated with support vector and fuzzy support vector machines

Abstract

This study introduces a novel approach integrating a support vector machine (SVM) with an optimal portfolio construction model. Leveraging the Radial Basis Function (RBF) kernel, the SVM identifies assets with higher growth potential. However, due to inherent uncertainties, some input points may not be precisely classified into their respective classes in various applications. To mitigate the influence of noise, a new fuzzy support vector machine (NFSVM) is employed to select assets. Here, each sample point is assigned a membership value using a fuzzy membership function, as documented in existing literature [1]. Additionally, the SVM model incorporates principal component analysis (PCA)to eliminate correlated technical indicators. Further, Markowitz’s mean-variance model (MV model) with cardinality constraints and without cardinality constraints is employed for the assets selected by SVM, FSVM, and NFSVM for optimal portfolio construction.The performance of the proposed model is experimentally assessed using a data set derived from the Nifty 50 and Euro Stoxx 50 index. The experimental results demonstrate that the optimal portfolio obtained from the NFSVM with the Markowitz mean-variance model outperforms the one generated by the SVM. This outcome substantiates the effectiveness and efficiency of the proposed model as an advanced approach for optimizing investment portfolios.

Keywords

• fuzzy support vector machines
• markowitz mean-variance model
• portfolio optimization
• classification
• prediction
• fuzzy membership function