Cross-Sectional Stock Return Analysis Using Support Vector Regression

J Liu*, Z Hu*, S Tan

Abstract

We adopt $\varepsilon$-Support Vector Regression, a nonlinear regression method, to analyse the relationship between stock return and explanatory factors. Computational results show $\varepsilon$-SVR outperforms significantly the Ordinary Least Squares linear regression with a much higher $R^2$ and a lower standard error.