Application of a Multi-factor Linear Regression Model for Stock Portfolio Optimization

Abstract

Multi-factor models of asset pricing indicate a linear relationship between the expected return of assets while exposing to one or more risks. In this study, the sensitivity of portfolio returns to 4 selected macroeconomic factors (Market Performance, Real GDP, Inflation, Unemployment) were examined by means of multiple linear regression analyses with regard to multi-factors. Experiments show that the generalized approach of moment estimators of risk premiums lead to better results on individual assets over historical averages. Especially when the factors are weakly correlated with assets which indicates that the factors selected should be less correlated with each other.