Machine Learning and Equity Investment

Student

Miao Yang

Advisor

Arindam Banerjee

Abstract

Back in the 1980s, equity Investment was dominated by stocks selections based on manual, qualitative fundamental analysis. Later as computers and statistics enter into the industry, quantitative investment emerges that practitioners develop quant models to evaluate company characteristics and build algorithmic stock portfolios. However, even nowadays, most quant approaches are still based on linear models, e.g., CAPM, Fama-French, and smart beta. In this project, we test machine learning models on stock return predictions, identify best performing methods, and show their predictive gains. This helps highlight the value of machine learning in financial innovation and suggest potentials in improving equity investment performance.