Factor investing is an investment strategy that aims to capitalize on specific characteristics associated with higher returns such as value, momentum, or quality which have historically been drives of above-average performance.
Machine Learning can find hidden patterns in data previously never discovered. Using machine learning, investment managers can uncover drivers of returns in the financial markets that have not previously been discovered. Moreover, machine learning can also analyze a wide range of structured and unstructured data such as social media activity, news sentiment, and financial reports which could further help in discovering patterns previously never found that could be used to reduce risk and increase their returns.
Traditional factor models used in investing assume linear relationships; however, machine learning model techniques such as neural networks can capture nonlinear interactions between factors, providing a more in-depth insight into how factors interact and affect the returns of the investments. Research supports the superiority of models using machine learning within factor models such as the Stambaugh and Yuan (2016) four-factor model which demonstrated superior performance in the time-series domain because of its machine-learning components.