Features Selection in the Age of Generative AI

By QTS Capital Management LLC Prepared by Ernest Chan, Chairman, and Nahid Jetha, CEO Features are inputs to machine learning algorithms. Sometimes also called independent variables, covariates, or just X, they can be used for supervised or unsupervised learning, or for optimization. For example, at QTS, we use more than 100 of them as inputs to dynamically calibrate the allocation between our Tail Reaper strategy and E-mini SP 500 futures. In general, modelers have no idea which features are useful a priori, or if they are redundant, for a particular application. Using all of the features can result in overfitting and poor out-of-sample performance, or worse, numerical instability and singularities during matrix inversion. Hence the need for a process called “feature selection”. Read more...