Statistical Learning in Finance

The statistical foundations of machine learning, taught through the problems of modern finance.

The course moves from the classical toolkit (linear regression, logistic regression, and k-nearest neighbors) to the methods behind modern applied work: support vector machines, decision trees and ensembles (bagging and boosting), neural networks, and unsupervised learning. Equal weight is given to the judgment that makes these tools useful: assessing models honestly, selecting features carefully, and recognizing when a method's assumptions break down.

Throughout the semester, students work hands-on with real financial datasets, applying each method to problems in asset pricing, portfolio optimization, and credit risk assessment.