Boston File: Binary Logistics Regression & Industry Modeling by Russ Koesterich

Most conventional financial models are intended to provide a point-estimate, often of an expected return. These estimates are necessary for portfolio construction and mean variance optimization. However, there is another potential way to frame financial outcomes. Rather than trying to model expected returns, the problem can be framed as an outcome, and the model’s objective function as a probability.

Leave a Reply

Your email address will not be published. Required fields are marked *