Title: Best subset selection in regularized sparse index tracking
Authors: Nick Koning - University of Groningen (Netherlands) [presenting]
Paul Bekker - University of Groningen (Netherlands)
Abstract: The goal of sparse index tracking is to select a portfolio with a limited number of stocks in order to replicate an index. We propose a novel approach that combines $l_1$- and $l_2$-regularization (also known as lasso and ridge regression) with best subset selection. In order to estimate the portfolio, we connect sparse index tracking to a new approach for best subset selection in the standard linear regression model as suggested previously. We provide a numerical comparison to forward stepwise selection on historical data of stock indexes and illustrate the sensitivity of the tracking performance to the values of the parameters.