Index Tracking by Means of Optimized Sampling

Abstract

We read often that investors are frequently satisfied with an average return, such as the return on an index. To develop a universally applicable model for constructing a portfolio that tracks an index as accurately as possible, one should focus primarily on its practical use in the period after optimization. The objective is to define a limited number of stocks for the portfolio and to maintain it with minimal transaction costs. The approach proposed here requires fewer undesirable assumptions than we see in other approaches to indexing. One-year backtests of the model with the MSCI Europe Index indicate that the optimized index portfolio performs much like the index.