Based On P - Norm Approximation Of Index Tracking Model And Empirical Research

Index tracking is a popular passive investment policy in recent years. We analyze one of the methods of index tracking, the optimization method of selecting a small part of the stocks contained in the index. The optimization method aims at minimizing the track error between the selected stocks and the index while the number of the stocks we picked is limited (sparse solution). Due to the constraint on the stock number, the traditional model is a mixed integer programming, which can only solve problem with limited scale. We use the p-norm model as an approximation of the0-norm constraint in order to solve larger scale problems. We also study the sparseness of the solutions using p-norm. We use CS1300index to conduct the empirical study and analyze the performance of the p-norm index tracking models.