Index Tracking Based Portfolio Selection Model And Its Empirical Analysis

Indexing investment, one of the most important methods of passive invest-ment management, is widely used by index funds and other institutional investors. As a practical technique of indexing strategy, index tracking is strongly supported by modern financial theory, especially by modern portfolio theory, and has gained considerable attention both in academic research and financial practice.Choosing a proper replication method is the core of index tracking problem. In this thesis, combining factor models with mean-variance model, we present a cardinality constrained index tracking model based on multifactor model. Our model minimizes the variance of the portfolio with constraints on factor betas and expected excess return. Taking into account the practical investment man-agement, we also consider restrictions on the total number of stocks contained in the portfolio. To solve this cardinality constrained index tracking model more efficiently, we reduce the original model to a new equivalent0-1mixed integer pro-gramming model via Lagrangian dual and conic optimization techniques. Then, using data from China Stock Market, we construct three artificial indices and compare the empirical performance of model with classical Mean-Variance index tracking model. Furthermore, we carry out empirical studies on model perfor-mance based on CSI300index. Empirical results indicate that our index tracking model is quite accurate and flexible. Finally, we extend our model by considering relative risk of portfolio.The thesis is organized as follows:In Chapter1, we introduce passive invest-ment management and indexing strategy. In Chapter2, we introduce theoretical results on index tracking and basic models, such as Mean-Variance index tracking model, Mean-Absolute-Deviation index tracking model and Single Factor index tracking model. In Chapter3, we present a cardinality constrained index tracking model based on multifactor model. In Chapter4, we investigate the empirical performance of our model using real data from China Stock Market. In Chapter5, we extend our model by considering relative risk of portfolio. Finally, we sum- marize the main results of the thesis and discuss some future research directions in Chapter6.