Možnosti využití statistických metod při tvorbě investičního portfolia

This thesis deals with stock investing and investment portfolio construction. The thesis develops a methodology to help the individual investor in navigating on the stock market. The methodology focuses on stock market dynamics and it is primarily based on graph theory. It mainly uses the minimum spanning tree method to capture the hierarchical structure and topological arrangement of the stock market. Based on this structure we look for stocks that exhibit high diversification potential. To identify these stocks, we use the characteristics and attributes of network graphs or use methods derived from social network analysis. From the selected stocks we design investment portfolios that are based on modern portfolio theory. In practical part of the thesis the developed methodology is applied to time series of stock prices of 492 companies that are in the SP&500 stock index in time window from October 2016 to October 2021. Based on the results of the methodology we argue that by using the minimum spanning tree graph it is possible to track the evolution of the hierarchical structure of the stock index over time and that this structure occupies a certain topological arrangement that changes depending on the movement of the whole index. Using the methodology we identify from the minimum spanning tree graph stocks with good diversification potential and construct investment portfolios from these stocks. All portfolios based on this methodology show decent results compared to the long-term performance of the SP&500 index.