Optimalizace portfolia akcií obchodovaných na burze NASDAQ

The topic of this thesis is the optimization of stock portfolios on the stock exchange NASDAQ. The main objective is to use a training and testing period to evaluate the performance of chosen portfolios obtained from a broader sample of NASDAQ-listed stocks using the mean-semivariance model. The next aim is to create a set of efficient portfolios based on a longer time series optimization and finally to make recommendations to two investors inspired, in their characteristics, by two real acquaintances of the author. The secondary aim is also to summarize the key features of selected investment vehicles and the fundamental concepts of modern portfolio theory. The mean-semivariance model is used to optimize the set of 107 preselected stocks, first from 2013 to 2018 and subsequently for the period of 2013 to the present day. Selected portfolios from the first set of efficient portfolios are utilized to depict how they would have fared in reality in the period of 2018 to 2023. Even though the portfolios were following the trends of the stock index NASDAQ Composite, overall, they both ended up being more profitable. The second optimization has two scenarios – one for the whole set of 107 stocks and one for the set with the exclusion of a stock with extreme characteristics. Then, 15 efficient portfolios are derived, the efficient frontier is constructed, and chosen portfolios are recommended to the investors.