Portfolio optimization for the purpose of building passive income by dividends

The thesis has two objectives. The first is to investigate the effectiveness of using the Markowitz mean-semivariance model for portfolio optimization with a focus on dividend yield. The second is to describe the entire process of portfolio selection, from examining the input data, to optimizing the portfolio itself and revising it, to applying the prediction model. The study uses historical data on stocks paying dividends continuously over the past 20 years to construct a portfolio that maximizes return while minimizing risk. The methodology involves calculating the expected return and volatility of each stock and using this data to construct a portfolio that meets the investor's risk, return and dividend yield preferences. The optimization process is performed using a Markowitz mean-semivariance model that takes into account the covariances between individual stocks and the overall portfolio risk measured as the negative deviation from the market portfolio. The results show that portfolios optimized using the Markowitz mean-semivariance model outperform dividend payout-oriented ETFs traded on exchanges. In the process, the paper derives parameters affecting portfolio diversification to avoid over-dependence on the performance of a single stock or sector. The process of portfolio review and adjustment of the necessary parameters to further improve the constructed portfolios and then selecting the optimal portfolio from the efficient frontier of portfolios using Sharpe ratio and total portfolio return at different levels of investor's risk aversion is also described. Finally, the research hypotheses generated during the construction of the portfolios are evaluated using correlation analysis and finally, the possibility of applying the GARCH model to predict portfolio volatility is demonstrated.