00000nam a22000004a 4500 UP-8027390931311479438 Buklod 20250122125659.0 a r |||| u| ta 250102s2024 xxu r |||| u|eng d DSTC rda eng LG 995 2024 S8 L49 Leynes, Hans Joshua C. author. Philippine Stock Exchange Portfolio Recommendation Using Forecasting Clustering, and Optimization Techniques Hans Joshua C. Leynes; Katrina D. Tan. UP School of Statistics, University of the Philippines-Diliman, Quezon City, July 2024. 193 pages colored illustrations 30 cm Includes index. ABSTRACT Capital appreciation and capital growth are essential to meet everyday needs this day and age. The passive income from stock investment has significantly grown. But to succeed in stock market, one must choose their investment wisely. Factors such as the volatility of stock returns and the abundance of assets one can invest in need to be considered. IN order to reduce the total risk of one's portfolio, investors can diversify their holdings across a variety of stocks and assets. In this paper, we aim to provide a way to tackle mentioned challenges by recommending diversified stock portfolios which maximize returns while minimizing risks from the Philippine Stock Exchange. This will be done through a four step process. First, we will use ARIMA-GARCH forecasting to predict stock returns and their volatility. Then, we will compare 2 approaches for selecting the stocks to be included in the recommended portfolio. The first approach involves selecting the stocks with the highest average Sharpe Ratio for each week which will be computed using the forecasted returns and volatility. For the second approach, we will use SOM clustering based on the forecasted stock returns and volatility to cluster the stocks. Then, from each cluster, we will select a stock that will be included in the portfolios based on their Sharpe ratio. After Selecting the stock that will be included in the portfolio using these two approaches, we will use MGARCH forecasting to generate the forecasted returns and covariance matrix for for the selected stocks. Lastly, we will compare the Genetic Algorithm and Markowitz Mean-Variance Model as approaches for finding optimal portfolio weight allocation, with the forecasted returns and covariance matrix as our inputs. We will assess the recommended portfolios by comparing their Sharpe Ratio and Daily Expected Returns per week to that of the Philippine Stock Exchange Index (PSEI) and BPI Odyssey Philippine High Conviction Equity Fund, which will be our baseline and aggressive fund respectively. Based on the results, all recommended portfolios outperformed the benchmarks in terms of Cumulative Returns and Sharpe Ratio for both test periods covered. The top performs in the terms of Cumulative Returns are the SOM Clustering-Genetic Algorithm portfolio (89.6%) for the Pre-Pandemic period (2020) and the Highest Sharpe Ratio-Genetic Algorithm portfolio (59.9%) for the Including Pandemic period (2021). In terms of overall Sharpe Ratio, the Highest Sharpe Ratio - Genetic Algorithm portfolio was the best performer for both the Pre-Pandemic (.09498) and Including Pandemic (.13625) periods. Forecasting Analysis Risk Diversification Portfolio Optimization Autoregressive Integrated Moving Average (ARIMA) Multivariate Generalized Autoregressive Conditional Heteroskedasticity (MGARCH) Dynamic Conditional Correlation (DCC) Sharpe Ratio Self Organizing Maps (SOM) Genetic Algorithm Markowitz Mean-Variance Model. Tan, Katrina D. author. FI UPD DSTC LG 995 2024 S8 L49 Thesis