##plugins.themes.huaf_theme.article.main##
Abstract
In modern financial investment, finding and building a powerful and fast-correcting tool in portfolio optimization has become an urgent need for investors to make more accurate decisions. Machine learning methods are becoming increasingly popular in financial research, from predictive analysis to complex ones such as in portfolio management and optimization. This paper uses Random Forest (RF) and Support Vector Regression (SVR) methods to forecast stock returns, and conducted a comparative test of two portfolio optimization strategies: MVP (Mean-Variance Portfolio) and MVF (Mean-Variance with Forecasting). The study used a portfolio of stocks in the 30 largest capitalization stocks (VN30) of the Ho Chi Minh City Stock Exchange. Following the application of the screening conditions, the research sample comprises 22 stocks. The research results showed that the combination of the SVR and MVF methods brought the best efficiency in portfolio optimization strategies.
Keywords: Portfolio optimization; Machine learning; VN30; Random Forest; Support Vector Regression.