Multi-objective portfolio optimization in stock market

Abstract

Portfolio optimization is an important research area in finance, because most investors would like to diversify their investments by holding a portfolio rather than investing in a single stock to earn a higher rate of return and to reduce risk. Multi-objective portfolio optimization can achieve an optimal portfolio which generates the highest rate of return with lowest risk. This paper proposes a new approach to portfolio construction, including stock selection and stock allocation. The approach then is tested using stock price data and other financial information of 113 stocks over a one-year period. K-means clustering is applied on this dataset in order to categorize data into several groups to finally identify a pool of potential stocks. Finally, we construct the efficient frontier of optimal portfolios using both the single-objective and multi-objective approaches and compare the results.