| Summary: | 碩士 === 長庚大學 === 工商管理學系 === 105 === Markowitz's mean-variance portfoilo optimization model measures expected returns and risks of portfolios with historical rate of return and corresponding variance, opening the portfolio selection research framework. However, Markowitz’s model has been inefficient in using quadratic programming to solve large-scale portfolio selection problems, owing to the proloneged computation time that rapidly increases with the asset number and thus the size of the covariance matrix. Furthermore, since Markowitz’s model represents a multi-criteria decision making problem, this study uses a multiobjective genetic algorithm to extract the portfolio efficient frontier so as to increase the diversity of the portfolio. In this study, use the ETF, options, and bond data from 2010 to 2014. There are 300 kinds of porfolio selections in Markowitz’s model, although there are 171 kinds of porfolio selections in Markowitz’s model by NSGA-II, which increases portfolio efficiency, and its construct the efficient frontier is also very close to all the combinations of the efficient frontier.
|
|---|
