Application of Bollinger Bands and Genetic Algorithms for Optimal Portfolio

Autor:	Chien-Chien Hu, 胡茜茜
Druh dokumentu:	學位論文 ; thesis
Popis:	103 Modern portfolio theory is based on the Mean Variance (MV) model proposed by Markowitz, using variance to measure risk. However, the real data often fails to meet the assumption of normality, which is required by the MV model. Bollerslev (1986) published the GARCH model to solve the above problem. In this study, we use the Bollinger Bands as a basis to change the weights, use GARCH model and DCC-GARCH model to predict risk, and use Genetic Algorithms to construct the optimal portfolio. The results show that, using historical monthly data, Genetic Algorithm achieves the best performance with population size of 100, mating rate of 0.8, and mutation rate is 0.01. The results are also better than that of the equal weighted portfolio.
Databáze:	Networked Digital Library of Theses & Dissertations
Externí odkaz:	http://ndltd.ncl.edu.tw/handle/44081333224745787345 Zobrazit plný text záznamu