Apply Grey Relation Analysis, Genetic Algorithms And Fuzzy Neural Network To Forecast Taiwan Stock Exchange Weighted Index

• Main Authors: Chen, Tsungching, 陳宗敬
• Other Authors: Lee, Liangchien
• Format: Others
• Language: zh-TW
• Published: 2012
• Online Access: http://ndltd.ncl.edu.tw/handle/65852516622807429189

碩士 === 義守大學 === 財務金融學系 === 100 === Stock markets have been an essential indication in economy market for many countries. The main reason for this is because the stock market index could reflect the expected profitability of listed companies in current period beforehand, also is able to predict the economic sentiment index. The Greek debt crisis in May 2010 triggering the subsequent debt crisis in Europe; the earthquake in Japan in March 2011, and the same month the civil wars in Libya, all of these factors had a profoundly impact on stock markets around the world, no exception for Taiwan stock exchange weighted index as well.In this case study, the gray relational analysis will be used to examine the factors which are highly relavant in affecting the Taiwan stock exchange weighted index (e.g. Tokyo Nikkei 225、Shanghai SSE Composite Index、Standard & Poor 500 Index、Shenzhen Component Index、 Frankfurt DAX、Dow Jones Industrial Average、London FT100、adjusted debit balance bearish、moving average convergence divergence、moving average). The main purpose of using gray relational analysis is to filter out the relationship between the target and the reference values. Moreover, gray relational will be used to analyze 23 variables which are relevant with Taiwan stock exchange weighted index, and filter out the top ten relevant variables to constrict the prediction model in order to forecasting Taiwan stock exchange weighted index. The result can be the standard to assist investors choosing the better investment. The figures of Taiwan stock exchange weighted index from 2008 to 2010, will be inserted to analyze this project. The test model will be constructed through genetic algorithms, fuzzy neural network and fuzzy genetic algorithm, moreover, the index of Taiwan stock exchange weighted and the relevant important factors will be applied into the empirical analysis, in order to define the actual effect of genetic algorithm, fuzzy neural networks and fuzzy genetic algorithms. Single solution best practices in the individual variables and global variables, almost all of the genetic algorithm are the minimum error, and optimal average and final average optimal methods are fuzzy neural networks with minimum error.