Evolutional Computing Methodology in finding Portfolio　Optimization on different risk aversion using Taiwan Morgan Index Stocks as an example

碩士論文[[abstract]]日常我們作決策是在有限資源前提及限制條件下，從一個或數個最可行的方案，選擇出最佳解。而股票投資眾多的研究，無疑都是試圖找出可依循的市場規則，進而獲取更高的報酬及較低之風險投資組合（即為多目標之求解之一例）。 本研究利用演化式計算之遺傳演算法，根據Markowitz平均數－變異數投資組合模式求投資組合之效率前緣，並根據Bodie, Kane 和 Marcus之效用函數之不同投資風險偏好行為特徵下，演算求解財務上投資者偏好及投資組合配置最佳化問題。本研究實驗觀察到利用遺傳演算法模擬逼近求取效率前緣，與傳統方法之圖形之趨近，且較傳統方法容易求解。另驗證投資整體摩根指數股及摩根指數電子股之比較時，投資電子股之報酬率雖較高，但其標準差較高（風險較高），亦證實市場中具有高風險、高報酬之特質。 另投資者在考量不同風險偏好時，研究發現愈風險趨避者之投資決策，其無風險資產之投資比重越高，其投資於風險性資產之比重較低，尤其投資於風險性較高之電子股相對地減少。[[abstract]]We frequently make our investment decision based on limited resource and information; we often select the best option according to so-called “feasible” alternatives. However, different ways of analyzing market trend are all trying to achieve one goal; in trying to predicate a trend in order to maximize returns and minimize risks. Our study utilize evolutional computing methodology（genetic algorithm）, base on Markowitz’s Mean-Variance Portfolio Model— efficient frontier of various investing portfolio , with calculating risk aversion analysis done using Bodie, Kane and Marcus’s model, to come up with methodology in identifying the best investing opportunities. This particular study utilize genetic algorithm to simulate most efficient frontier, similar to traditional graphical convexity, but easier in discovering the solution. This study also verified the fact that, when comparing Taiwan Morgan Index and Taiwan Semiconductor Morgan Index, Semiconductor sector has higher annual return, but at the same time, it also has to bear higher risks. Thus, validate the theory that there is a distinctive characteristic of high-risk high-return in the stock market. When making investing decisions, one who wants to avoid high lost, would normally choose riskyless asset