中文摘要 股票投資是現代大部分人會面對到的課題。投資者無非是希望能從股市中獲取報酬,但卻因知識與資訊的匱乏,導致判斷錯誤造成財富的損失。過去已有許多研究期望找出一個適當的股價預測方法,來降低買賣股票的風險。 本研究利用近期發展的馬氏田口系統(Mahalanobis-Taguchi System, MTS)應用在股價預測上。從文獻與專家訪問中,找出判斷股價漲跌的顯著指標,並考量投資者實際買賣行為,將指標與買賣過程整合納入模式之中。利用MTS建立一個資訊辨識系統來區分正常群體與異常群體,並以田口方法進行參數優化以縮減模式,提供一個投資建議予投資者,達到穩健與降低時間成本之成效。
Abstract Security investment is an important issue that modern most people will face. Investors only hope to obtain the remuneration from the stock market, but because knowledge and information are deficient, result in judging the mistake causes the losses of the wealth. In the past a lot of research expected to find out an appropriate stock price prediction method reducing the risk. This research applies Mahalanobis-Taguchi System (MTS) developed recently in stock price forecasting. From literature and visiting expert, find out the notable Indicators of judging the pattern of the stock price and consider investor’s actual behavior, combine those factors to MTS. Utilize MTS construct an information recognition System which differentiate between normal group and abnormal group. And then the model optimizes the indexes to reduce the model which give a suggestion to investor arrived the result of robustness and cost down by Taguchi Method.