應用滾動灰預測模型與自組性演算法建立股票買賣決策系統

Abstract

由於目前全球經濟不景氣，民眾薪水低且物價高，銀行利息遠不如投資股票之獲利，使股票買賣成為一般民眾投資的主要途徑。然而，投資股票時必須要能準確掌握股價走勢，以決定適當之股票買賣時間點才能獲利，這對資訊有限之投資散戶而言尤其重要，因為散戶不像專業投顧公司，有足夠的專業知識及分析工具來買賣股票。因此，本研究之主要目的是針對股票投資散戶，使用一般投資者皆可取得之股票資訊，建立一套股票買賣決策系統。另外，為提升僅使用單一預測方法之預測準確度，本研究結合滾動灰預測模型與自組性演算法來預測股價。首先，應用滾動灰預測模型(Rolling Grey Forecasting Model, RGM)建立股價預測模型，以預測隔日股票價格及漲跌幅度；再應用自組性演算法(Group Method of Data Handling, GMDH)整合八項股票技術指標之資訊，建立股價走勢預測模型，以預測隔日之股價走勢。最後，結合股價預測模型及股價走勢預測模型所得之資訊，建立一個股票買賣決策系統，以提高使用單一預測方法之預測準確度並提高投資報酬率，以提供一般投資者股票買賣決策參考。本研究最後利用台灣五十成分股中之八支股票來說明本研究方法確實有效。

Due to the economic recession in recent year, the interest rate of banks is much lower than the profit from the stock investment. Stock investment becomes a popular investment tool to gain profit. Stock investors must know the proper timing of buying and selling stocks to gain high profit. However, it is difficult to decide the proper trading time for the retail investors whose knowledge and information on stocks are less than investment consulting company. Therefore, the main objective of this study is to utilize the information of stock prices and technical indices which a retail investor can easily obtain to construct a stock trading decision system. First, a Rolling Grey Forecasting Model (RGM) is constructed to forecast the stock prices and the change of stock prices between different time points. Group Method of Data Handling (GMDH) is then utilized to forecast the trend of stock prices of next time point. The proposed stock trading decision system is constructed by integrating the change RGM and GMDH models. Finally, eight Taiwanese stocks are utilized to demonstrate the effectiveness of the proposed method.