股票的價值應該來自於公司的獲利,因此股價應該合理的反應出公司營運的狀況及獲利的程度,本研究試圖由代表公司營運及獲利狀況的財務比率,找出預測股價漲跌的參考規則。本研究使用資料探勘技術中的分類樹模型來作為分析的工具,分類樹模型在尋找自變數與因變數的規則上,常有不錯的表現,而Salford System所研發的CART軟體則是本研究中的主要分析工具。 本研究以台灣地區的電子類股作為研究標的,研究樣本期間為民國八十四年第一季起至民國九十年第二季止,資料來源的台灣經濟新報。資料樣本以季為單位,自變數為公司的各項財務比率資料,因變數為季股價報酬率資料。研究樣本群以民國八十九年第二季(含)以前為學習樣本群,共計1885筆,用來成長分類樹模型,而民國八十九年第二季以後資料為測試樣本群,共計674筆,用來驗證分類樹模型的分類預測能力。 經由分類正確率及模擬投資報酬率的分析,均顯示研究所得的CART分類樹模型有效。該模型指出,在所有使用的公司財務比率中,與去年同期比較的利益成長率對分類預測季股價報酬率最具成效。透過本研究可以了解,的確可以利用公司財務比率資料對股價報酬率的趨勢作一定程度的預測,而CART分類樹模型也的確具有相當的研究成效,值得進一步加以應用。
The stock price should reasonably reflect the status of the company’s management and profitability. This paper tried to use CART to build a model in order to forecast the stock price rising or falling from the company’s financial ratios. The CART software developed by Salford-Systems is the main analysis tool in this study. This study uses the electronic stocks in Taiwan from the first season of 1995 through the second season of 2001 as samples, and all data informations are from Taiwan Economic Journal (TEJ). The independent variables are the company’s financial ratios and the dependent variable is the stock return. The samples before the second season of 2000 are used as the learning samples, and the samples after that are used as the testing samples. Through the analyses of rates of accuracy and the returns of simulated investment, it is proved that the classification tree constructed in this study is worked, and on all financial ratios, the growing rates of profit are the significant attributions on classification.