本研究的目的是試圖經由眾多股市技術分析指標中,萃取出相關因素做為個股情緒指數,以做為股價預測模型之參考。本研究以台灣上市公司之營建類股指數及7檔個股為研究對象。研究期間為2002年1月至2012年12月之2,737筆日資料。研究變數包括5K、5D、9MACD、12MACD、26MACD、DI、9EMA、26EMA、9DIF、12DIF、26DIF、5BIAS、10BIAS、5RSI、10RSI、15RS、30RSI等17個指標。本研究以因素分析模型之主成分分析法及直交轉軸程序,分別針對8檔標的股之技術指標,各萃取出三個構面,並將三個構面與股價報酬進行複迴歸分析。實證顯示:三個標的股因素分析之情緒指數具高度一致性,三種情緒指數對個股股價報酬都具有顯著影響效果,除了MACD之情緒指數對少數個股股價報酬不具有顯著效果外。建議藉由技術指標建立個股之情緒指數,可有效解釋及預測個股報酬。
The purpose of this study is to construct stock sentiment indexes based upon technical analysis indexes for the establishment of stock price forecasting model. The sample consists of construction sector index and 7 major firms from the sector. The daily data ranges from January, 2002 to December, 2012, a total of 2,737 observations, were collected. Seventeen technical indexes include 5K、5D、9MACD、12MACD、26MACD、DI、9EMA、26EMA、9DIF、12DIF、26DIF、5BIAS、10BIAS、5RSI、10RSI、15RS、30RSI were adopted. The principle factor analysis with orthogonal rotation procedure was executed for the eight stock series, and 3 factors were extracted from each series. Then regression analyses were performed. The results show that the 3 factors extracted from each series are highly consistent and demonstrated statistically significant impacts on stick returns, beside 1 factor - MACD extracted from two major firms are not demonstrated statistically significant impacts on stick returns. Thus, the procedure proposed by this study might be useful in explaining and forecasting stock returns.