本研究所採取的情緒指標分別為賣買權成交量比、賣買權未平倉量比、週轉率、融資餘額、融券餘額以及VIX指數，並將情緒指標依過去分為百分之十的極高值和極低值兩組，試圖探討當情緒指標達極端值時，對市場是否代表超賣或超買。樣本期間為2004年1月2日至2011年6月30日，以情緒指標所發出的訊號為事件日，再以事件研究法檢定事件日前與事件日後是否有異常報酬。檢定平均異常報酬率，有兩種檢定法分別為有母數與無母數檢定，本研究將同時採用有母數檢定(普通橫剖面法)及無母數檢定(Wilcoxon符號順序檢定)，以增加結論上的穩定性。檢定結果發現，不論是普通橫剖面法與Wilcoxon符號順序檢定法皆顯示事件日前與事件日後的平均市場報酬率有顯著的異常現象，代表情緒指標在股市裡具有預測能力及有可能帶來超額的報酬。
The thesis tries to employ sentiment indicators to figure out the possible abnormal returns in the Taiwan stock market. These indicators used are the put-call volume ratio, the put-call open interest ratio, the turnover rate, the margin loan balance, the short bal-ance, and the VIX. The highest 10% and the lowest 10% values of each index are con-sidered as the extreme values which represent the stock market might be overbought or oversold. The events generated from the extreme values are used to examine whether there are abnormal returns after the events. The sample period was from January 2, 2004 to June 30, 2011 and the research method is the event studies using parametric and nonparametric tests. The empirical results show that both tests indicate the average market returns after events are significantly different from those before events. There-fore, the sentiment indicators may have some ability to predict the stock market and generate excess returns.
