關鍵字: 情緒指數、CBSI、石油期貨、格蘭傑因果檢定、Garch-in-Mean ;This thesis explores whether the sentiment index can forecast the changes of oil futures prices, focusing on studying changes in the West Texas Intermediate and Brent crude oil futures prices. The sample period is from 1994 to 2019, using eleven types of monthly-, weekly- and daily-frequency sentiment indexes. We use each individual sentiment index to make prediction and compare the predictive power among them. The research methods include the standard predictive regression model, Granger causality test and Garch-in-Mean model. The empirical results show that some of the sentiment indexes effectively predict oil futures returns. Furthermore, the result in the Garch-in-Mean test also shows that the professional investor sentiment index - CBSI sentiment index has a smart money effect on the West Texas Intermediate crude oil futures returns. In addition, this index also has a considerable predictive ability. Through this study, investors can understand the efficacy between each sentiment index, and also use these sentiment indexes to judge the changes in oil futures prices when investing in the oil futures markets.
