This study analyzes whether the decomposition of i nvestment sentiment i ncreases the financial market analysis power during the period from Jan. 1, 2005 to Sept. 30 of 2020. This paper finds that the model that classifies i nvestment sentiment has higher analysis power than the model that does not. This paper also finds that the model with decomposed i nvestment sentiment has higher analysis power compared to when only macroeconomic variables are used. In addition, we found a higher predictive power for the stock market by decomposing i nvestor sentiment. These findings suggest that decomposition of i nvestment sentiment i s a meaningful process for better sentiment analysis. 

main words : sentiment i ndex, Tone i ndex, category decomposition, SARIMA​