Can extracted sentimental features from stock forum account for the stock return?

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

Abstract

This paper provides novel evidences for the link between stock return and sentiment of investors. Firstly, an efficient method is employed to improve the accuracy of sentiment polarity classification at the level of paragraphs for Chinese online reviews. Next, we choose Bull Index (BI), Difference Index of Sentiment (DIS) and turnover (TR) as indicators of sentiment indexes and we select stock return (Re) and trade volume (TV) as variables of stock indexes. Then, multiple liner regression and VAR model are used to discover the relationship between sentiment indexes and stock indexes. The results show that grouped sentiment indexes can significant forecast stock indexes than single explanatory variable. Besides that, both BI and DIS have mutual granger cause relationship with Re and BI is the granger cause of TV.

Original languageEnglish
Title of host publication14th International Conference on Services Systems and Services Management, ICSSSM 2017 - Proceedings
EditorsXiaoqiang Cai, Jiafu Tang, Jian Chen
PublisherInstitute of Electrical and Electronics Engineers Inc.
ISBN (Electronic)9781509063697
DOIs
StatePublished - 28 Jul 2017
Event14th International Conference on Services Systems and Services Management, ICSSSM 2017 - Dalian, China
Duration: 16 Jun 201718 Jun 2017

Publication series

Name14th International Conference on Services Systems and Services Management, ICSSSM 2017 - Proceedings

Conference

Conference14th International Conference on Services Systems and Services Management, ICSSSM 2017
Country/TerritoryChina
CityDalian
Period16/06/1718/06/17

Keywords

  • Chinese online reviews
  • Multiple liner regression
  • Sentiment polarity
  • Text mining
  • VAR

Fingerprint

Dive into the research topics of 'Can extracted sentimental features from stock forum account for the stock return?'. Together they form a unique fingerprint.

Cite this