Clustering News to Create Sentiment Indexes that Help Predict Stock Prices

• Item type: SIG Technical Reports(1)
• 公開日: 2017-07-10
• タイトル: Clustering News to Create Sentiment Indexes that Help Predict Stock Prices
• タイトル / 言語: en
• タイトル: Clustering News to Create Sentiment Indexes that Help Predict Stock Prices
• 言語: eng
• 資源タイプ / 資源タイプ識別子: http://purl.org/coar/resource_type/c_18gh
• 資源タイプ: technical report
• 著者所属: Chuo University
• 著者所属: CyberAgent, Inc.
• 著者所属(英): en
• 著者所属(英): Chuo University
• 著者所属(英): en
• 著者所属(英): CyberAgent, Inc.
• 著者名: Hiroshi, Ishijima; Takuro, Kazumi
• 著者名(英): Hiroshi, Ishijima; Takuro, Kazumi
• 論文抄録 / 内容記述タイプ: Other
• 論文抄録 / 内容記述: The purpose of this paper is to quantify the market sentiment as three indexes and examine whether they can help predict Japanese stock prices. Sentiment analysis is gaining increasing interest in both academia and business. Along these lines, Ishijima et al. (2014) created a sentiment index that quantifies the positive or negative emotion that might appear in entire headlines and articles of the Nikkei which is the most popular business newspaper in Japan. They concluded that the sentiment index significantly predicts Japanese stock prices three days in advance. We re-examine their results by suggesting a new sentiment index quantified from the headlines and articles limited to the economy-and-business news. To the best of our knowledge, this is the first paper that applies Latent Dirichlet Allocation (LDA) to cluster the Nikkei on an eight-year daily basis. We then explore the implication on how the new sentiment index can help explain Japanese stock prices. Our findings are three-fold: (i) Sentiment index created from the headlines and articles limited to the economy-and-business news significantly allows us to predict the Nikkei 225 and the market trading volume of the next business day. (ii) We cannot observe the return reversal referred to in the literature. (iii) Sentiment index will follow Japanese stock prices.
• 論文抄録(英) / 内容記述タイプ: Other
• 論文抄録(英) / 内容記述: The purpose of this paper is to quantify the market sentiment as three indexes and examine whether they can help predict Japanese stock prices. Sentiment analysis is gaining increasing interest in both academia and business. Along these lines, Ishijima et al. (2014) created a sentiment index that quantifies the positive or negative emotion that might appear in entire headlines and articles of the Nikkei which is the most popular business newspaper in Japan. They concluded that the sentiment index significantly predicts Japanese stock prices three days in advance. We re-examine their results by suggesting a new sentiment index quantified from the headlines and articles limited to the economy-and-business news. To the best of our knowledge, this is the first paper that applies Latent Dirichlet Allocation (LDA) to cluster the Nikkei on an eight-year daily basis. We then explore the implication on how the new sentiment index can help explain Japanese stock prices. Our findings are three-fold: (i) Sentiment index created from the headlines and articles limited to the economy-and-business news significantly allows us to predict the Nikkei 225 and the market trading volume of the next business day. (ii) We cannot observe the return reversal referred to in the literature. (iii) Sentiment index will follow Japanese stock prices.
• 書誌レコードID / 収録物識別子タイプ: NCID
• 書誌レコードID / 収録物識別子: AN10505667
• 書誌情報: 研究報告数理モデル化と問題解決（MPS）; 巻 2017-MPS-114, 号 10, p. 1-4, 発行日 2017-07-10
• ISSN / 収録物識別子タイプ: ISSN
• ISSN / 収録物識別子: 2188-8833
• Notice: SIG Technical Reports are nonrefereed and hence may later appear in any journals, conferences, symposia, etc.
• 出版者 / 言語: ja
• 出版者: 情報処理学会