Topic Tones of Analyst Reports and Stock Returns: A Deep Learning Approach
45 Pages Posted: 20 Aug 2018 Last revised: 20 May 2021
Date Written: August 8, 2018
We apply a deep neural network supervised learning (DNN) approach to extract text topics from analyst reports based on whether the topics are used to justify the quantitative numbers (justification type), such as the target price, or not (qualitative type). A baseline model without using text information has an adjusted R squared of 2.3% in predicting the cumulative two-day abnormal returns. When we include the topic tones, the adjusted R squared increases to 15.4%. This significant increase of R squared is mainly driven by the qualitative type topics and not much driven by the justification type topics.
Keywords: Textual Analysis; DNN Approach; Topic Tones; Information Content
JEL Classification: C89, G11, G12, G14
Suggested Citation: Suggested Citation