Style analytics: news impact on style investing

Abstract

Behavioral finance and its anomalies have been growing steadily since the 1990s. These various anomalies challenge the rules of traditional finance and consider the investor as irrational. Different characteristics affecting the investor’s behavior are constantly being studied, such as the effects of cognitive psychology or the impact of external information. This work aims to analyze most impactful textual features that affect the investor throughout his or her information processing. Through a survey on different investment styles, we collected information from selected news: the tone of the news, the author’s or speaker’s opinion(s) and recommendation(s), the trust in the news, the excitement felt by the readers after reading the news, their investment intention(s) and their risk considering these investment intention(s). Data analyses are then carried out via decision trees. Analyses of the first collected data suggest that the tone of the text and the recommendations of the various speakers are the two most impactful features in the investment process, followed by the author's excitement and opinions. In addition, we note that the authors of the articles also play their role in the investment process. Indeed, it appears that, the more homogeneous the information about an investment style is, the simpler is the tree. Another important part of this work consisted in textual augmentation. Based on the analyzed data, we would give labels to the different news included in the survey. In order to be able to apply these labels to other un-surveyed texts, we have developed algorithms to create, via a synonym’s generator, new news from the previous ones. Finally, we added a discussion to bring some criticisms and comments about our tools, possible improvements and suggestions for additional investigations.