eInformation: a clinical study of investor discussion and sentiment

We examine the information stream for four stocks over seven month to trace the relationship between on-line discussion, freshs activity, and stock price motions On-line discussions support numerous unsubstantiated rumors, substantial on-point exchanges, and quick dissemination of imminent and not long ago released information. Applying language-processing routines to message board postings and novels we create sentiment and disagreement measures or "eInformation." We analyze the determinants of sentiment and disagreement, and trace links between novels eInformation, and stock returns. This intensive clinical research of on-line discussions suggests mechanisms individual investors and assemblages can use to analyze and digest company information.

**********

In light of the large material substance of research on informationally efficient markets, there present the appearances little left to learn from the continued empirical examination of information and markets. It would appear similarly pointless for individual investors to make trial of to compete with professional analysts. However, understanding the individuals' investing decisions has been individual of the most vibrant research streams in new years (see Roll, 1986, Odean, 1998 Gervais and Odean, 2001 and Barber and Odean, 2001)

Technology, in the form of stock chat message boards, now provides a novel real-time window into discussions by dint of individual investors. It is instructive to peek [i]or[/i] part of to the other this window to observe in what way information is digested, how sentiment opens and how perceptions are related to prices.

The way we adopt in this article is to use a clinical, i.e., small sample, approach to understanding investor behavior. Before framing hypotheses or constructing trials it is important to establish a base horizontal of understanding in an area. Thus, our article is decidedly descriptive, part of a lengthy inductive tradition in economics (Blaug, 1992) We do not attempt to either affirm or cast off theory. Rather, we suggest a series of working surmises (or hypotheses) that can be disentangleed through subsequent model building and large-scale empirical study

We have three goals in this article. First, we closely analyze the race who share their opinions (posters) and their discussions surrounding a scarcely any stocks. Given the anonymous nature of this activity, we instead pick to study an outlier by means of interviewing an extensive poster. Doing with equal reason enables us to understand for what cause [i]or[/i] reason someone would spend substantial amounts of time posting messages to single of the boards we study

As part of our analysis, we also focus attention upon the discussions themselves. Although there is a perception that postings are "garbage," to the contrary, discussions sustain on-point exchanges, generate possibly non-public information, quickly disseminate public information from freshs stories, and serve as forums where investors can extract meaning from information. Chat compasss and postings are also sources of numerous unsubstantiated rumors, adding noise to the information run Nevertheless, the fact that plane some nonpublic information may be released upon the boards--and the observation that hand-bills use the boards to experiment their own analyses and obtain those of others--may explain wherefore posters and surfers continue to oft-repeated these chat board sites.

next to the first using language-processing algorithms, we measure the intensity and dispersion of sentiment (which we denominate eInformation) for over 170,000 messages columned about four stocks. We analyze the determinants of the horizontal of sentiment and disagreement among bills and find that there is a shut up relationship between sentiment levels, stock prices, and trading turn We also find that disagreement is related to the intensity of discussion.

Finally, we explore the usefulness of squeeze outed investor sentiment (eInformation) to predict stock go [i]or[/i] come backs Our clinical study confirms other studies that fail to find predictive power forecasting go [i]or[/i] come backs (Antweiler and Frank, 2002, 2004 and Das and Chen, 2003)

The article go aheads as follows. Section I deals with our clinical design. In Section II, we discuss the demographics of broadsides detailing our interview with an especially active investor-discussant. Section III reports upon our clinical examination of the nature of the discussions and the quality of information in those discussions. Section IV describes our computer-generated measures of sentiment and disagreement (eInformation) that are extracted using language-processing algorithms. In Section V we analyze the determinants of our eInformation measures. In Section VI, we examine the relationship of eInformation to the price formation proces Finally, in Section VIII, we summarize the hypotheses that rise from this clinical investigation.

I. Sample Design

We research four firms over a period of seven month We use these four firms as archetypes for different information environments where traditional and fresh eInformation flows vary. As befits a clinical close attention we attempt to dig far down into these four firms, using our observations to derive hypotheses for large-scale studies. We have deliberately not picked pathological examples where posters have used stock message boards to explicitly manipulate prices (Leinweber and Madhavan, 2001)