Measuring Effects on Stock Returns of Sentiment Indexes Created From Stock Message Boards

Abstract

Various techniques and sources of information exist to aid investors in predicting future stock returns. However, no effective proxy for retail investors, such as stock message board users, has been established. This study provides guidelines for creating an effective proxy. The heart of such proxies is sentiment indexes, and in the past the indexes have had low predictive power. Introducing four methodological improvements for applying text classifiers and two probability measurements, we contrast eight widely applied text classifiers to stock message board data. Based on the classifier results and incorporating our new methods, the new sentiment index proves to be a significant "same-day positive but next-day negative" directional indicator.JEL Classification: G11, G12

Time-variation in the impact of news sentiment

“…2.1 News sentiment and contextual analysis Antweiler and Frank (2004) were the first to apply contextual analysis and develop news sentiment measures to understand stock returns; using a Naïve Bayes algorithm to assign trading signals on the basis of messages posted to internet message boards they find that while such messages are able to predict market volatility, their effect on stock returns while statistically significant is economically small. In a similar context, Zhang et al (2012) incorporate several methodological improvements in order to create news sentiment indices that are significant directional indicators. Tetlock (2007) undertakes a different approach that has become more prevalent in the literature; classifying words on the basis of categories from the Harvard psychosocial dictionary, he assigns quantitative scores to content in the Wall Street Journal's "Abreast of the Market" column and reports that high levels of media pessimism predict declining market prices which are followed by price reversals.…”

Influence of Bloomberg’s Investor Sentiment Index: Evidence from European Union Financial Sector

“…At the same time as the investor sentiment indexes, the literature has developed that attempts to explain return on assets through textual analysis of the news [13,68]. There is no clear evidence of its explanatory capacity, since there are papers that argue it has greater potential than sentiment indexes [39,[69][70][71] but we also find papers arguing otherwise, as a consequence of the different linguistic perception of each investor, the market where the news is from, the asymmetry between words with negative and positive connotations, the language in which the news is given and the analysis of words out of context [14,19,32,33,35,36,62,66,71,72].…”

“…In summary, based on the literature reviewed, we can group the empirical studies together by two fundamental characteristics: on the one hand, those papers that do not consider an asset valuation model to measure the relationship between investor sentiment and market returns of assets [12,17,28,37,79,80], versus those who do [7,9,11,62,64,66,67,81,82]; and on the other, studies that develop their own sentiment indexes [11,35,41,71,82] versus those using indexes developed by specialized investors or economic agents [9,36,64,[82][83][84].…”

Private investor extrapolation bias – evidence through qualitative content analysis (QCA)

“…As QCA facilitates the manual coding of textual data that require a certain degree of interpretation (Schreier, 2012), this qualitative data collection method is appropriate for coding message board posts. Message board posts have been suggested to imply private investor sentiment (Antweiler and Frank, 2004; Das et al , 2005; Das and Chen, 2007; Zhang and Swanson, 2010; Zhang et al , 2012). The susceptibility of private investors to extrapolation bias is examined by testing whether after-hours sentiment is positively predicted by daily share return.…”