1,721,005 research outputs found
What do they mean? Using Media Richness as an Indicator for the Information Value of Stock Analyst Opinion regarding post-earnings Firm Performance
Full text linkIn this research the impact of media-richness on the investor reaction to earnings announcements is investigated. To this end, unstructured (high-richness) sources of analyst opinion are subjected to text-mining and combined with structured (low-richness) sources of analyst opinion, as well as other commonly used structured data relevant to company performance. Results indicate that equivocality is a major problem faced by investors, while uncertainty as understood by media-richness theory appears to be less dominant
Topic Modelling Methodology: Its Use in Information Systems and other Managerial Disciplines
No full textOver the last decade, quantitative text mining approaches to content analysis have gained increasing traction within information systems research, and related fields, such as business administration. Recently, topic models, which are supposed to provide their user with an overview of themes being dis-cussed in documents, have gained popularity. However, while convenient tools for the creation of this model class exist, the evaluation of topic models poses significant challenges to their users. In this research, we investigate how questions of model validity and trustworthiness of presented analyses are addressed across disciplines. We accomplish this by providing a structured review of methodological approaches across the Financial Times 50 journal ranking. We identify 59 methodological research papers, 24 implementations of topic models, as well as 33 research papers using topic models in In-formation Systems (IS) research, and 29 papers using such models in other managerial disciplines. Results indicate a need for model implementations usable by a wider audience, as well as the need for more implementations of model validation techniques, and the need for a discussion about the theoretical foundations of topic modelling based research
They Talk but What do they Listen to? Analyzing Financial Analysts' Information Processing using Latent Dirichlet Allocation
No full textIn this study, we examine stock analyst information processing behaviour on the example of information transfer between analyst conference calls and analyst reports. From a theoretical perspective, the study contributes to an understanding of analysts’ recommendation biases resulting from their information processing. It provides new insights on how information is actually used by analysts, while practical implications for both sides of conference calls and other market participants are examined. Results indicate that analysts are exposed to new information during conference call events, which they conse-quently incorporate in their reporting
They Talk but What do they Listen to? Analyzing Financial Analysts' Information Processing using Latent Dirichlet Allocation
No full textIn this study, we examine stock analyst information processing behaviour on the example of information transfer between analyst conference calls and analyst reports. From a theoretical perspective, the study contributes to an understanding of analysts’ recommendation biases resulting from their information processing. It provides new insights on how information is actually used by analysts, while practical implications for both sides of conference calls and other market participants are examined. Results indicate that analysts are exposed to new information during conference call events, which they conse-quently incorporate in their reporting
Supporting financial Decisions by identifying relevant Conference Call Topics
No full textThe ever rising amount of business communications results in a growing amount of qualitative data relevant to many decision situations. This increase in information volume and velocity threatens to overburden decision makers. We provide a structured approach towards this problem using topicmodels to reduce information overload by filtering content and by providing context-relevant information to decision makers. Building upon theoretical considerations related to phases of the decision process established by Herbert A. Simon, we implement the proposed approach on the example of a large document collection of stock analyst reports and analyst conference calls using Latent Dirichlet Allocation (a topic model). Thereby, we extract investment-relevant topics from the model and discuss the opportunities for decision support resulting from the chosen approach
How to conquer information overload? Supporting financial decisions by identifying relevant conference call topics
No full textThe ever rising amount of business communications results in a growing amount of qualitative data relevant to many decision situations. This increase in information volume and velocity threatens to overburden decision makers. We provide a structured approach towards this problem using topicmodels to reduce information overload by filtering content and by providing context-relevant information to decision makers. Building upon theoretical considerations related to phases of the decision process established by Herbert A. Simon, we implement the proposed approach on the example of a large document collection of stock analyst reports and analyst conference calls using Latent Dirichlet Allocation (a topic model). Thereby, we extract investment-relevant topics from the model and discuss the opportunities for decision support resulting from the chosen approach
Mutual prediction and the drivers of crowd wisdom
No full textWe examine the drivers of crowd wisdom in the financial domain by relating analyst report and social media sentiment via Granger causality (GC) testing based on the wisdom of crowds (WoC) theory. The significance of a large number of the tested time series indicates that analyst reports and social media content are suitable for mutual prediction. We elaborate on the conditions under which crowd cognitive diversity matters, and we derive related measures. The results suggest that the WoC theory can partially explain the GC between the two media types and that both professional analysts and the crowd can outperform one another under favorable circumstances
Topic Modelling Methodology: Its Use in Information Systems and other Managerial Disciplines
No full textOver the last decade, quantitative text mining approaches to content analysis have gained increasing traction within information systems research, and related fields, such as business administration. Recently, topic models, which are supposed to provide their user with an overview of themes being dis-cussed in documents, have gained popularity. However, while convenient tools for the creation of this model class exist, the evaluation of topic models poses significant challenges to their users. In this research, we investigate how questions of model validity and trustworthiness of presented analyses are addressed across disciplines. We accomplish this by providing a structured review of methodological approaches across the Financial Times 50 journal ranking. We identify 59 methodological research papers, 24 implementations of topic models, as well as 33 research papers using topic models in In-formation Systems (IS) research, and 29 papers using such models in other managerial disciplines. Results indicate a need for model implementations usable by a wider audience, as well as the need for more implementations of model validation techniques, and the need for a discussion about the theoretical foundations of topic modelling based research
A Study on Mutual Prediction
No full textThe predictive power of stock analyst reports has been used to relate report contents to stock returns or describe herding behavior of analysts themselves. In this paper, the sentiment of analyst reports is related to that of a large social media data set via Granger Causality testing on the basis of wisdom of crowds theory based considerations, in order to investigate whether the two types of content are inherently related or not. Results show strong significance for a large number of the tested time series, indicating that the two types of content are indeed suitable for mutual prediction. In addition, we elaborate on the conditions under which cognitive diversity of the crowd matters. Furthermore, a second analysis stage provides evidence for which type of company and news environment a particular direction of granger cause arises between the two types of content
A Study on Mutual Prediction
No full textThe predictive power of stock analyst reports has been used to relate report contents to stock returns or describe herding behavior of analysts themselves. In this paper, the sentiment of analyst reports is related to that of a large social media data set via Granger Causality testing on the basis of wisdom of crowds theory based considerations, in order to investigate whether the two types of content are inherently related or not. Results show strong significance for a large number of the tested time series, indicating that the two types of content are indeed suitable for mutual prediction. In addition, we elaborate on the conditions under which cognitive diversity of the crowd matters. Furthermore, a second analysis stage provides evidence for which type of company and news environment a particular direction of granger cause arises between the two types of content
- …
