Generating Domain-Specific Dictionaries Using Bayesian Learning

Pröllochs, Nicolas; Feuerriegel, Stefan; Neumann, Dirk

doi:10.2139/ssrn.2522884

Cited by 28 publications

(48 citation statements)

References 25 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Nevertheless, practitioners might prefer alternatives that further extend linguistic terms t with a metric w t discriminating the relative importance (e. g. "risk" may be a stronger indicator of risk than "may"). The benefit of our approach is that it can be easily replaced by such a scoring mechanism, i. e. changing R d to t w t tf t /#total words and O d analogously [51]. For reasons of reproducibility and ease-of-use, we demonstrate the outcomes for a simple categorization without weighting in the following.…”

Section: Language Analysismentioning

confidence: 99%

Business analytics for strategic management: Identifying and assessing corporate challenges via topic modeling

Pröllochs

Feuerriegel

2020

Information & Management

Self Cite

View full text Add to dashboard Cite

Strategic management requires an assessment of a firm's internal and external environment. Our work extends the body of management tools (e.g. SWOT analysis or growth-share matrix) by proposing an automated text mining framework. Here we draw upon narrative materials from firms (e.g. financial disclosures) and perform topic modeling in order to identify the key issues faced by an organization. We then quantify the use of language along two dimensions: risk and optimism. This reveals a firm's strengths and weaknesses by identifying business units, activities and processes subject to risk, while also comparing it to competitors or the market.

show abstract

Section: Language Analysismentioning

confidence: 99%

Business analytics for strategic management: Identifying and assessing corporate challenges via topic modeling

Pröllochs

Feuerriegel

2020

Information & Management

Self Cite

View full text Add to dashboard Cite

show abstract

“…(Varazzo, 2011). Text processing is done using the Natural Language Toolkit (Bird et al, 2009) and the package sentimentAnalysis (Pröllochs et al, 2015) for the statistical programming language R (R Core Team, 2016). All machine learning models are implemented using scikit-learn (Pedregosa et al, 2011).…”

Section: Data Preparation and Consolidationmentioning

confidence: 99%

Separating the signal from the noise – Financial machine learning for Twitter

Schnaubelt

Fischer

Krauß

2020

Journal of Economic Dynamics and Control

View full text Add to dashboard Cite

Most statistical arbitrage strategies in the academic literature soley rely on price time series. By contrast, alternative data sources are of growing importance for professional investors. We contribute to bridging this gap by assessing the price-predictive value of more than nine million tweets on intraday returns of the S&P 500 constituents. For this purpose, we design a machine learning pipeline addressing specific challenges inherent to this task. At first, we engineer domain-specific features along three categories, i.e., directional indicators, relevance indicators and meta features. Next, we leverage a random forest to extract the relationship between these features and subsequent stock returns in a low signal-to-noise setting. For performance evaluation, we run a rigorous eventbased backtesting study across all tweets and stocks. We find annualized returns of 6.4 percent and a Sharpe ratio of 2.2 after transaction costs. Finally, we illuminate the machine learning black box and unveil sources of profitability: First, results are both driven and limited by the temporal clustering of tweets, i.e., the majority of profits stem from tweets clustered closely together in time, corresponding to high-event situations. Second, the importance of included features follows an economic rationale, e.g., tweets with positive sentiment tend to yield positive returns and vice versa. Third, we find that stocks of medium market capitalization and from the consumer and technology sectors contribute most to our results, which we interpret as a trade-off between tweet coverage and tweet relevance.

show abstract

“…A viable trade-off is commonly presented by drawing upon sentiment dictionaries as a form of feature engineering. Here the idea is to incorporate domain knowledge in the form of predefined dictionaries that label terms into different semantic categories [44]. The conventional assumption is that the overall sentiment hints the economic outlook [45,46,7].…”

Section: Sentiment-based Machine Learningmentioning

confidence: 99%

Long-term stock index forecasting based on text mining of regulatory disclosures

Feuerriegel

Gordon

2018

Decision Support Systems

Self Cite

View full text Add to dashboard Cite

Share valuations are known to adjust to new information entering the market, such as regulatory disclosures. We study whether the language of such news items can improve short-term and especially long-term (24 months) forecasts of stock indices.For this purpose, this work utilizes predictive models suited to high-dimensional data and specifically compares techniques for data-driven and knowledge-driven dimensionality reduction in order to avoid overfitting. Our experiments, based on 75,927 ad hoc announcements from 1996-2016, reveal the following results: in the long run, text-based models succeed in reducing forecast errors below baseline predictions from historic lags at a statistically significant level. Our research provides implications to business applications of decision-support in financial markets, especially given the growing prevalence of index ETFs (exchange traded funds). reflect publicly-available information. Based on this premise, one can expect price changes whenever new information enters the market. In practice, regulations ensure that stock-relevant information is revealed primarily via regulatory disclosures in order to provide equal access for all market participants. Such materials disclose, for instance, quarterly earnings, but also management changes, legal risks and other events deemed important [2]. Accordingly, financial disclosures present an alluring and potentially financially-rewarding means of forecasting changes in stock valuations [3].In this respect, corporate news conveys a broad spectrum of information concerning the past performance and current challenges of the business [4], as well as frequently hinting at the future outlook. Research has followed this reasoning and empirically quantified the impact of the narrative content on the subsequent stock market responses [cf. 5, 6, 7]. Moreover, researchers have also demonstrated the prognostic capability of financial disclosures with respect to individual stock market returns in the short term [e. g. 8, 9, 10]. Accordingly, news-based forecasting has received considerable traction and, as a result, various publications have evaluated different news datasets, forecasted indicator/markets, preprocessing operations from the field of natural language processing and forecasting algorithms.Here we refer to the literature, which provides a thorough overview [3].Forecasting the development of stock indices is highly demanded by multiple stakeholders in financial markets. The underlying reason is that households are investing their money no only in individual stocks, government bonds or savings accounts; rather, they increasingly prefer exchange-traded funds (ETFs). These ETFs replicate the movements of marketable securities, with stock indices being the most prominent example. As part of their benefits, ETFs are traded on stock exchanges but often with higher liquidity and lower fees. Hence, private investors demand for decision support in better understanding the development of markets, as well as for obtaining prognostic support. For ...

show abstract

Generating Domain-Specific Dictionaries Using Bayesian Learning

Cited by 28 publications

References 25 publications

Business analytics for strategic management: Identifying and assessing corporate challenges via topic modeling

Business analytics for strategic management: Identifying and assessing corporate challenges via topic modeling

Separating the signal from the noise – Financial machine learning for Twitter

Long-term stock index forecasting based on text mining of regulatory disclosures

Contact Info

Product

Resources

About