When machines trade on corporate disclosures: Using text analytics for investment strategies

2023 | Decision Support Systems | Citations: 0

Authors: Schmitz, Hans Christian; Lutz, Bernhard; Wolff, Dominik; Neumann, Dirk

Abstract: Can you make profits by trading on corporate disclosures using machine learning? ... Expand

Abstract: Can you make profits by trading on corporate disclosures using machine learning? In this study, we aim to obtain a conservative estimate of profitability, while accounting for the combination of several important realworld aspects. Specifically, we consider the holistic research problem that combines model predictions based on the textual content of corporate disclosures and trading strategies while accounting for transaction costs, order clearance periods, post-publication returns, and liquidity filtering. Furthermore, we aim to understand how the resulting profits are influenced by different model and trading strategy parameters. Based on 354,992 form 8-K filings and 10,204 ad hoc announcements, we find that the proposed trading strategies yield up to 7.81 % and 9.34 % out-of-sample annualized return. In addition, our results suggest that machine learning models should be provided with additional features about prior disclosures, while being trained on the ternary prediction problem that allows for predictions of neutral market reactions. We complement our results with several sensitivity analyses that show how profitability is influenced by transaction costs, different ensemble sizes, return neutrality thresholds, and liquidity filtering. Ultimately, we provide useful insights for practitioners by describing how the machine learning models arrive at decisions in terms of Shapley Additive Explanations. Collapse

Topics: shapley additive explanation accounting telephone electronic mail

Methods: machine learning sensitivity analysis term frequency–inverse document frequency GloVe random forest classification

Theories: transaction cost economics game theory

A taxonomy for deriving business insights from user-generated content

2023 | European Conference On Information Systems | Citations: 0

Authors: Weingart, Pablo; Wambsganss, Thiemo; Soellner, Matthias

Abstract: Deriving business insights from user-generated content (UGC) is a widely invest ... Expand

Abstract: Deriving business insights from user-generated content (UGC) is a widely investigated phenomenon in information systems (IS) research. Due to its unstructured nature and technical constraints, UGC is still underutilized as a data source in research and practice. Using recent advancements in machine learning research, especially large language models (LLMs), IS researchers can possibly derive these insights more effectively. To guide and further understand the usage of these techniques, we develop a taxonomy that provides an overview of business insights derived from UGC. The taxonomy helps both practitioners and researchers identify, design, compare and evaluate the use of UGC in this IS context. Finally, we showcase an LLM-supported demo application that derives novel business insights and apply the taxonomy to it. In doing so, we show exemplary how LLMs can be used to develop new or extend existing natural language processing (NLP) applications in the realm of IS. Collapse

Topics: database system social network online review social media user-generated content

Methods: natural language processing structured literature research qualitative interview deep learning NLP application

Automated mortgage origination delay detection from textual conversations

2021 | Decision Support Systems | Citations: 5

Authors: Brahma, Arin; Goldberg, David M.; Zaman, Nohel; Aloiso, Mariano

Abstract: For modern mortgage firms, the process of setting up and verifying a new loan, k ... Expand

Abstract: For modern mortgage firms, the process of setting up and verifying a new loan, known as origination, is complex and multifaceted. The literature notes that this process is rife with delays that can stunt the firm’s business opportunities, but no modern analytical techniques have been developed to address the problem. In this paper, we suggest the use of text analytic and machine learning techniques to predict likely delays. In collaboration with a large national mortgage firm, we derive a large dataset of transcripts from employees’ communications per­ taining to potential loans. We first use information retrieval to generate an initial list of “seed terms,” or terms most associated with loans that were delayed. We then use an array of machine learning approaches to generate predictive models based upon these seed terms. We find that these approaches are comparable in performance to less interpretable state-of-the-art approaches utilizing word embeddings. The resultant models offer interpretable and high-performing solutions to mitigate the risk of delays through early risk detection. Collapse

Topics: business process management information retrieval Python innovation management decision making

Methods: machine learning natural language processing computational algorithm BERT lasso regression

Spline-rule ensemble classifiers with structured sparsity regularization for interpretable customer churn modeling

2021 | Decision Support Systems | Citations: 0

Authors: De Bock, Koen W.; De Caigny, Arno

Abstract: An important business domain that relies heavily on advanced statistical- and ma ... Expand

Abstract: An important business domain that relies heavily on advanced statistical- and machine learning algorithms to support operational decision-making is customer retention management. Customer churn prediction is a crucial tool to support customer retention. It allows an early identification of customers who are at risk to abandon the company and provides the ability to gain insights into why customers are at risk. Hence, customer churn pre­ diction models should complement predictive performance with model insights. Inspired by their ability to reconcile strong predictive performance and interpretability, this study introduces rule ensembles and their extension, spline-rule ensembles, as a promising family of classification algorithms to the customer churn pre­ diction domain. Spline-rule ensembles combine the flexibility of a tree-based ensemble classifier with the simplicity of regression analysis. They do, however, neglect the relatedness between potentially conflicting model components which can introduce unnecessary complexity in the models and compromises model inter­ pretability. To tackle this issue, a novel algorithmic extension, spline-rule ensembles with sparse group lasso regularization (SRE-SGL) is proposed to enhance interpretability through structured regularization. Experiments on fourteen real-world customer churn data sets in different industries (i) demonstrate the superior predictive performance of spline-rule ensembles with sparse group lasso over a set well yet powerful benchmark methods in terms of AUC and top decile lift; (ii) show that spline-rule ensembles with sparse group lasso regularization significantly outperform conventional rule ensembles whilst performing at least as well as conventional splinerule ensembles; and (iii) illustrate the interpretable nature of a spline-rule ensemble model and the advantage of structured regularization in SRE-SGL by means of a case study on customer churn prediction for a telecommu­ nications company. Collapse

Topics: partial dependence plot logistics management marketing intelligence decision making missing data

Methods: computational algorithm machine learning case study experiment logistic regression

FinTech as a Game Changer: Overview of Research Frontiers

2021 | Information Systems Research | Citations: 25

Authors: Hendershott, Terrence; Zhang, Xiaoquan (Michael); Zhao, J. Leon; Zheng, Zhiqiang (Eric)

Abstract: Technologies have spawned finance innovations since the early days of computer a ... Expand

Abstract: Technologies have spawned finance innovations since the early days of computer applications in businesses, most recently reaching the stage of disruptive innovations, such as mobile payments, cryptocurrencies, and digitization of business assets. This has led to the emerging field called financial technology or simply FinTech. In this editorial review, we first provide an overview on relevant technological, pedagogical, and managerial issues pertaining to FinTech teaching and research, with a focus on market trading, artificial intelligence, and blockchain in finance. And then we introduce the articles appearing in this special section. We hope that our discussions of potential research directions and topics in FinTech will stimulate future research in the fields of information systems and finance toward making their unique marks in the FinTech evolution and the associated business and societal innovations. Collapse

Topics: electronic finance artificial intelligence mobile system fake news blockchain

Methods: linear regression analysis machine learning word embedding classification method ordinary least square

Stratifying no-show patients into multiple risk groups via a holistic data analytics-based framework

2020 | Decision Support Systems | Citations: 0

Authors: Simsek, Serhat; Tiahrt, Thomas; Dag, Ali

Abstract: Accurate prediction of no-show patients plays a crucial role as it enables resea ... Expand

Abstract: Accurate prediction of no-show patients plays a crucial role as it enables researchers to increase the efficiency of their scheduling systems. The purpose of the current study is to formulate a novel hybrid data mining-based methodology to a) accurately predict the no-show patients, b) build a parsimonious model by employing a comprehensive variable selection procedure, c) build a model that does not suffer due to data imbalance, and d) provide healthcare agencies with a patient-specific risk level. Our study suggests that an Artificial Neural Network (ANN) model should be employed as a classification algorithm in predicting patient no-shows by using the variable set that is commonly selected by a Genetic Algorithm (GA) and Simulated Annealing (SA). In addition, we used Random Under Sampling (RUS) to improve the performance of the model in predicting the minority group (no-show) patients. The patient-specific risk scores were justified by applying a threshold sensitivity analysis. Also, the web-based decision support tool that can be adopted by clinics is developed. The clinics can incorporate their own intuition/incentive to make the final decision on the cases where the model is not confident enough (i.e. when the estimated probabilities fall near the decision boundary). These insights enable health care professionals to improve clinic utilization and patient outcomes. Collapse

Topics: decision support system analytical information system programming language decision making website

Methods: computational algorithm machine learning sensitivity analysis artificial neural network optimization model

LIFE-INTEGRATED STRESS ASSESSMENT

2019 | European Conference On Information Systems | Citations: 0

Authors: Gimpel, Henner; Regal, Christian; Schmidt, Marco

Abstract: Stress is an individual, societal, and economical problem and the main cause of ... Expand

Abstract: Stress is an individual, societal, and economical problem and the main cause of psychological and physiological illnesses. First efforts towards technological support of stress management and stressadaptive IT systems use smartphone data to assess stress, enabled by the vast amount of data produced by smartphones’ rich sensing capabilities. However, these systems require the user’s active cooperation. We aim at life-integrated, continuous, smartphone-based stress assessment, which does not influence the user’s daily habits. The proposed system uses hardware and software sensors to collect data on the user’s behavior and environmental context. A prototype demonstrates the general feasibility of lifeintegrated mobile stress sensing and collects real-life evidence within a public field study, but also highlights that efficient resource consumption and privacy are even more important for unobtrusive applications. The evaluation of collected data yields a person-independent regression model that explains approximately 41 % of the variance in stress. A regression model built with the same data predicts perceived stress with an accuracy of 81 %. These results pave the way for the design of adaptive systems, which continuously sense the individual’s mental state and execute regulating measures to better fit the user’s needs. Collapse

Topics: smartphone usability personalization Android critical success factor

Methods: design artifact field study regression analysis method survey design science

Generating Business Intelligence Through Social Media Analytics: Measuring Brand Personality with Consumer-, Employee-, and Firm-Generated Content

2019 | Journal of Management Information Systems | Citations: 0

Authors: Hu, Yuheng; Xu, Anbang; Hong, Yili; Gal, David; Sinha, Vibha; Akkiraju, Rama

Abstract: Social media platforms provide an enormous public repository of textual data fro ... Expand

Abstract: Social media platforms provide an enormous public repository of textual data from which valuable information can be extracted. We show that firms can extract business intelligence from social media data bearing on an important business application, measuring brand personality. Specifically, we develop a text analytics framework that integrates different distinct sources of social media data generated by consumers, employees, and firms, to measure brand personality. Based on Elastic-Net regression analyses of a large corpus of social media data, including self-descriptions of 1,996,214 consumers who followed the sample of brands on social media, 312,400 employee reviews of the brands' firms, and 680,056 brand official tweets, we develop a brand personality model that achieves prediction accuracy as high as 0.78. Among key insights, we find that the profile of individuals who choose to associate with brands on social media is an important predictor of brand personality; this provides the first real-world evidence for a consumer identity-brand personality link. We also identify a link between an organization's internal corporate environment as perceived by employees and brand personality as judged by consumers. We further illuminate the practical implication of our predictive model by building a cloud-based information system that allows managers and analysts to explore and track personality of their own brands and their competitors' brands. Collapse

Topics: personality Twitter social media social media analytics cloud computing

Methods: survey elastic net regression product review natural language processing descriptive statistic

Mining Online Reviews to Uncover Consumer Brand Engagement

2019 | International Conference on Information Systems | Citations: 0

Authors: Kulkarni, Uday; Deokar, Amit; Ajjan, Haya

Abstract: We build on the theoretical foundations of consumer engagement from the marketin ... Expand

Abstract: We build on the theoretical foundations of consumer engagement from the marketing literature to propose a novel way to measure consumer brand engagement (CBE) using machine learning and natural language processing of consumer-generated online content. We conceptualize customer-written product reviews as more than just eWOM influencing purchase decisions, but as indicators of CBE. Our method is operationalized through a general-purpose artifact that allows continuous, time-variant, and flexible measurement of CBE. We demonstrate the feasibility of our approach through a large dataset of product reviews of multiple brands of a Fortune 500 garment retailer. Our contribution has implications for research, in that, it creates an opportunity to investigate the antecedent and consequent relationships between CBE and other critical marketing constructs such as intention to purchase and customer loyalty. Further, the ability to measure the time-variant nature of CBE allows for testing leading and lagging relationships between CBE and other key business indicators. Collapse

Topics: online review marketing management online word of mouth word of mouth analytical information system

Methods: machine learning product review natural language processing deep learning regression analysis method

A for Effort? Using the Crowd to Identify Moral Hazard in New York City Restaurant Hygiene Inspections

2019 | Information Systems Research | Citations: 17

Authors: Mejia, Jorge; Mankad, Shawn; Gopal, Anandasivam

Abstract: From an upset stomach to a life-threatening foodborne illness, getting sick is a ... Expand

Abstract: From an upset stomach to a life-threatening foodborne illness, getting sick is all too common after eating in restaurants. Although health inspection programs are designed to protect consumers, such inspections typically occur at wide intervals of time, allowing restaurant hygiene to remain unmonitored in the interim periods. Information provided in online reviews may be effectively used in these interim periods to gauge restaurant hygiene. In this paper, we use textual information from online reviews of restaurants to effectively identify cases of hygiene violations in restaurants, even after the restaurants have been inspected and certified, thereby identifying moral hazard. Using a data set of restaurant hygiene inspections in New York City from 2010 through 2016 and the associated set of online reviews for the same set of restaurants from Yelp, we use supervised machine learning techniques to develop a hygiene dictionary specifically crafted to identify hygiene-related concerns. The use of this dictionary and the related word counts in online reviews allows us to identify systematic instances of moral hazard, wherein restaurants with positive hygiene inspection scores are seen to regress in their hygiene maintenance within 90 days of receiving the inspection scores. To the extent that social media provide some visibility into the hygiene practices of restaurants, we argue that the effects of information asymmetry that lead to moral hazard may be partially mitigated in this context. Based on our work, we also provide strategies for how cities and policymakers may design effective restaurant inspection programs, through a combination of traditional inspections and the appropriate use of social media. Collapse

Topics: online review moral hazard social media user-generated content crowdsourcing

Methods: machine learning regression analysis method longitudinal research naïve bayes parametric test

Theories: lemon market theory