LSTM-ReGAT: A network-centric approach for cryptocurrency price trend prediction

2023 | Decision Support Systems | Citations: 0

Authors: Zhong, Chao; Du, Wei; Xu, Wei; Huang, Qianhui; Zhao, Yinuo; Wang, Mingming

Abstract: Predicting price trends of cryptocurrencies is a challenging task due to the hig ... Expand

Abstract: Predicting price trends of cryptocurrencies is a challenging task due to the highly speculative cryptocurrency market. Prior studies mainly investigate predictors such as historical trading data, macroeconomic development, and public interests in cryptocurrencies for price trend prediction while ignoring the predictive role of the re­ lations between cryptocurrencies. In fact, the price movement of a cryptocurrency may be affected by those of other cryptocurrencies, thus, incorporating cryptocurrency interrelations can further improve the prediction performance. In this paper, we propose a novel end-to-end model for price trend prediction using long short-term memory (LSTM) and relationwise graph attention network (ReGAT), in which both individual cryptocurrency features and cryptocurrency relations are considered. Specifically, a cryptocurrency network is built by using shared features between cryptocurrencies. Further, LSTM is used to profile sequential patterns of individual cryptocurrency features. To fully extract network features, ReGAT is designed to aggregate information conveyed by different types of relations while automatically differentiating their importance. Finally, the conventional and network features are concatenated to predict the price trend. The effectiveness of LSTM-ReGAT is validated using real-world cryptocurrency market data. The trading simulations for Bitcoin and portfolios reveal that our model obtains the highest profits. Our study provides insightful implications for investment decision support in cryp­ tocurrency market. Collapse

Topics: cryptocurrency Bitcoin database system Google+ information system use

Methods: machine learning time series analysis longitudinal research simulation support vector machine

Perspective—Discovery Within Validation Logic: Deliberately Surfacing, Complementing, and Substituting Abductive Reasoning in Hypothetico-Deductive Inquiry

2018 | Organization Science | Citations: 0

Authors: Behfar, Kristin; Okhuysen, Gerardo A.

Abstract: We propose a more explicit role for abductive reasoning, or the development of i ... Expand

Abstract: We propose a more explicit role for abductive reasoning, or the development of initial explanation, in hypothetico-deductive (H-D) inquiry. We begin by describing the roots of abduction in pragmatism and its role in exploration and discovery. Recognizing that pragmatism treats abductive reasoning as inevitable, we argue that it can also be a deliberate form of reasoning in scientific inquiry, articulating the unique place it can have in hypothetico-deductive theorizing. We explain the opportunities from surfacing abductive reasoning in H-D where it already exists; from explicitly acknowledging abductive reasoning as a complement in building logical chains in H-D; and from using abductive reasoning as a substitute for H-D logic when a body of knowledge exhibits inconsistent, contradictory, or discrepant results. We elaborate strategies for data search and selection, data production and compilation, and analytical corroboration. Our overall argument is that the deliberate use of abductive reasoning in hypothetico-deductive projects has distinct advantages stemming from an explicitly tight connection between data and theory. We end by explaining the benefits of actively recognizing the role of abductive reasoning in organizational and management theorizing. The article was written and prepared by the U.S. goverenment employee(s) on official time and is therefore in the public domain. Collapse

Topics: IT career

Methods: theory development mixed method data dredging post-hoc analysis experiment

Theories: theory of abduction organizational behavior theory

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

2018 | Decision Support Systems | Citations: 0

Authors: Feuerriegel, Stefan; Gordon, Julius

Abstract: Share valuations are known to adjust to new information entering the market, suc ... Expand

Abstract: 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). Collapse

Topics: decision support electronic mail decision making decision support system electronic trading

Methods: experiment machine learning dimensionality reduction data transformation time series analysis

A Tree-Based Approach for Addressing Self-Selection in Impact Studies with Big Data

2016 | Management Information Systems Quarterly | Citations: 0

Authors: Yahav, Inbal; Shmueli, Galit; Mani, Deepa

Abstract: In this paper, we introduce a tree-based approach adjusting for observable self- ... Expand

Abstract: In this paper, we introduce a tree-based approach adjusting for observable self-selection bias in intervention studies in management research. In contrast to traditional propensity score (PS) matching methods, including those using classification trees as a subcomponent, our tree-based approach provides a standalone, automated, data-driven methodology that allows for (1) the examination of nascent interventions whose selection is difficult and costly to theoretically specify a priori, (2) detection of heterogeneous intervention effects for different preintervention profiles, (3) identification of pre-intervention variables that correlate with the self-selected intervention, and (4) visual presentation of intervention effects that is easy to discern and understand. As such, the tree-based approach is a useful tool for analyzing observational impact studies as well as for post-analysis of experimental data. The tree-based approach is particularly advantageous in the analyses of big data, or data with large sample sizes and a large number of variables. It outperforms PS in terms of computational time, data loss, and automatic capture of nonlinear relationships and heterogeneous interventions. It also requires less user specification and choices than PS, reducing potential data dredging. We discuss the performance of our method in the context of such big data and present results for very large simulated samples with many variables. We illustrate the method and the insights it yields in the context of three impact studies with different study designs: reanalysis of a field study on the effect of training on earnings, analysis of the impact of an electronic governance service in India based on a quasi-experiment, and performance comparison of contract pricing mechanisms and durations in IT outsourcing using observational data. Collapse

Topics: outsourcing price management big data logistics management business process outsourcing

Methods: propensity score method experimental group logistic regression simulation survey

Seizing the opportunity: towards a historiography of information systems

2012 | Journal of Information Technology | Citations: 7

Authors: Mitev, Nathalie; De Vaujany, François-Xavier

Abstract: Historical perspectives are only timidly entering the world of IS research compa ... Expand

Abstract: Historical perspectives are only timidly entering the world of IS research compared to historical research in management or organisation studies. If major IS outlets have already published history-oriented papers, the number of historical papers – although increasing – remains low. We carried out a thematic analysis of all papers on History and IS published between 1972 and 2009 indexed on ABI and papers indexed in Google Scholar™ for the same period. We used a typology developed by theorists Üsdiken and Kieser, who classify historical organisation research into supplementarist, integrationist and reorientationist approaches. We outline their links with the epistemological stances well known in IS research, positivism, interpretivism and critical research; we then focus on their differences and historiographical characteristics. We found that most IS History papers are supplementarist descriptive case studies with limited uses of History. This paper then suggests that IS research could benefit from adopting integrationist and reorientationist historical perspectives and we offer some examples to illustrate how that would contribute to enriching, extending and challenging existing theories. Collapse

Topics: Google+ database system outsourcing organizational change offshoring

Methods: archival research literature sample literature study survey qualitative content analysis

Theories: organizational theory

A linguistic analysis of group support systems interactions for uncovering social realities of organizations

2011 | ACM Transactions on Management Information Systems | Citations: 0

Authors: Kuo, Feng-Yang; Yin, Chun-Po

Abstract: Talks are actions, and language represents the medium through which we encounter ... Expand

Abstract: Talks are actions, and language represents the medium through which we encounter reality, carry out practical reasoning, and construct social actions. This study applies the speech act theory to analyze the data collected in a study by Trauth and Jessup [2000] and confirms previous research findings that both the topic and the group size influence the pattern of discussion, especially when issues are threatening. It also shows that the abundance of speech acts like assertives, directives, and expressives can be accounted for by a few simple recurring patterns, indicating participants are rather close-minded. More important, linguistic analysis helps uncovering defensive speech routines that inhibit the generation of valid information and create self-sealing patterns of escalating error. Linguistic analysis may therefore complement positivist and interpretive analysis by examining if participants' engagement is superficial or profound, if consensus is reached or blocked, and if certain speech acts lead to dysfunctional organizational learning. Hence, in the era of participatory Web in which language is the primary medium for interactive sharing and dynamic collaboration, linguistic analysis can be applied to study the promises and declarations that people rely on to initiate, coordinate, and complete social actions. Collapse

Topics: group support system anonymity idea generation social network brainstorming

Methods: linguistic analysis qualitative content analysis data dredging survey mixed method

Theories: speech act theory

Profitability of technical analysis in financial and commodity futures markets — A reality check

2010 | Decision Support Systems | Citations: 0

Authors: Yen, Stéphane Meng-Feng; Hsu, Ying-Lin

Abstract: Based on the SPA test (test for superior predictive ability), Sortino and revers ... Expand

Abstract: Based on the SPA test (test for superior predictive ability), Sortino and reversed Sortino ratios, we examined the profitability of a universe of 8061 technical trading rules in ten futures markets including five financial and five commodity underlying assets. We tested whether the best performing rule really beats its buy-and-hold benchmark strategy in bullish and bearish markets, respectively, during the in-sample testing period. The best rules' performance relative to the benchmark is also tested during the one-year out-of-sample period for all ten sets of data. A novel set of multi-indicator rules, MFI–RSI, and four popular categories of single-indicator rules, filter rules, moving averages, on-balance volume averages and momentum strategy in volume, were employed to form our universe of trading rules. The results on the SPA test suggest market efficiency in nine of the ten futures markets, while the results on the Sortino and reversed Sortino ratios reveal persistent outperformance of the best ‘downside’ and ‘upside’ rules relative to the buy-and-hold benchmark across time in four and three futures markets, respectively. Collapse

Topics: accounting anonymity electronic trading decision support system

Methods: data dredging parametric test Student's t-test descriptive statistic literature study

Theories: transaction cost economics