Artificial neural networks in forex rate prediction: vetting the past and charting the future using bibliometric analysis

Abstract

Purpose The present paper aims to systematically map the research landscape of artificial neural networks (ANNs) in forex rate forecasting by particularly (1) uncovering significant research trends, key players, scientific collaborations, hot topics, emerging themes, and primal knowledge dimensions; and (2) discovering potential areas for future research in the concerned field. Design/methodology/approach To delve deeply into the field, the present study employed the fusion approach of bibliometric analysis (quantitative) and content analysis (qualitative) to analyse 487 articles published in Scopus-indexed journals during 1993–2024. The extracted data was analysed using RStudio (Biblioshiny) and VOSviewer software tools. Findings The analysis revealed the overall upward trend of the research with (1) the proliferation of publications since 2019; (2) China as the most productive country; (3) “Expert Systems with Applications” as the prominent journal; and (4) “exchange rate prediction” and “genetic algorithm” as the trendy areas, whereas “quantitative trading”, “hybrid models”, and “long short-term memory” are the emerging themes of the field. Additionally, model optimization, technical analysis, model hybridization, and modelling data complexity were discovered as the primal knowledge dimensions in the field. Originality/value To the best of the authors' knowledge, the current study is the first that systematically deconstructs the social, conceptual, and intellectual structure of ANN research in forex rate forecasting. Its main contribution lies in equipping (1) researchers with potential areas for future investigations and advancements in the field; and (2) practitioners with means of overcoming modelling challenges and improving the forecasting accuracy of ANNs, thereby enhancing their forecasting-based decision-making capacity.