A Systematic literature review on Cryptocurrency forecasting

Abstract

This study systematically reviews scientific research on predicting cryptocurrency markets. A total of 790 articles obtained from the Web of Science database were included in the analysis, and the structure of the literature was evaluated using bibliometric methods. The preliminary investigation indicated that studies examining the prediction of cryptocurrencies have undergone a substantial increase since 2016. While a significant proportion of the extant literature pertains to Bitcoin, the first cryptocurrency, it is evident that other cryptocurrencies, such as Ethereum, have also attracted the attention of researchers over the years. The analysis yielded four primary categories: machine learning-based prediction methods, financial risk and volatility analyses, behavioral and technical determinants, and finally, advanced deep learning methods. In the context of cryptocurrency prediction, studies have underscored the significance of attributes, emphasizing their role in enhancing the efficacy of prediction models. These studies have also highlighted the impact of integrating machine learning and deep learning-based models with conventional methods in enhancing the performance of established models. The study emphasizes the necessity to direct future research towards the integration of behavioral indicators and the examination of multiple market relationships.