Developing a New Model Based on Artificial Intelligence Techniques for Predicting Bitcoin Fluctuations

要旨

In view of the significance of bitcoin in the global economy, various models have been developed to analyze the behavior of bitcoin prices. The models based on advanced techniques are considered more dependable and precise as they can account for the linear or nonlinear structures involved in the bitcoin pricing process. Furthermore, artificial neural networks (ANNs) are powerful techniques for forecasting in the presence of complex and nonlinear structures, while fuzzy inference systems have the capability of modeling human knowledge and inherent uncertainty. Therefore, the integration of these two techniques, known as ANFIS, has been applied in various areas of time series prediction. This paper proposes the use of an ANFIS model to overcome the linearity and limitations of traditional models, resulting in more accurate predictions. The experimental outcomes of the bitcoin price prediction demonstrate that the proposed model surpasses other methods and improves the precision of bitcoin price prediction. Hence, the proposed model can be a suitable option for forecasting financial time series.