Bitcoin Crypto-Asset Prediction: With an Application of Linear Autoregressive Integrated Moving Average Method, and Non-Linear Multi-Layered and Feedback Artificial Neural Network Models

This study aims to examine the forecasting performance of two widely used methods in time series analysis: the ARIMA and the MLP-ANN models, focusing on Bitcoin (BTC) price data. ARIMA represents a linear forecasting approach, while MLP-ANN is a nonlinear forecasting method. Both models were evaluated using R-Studio, and the stationarity of the dataset was validated through unit root tests. The dataset consists of weekly BTC price observations from 2020 to 2022. The analysis results indicate that the ARIMA model outperformed the MLP-ANN model in predicting BTC prices. This finding contradicts the growing consensus that nonlinear models are better suited to capture the complex dynamics of financial data. The study contributes to the cryptocurrency forecasting literature by providing empirical evidence on the strengths and weaknesses of both linear and nonlinear models.