Estimating and Forecasting Bitcoin Daily Returns Using ARIMA-GARCH Models

This paper provides an evaluation of the forecasting performance of hybrid ARIMA-GARCH model in forecasting Bitcoin daily price returns. We combined ARIMA and GARCH model with Normal, Student�s t and Skewed student�s t distributions. To make the series stationary, Bitcoin daily price data was transformed to Bitcoin daily returns. By using Box-Jenkins method, the appropriate ARIMA model was obtained and for capturing volatilities of the returns series GARCH (1, 1) models with Normal, Student�s t and Skewed student�s t distributions was used. To evaluate the performance of the models, the study employs two measures, RMSE and MAE. The results reveal that ARIMA (2, 0, 1) -GARCH (1, 1) with Normal distribution outperform the other three in terms of out-of-sample forecast with minimum RMSE and MAE. The findings can aid investors, market practitioners, financial institutions, policymakers, and scholars.