Next-Day Bitcoin Price Forecast

摘要

This study analyzes forecasts of Bitcoin price using the autoregressive integrated movingaverage (ARIMA) and neural network autoregression (NNAR) models. Employing the static forecastapproach, we forecast next-day Bitcoin price both with and without re-estimation of the forecastmodel for each step. For cross-validation of forecast results, we consider two different trainingand test samples. In the first training-sample, NNAR performs better than ARIMA, while ARIMAoutperforms NNAR in the second training-sample. Additionally, ARIMA with model re-estimationat each step outperforms NNAR in the two test-sample forecast periods. The Diebold Marianotest confirms the superiority of forecast results of ARIMA model over NNAR in the test-sampleperiods. Forecast performance of ARIMA models with and without re-estimation are identical for theestimated test-sample periods. Despite the sophistication of NNAR, this paper demonstrates ARIMAenduring power of volatile Bitcoin price prediction.