A New Forecasting Framework for Bitcoin Price with LSTM

Abstract

Long short-term memory (LSTM) networks are a state-of-the-art sequence learning in deep learning for time series forecasting. However, less study applied to financial time series forecasting especially in cryptocurrency prediction. Therefore, we propose a new forecasting framework with LSTM model to forecasting bitcoin daily price with two various LSTM models (conventional LSTM model and LSTM with AR(2) model). The performance of the proposed models are evaluated using daily bitcoin price data during 2018/1/1 to 2018/7/28 in total 208 records. The results confirmed the excellent forecasting accuracy of the proposed model with AR(2). The test mean squared error (MSE), root mean square error (RMSE), mean absolute percentage error (MAPE), and mean absolute error (MAE) for bitcoin price prediction, respectively. The our proposed LSTM with AR(2) model outperformed than conventional LSTM model. The contribution of this study is providing a new forecasting framework for bitcoin price prediction can overcome and improve the problem of input variables selection in LSTM without strict assumptions of data assumption. The results revealed its possible applicability in various cryptocurrencies prediction, industry instances such as medical data or financial time-series data.