Forecasting Using Time Series Analysis Method in Crypto Currency Period 2015 – 2022

Cryptocurrency is a digital currency that is currently much interested as an alternative investment. ARIMA, GARCH, and Holt's Winter method are one of the methods used for forecasting time series data. This research aims to create a model and predict the price of bitcoin. The data used is secondary data in the form of daily closing price data for Bitcoin, Ripple, and Litecoin, as much as 2,520 daily closing price data starting from August 07, 2015 to June 30, 2022, to predict Bitcoin, Ripple, and Litecoin prices for the following 30 periods starting on July 01, 2022 to July 31, 2022. This study determines the forecasting model with an error below 5% with MSE, MAPE, and U Theil. The study results indicate that the Holt Winter forecasting model is suitable for Bitcoin, Ripple, and Litecoin. The Holt Winter Bitcoin model produces a MAPE value = 2.605%, Holt's Winter Ripple produces a MAPE value = 4.334%, and Holt's Winter Litecoin produces a MAPE value = 3.598%