A Comparative Analysis of Enhanced Exponential Smoothing Method Techniques for Bitcoin Price Prediction

Abstract

Since the most recent cryptocurrency price surge and collapse, Bitcoin has gained recognition as an investing platform. Good forecasts are necessary to support investment decisions due to the market's extreme volatility. In this work, we introduced improved exponential smoothing methods for Bitcoin price prediction. Simple to sophisticated models can be smoothed using exponential smoothing techniques. Their adaptability enables them to cope with varying degrees of trend or seasonality in the data on Bitcoin prices. Utilising historical Bitcoin data from 2012 to 2020, we compared the models with their Mean Absolute Error (MAE) and Root Mean Squared Error (RMSE). We made a comparison among four different models; Autoregressive integrated moving average (ARIMA), Seasonal autoregressive integrated moving average (SARIMA), Fb-Prophet, and Silverkite. The MAE score of the Silverkite model is 2.062 and the RMSE score is 5.083, which is minimal as compared to other models. Exponential smoothing techniques enable speedier computations and real-time forecasting since they are computationally less expensive than more sophisticated approaches.