FORECASTING THE DYNAMICS OF THE ETHEREUM CRYPTOCURRENCY RATE

The study conducts an analysis of the cryptocurrency market to select a more relevant coin with a large capitalization. Due to its gaining popularity, the cryptocurrency ETH was chosen for the study. Literature review confirms the relevance and shows a low level of study of this cryptocurrency within the topic. The values of the series for the period from 14.03.2020 to 30.06.2020 are considered. The stationarity tests with the analysis of ACF and ChakF charts of the selected cryptocurrency series, as well as the extended Dickey-Fuller tests are carried out. The initial data were transformed to reduce dispersion (logarithm) and to exclude seasonality and trend (taking the second difference). Forecasting models were built by means of autoregressive model — ARIMA (p, d, q) integrated moving average and by means of statistical methods (analysis of ordinary and adjusted coefficients of determination) and Akaike and Schwartz information criteria the only appropriate model was selected. In order to be able to predict values, the model was diagnosed by three parameters (residuals white noise; the model is stationary and reversible) using residuals analysis and unit root test, and after obtaining a successful result, 2 values were predicted for the period from 01.07.2020 to 02.07.2020. A comparison of the real and predicted cryptocurrency price values shows that the sum of squares of deviation is significantly lower than the selected accuracy of the results. Visual analysis of the chart also shows that the model is able to form price peaks and declines. It is concluded that this method is effective in short-term cryptocurrency forecasts.

cryptocurrency, cryptocurrency market, Ethereum, time series, autoregressive model, moving average model, forecasting, financial time series, stationarity, autocorrelation function

Basova A. S., Vladimirova D. B. Forecasting the dynamics of the Ethereum cryptocurrency rate. Business. Education. Law, 2022, no. 4, pp. 199—204. DOI: 10.25683/VOLBI.2022.61.499.