Wavelet analysis and forecasting of financial time series

Abstract

Models and methods of forecasting financial time series are considered in the work. The main advantages and disadvantages of traditional models and neural networks for forecasting without data preprocessing are analyzed. Wavelet analysis and a recurrent neural network with long short-term memory (LSTM) were applied to predict the exchange rate of cryptocurrency. The obtained results are compared with the results of existing approaches, the efficiency is determined and a solution is proposed. Figs.: 6. Tabl.: 2. Refs.: 13 titles.
 Keywords: financial time series, neural networks, forecasting, data preprocessing, wavelet analysis, recurrent neural network with long short-term memory, cryptocurrency exchange rate.