The value of shares prediction in an unstable economy using neural networks

Abstract

The relevance of this research topic is due to the need to develop algorithmic provision of the market value forecasting for shares in Ukraine and the introduction of the concept for increasing information transparency of the domestic stock market. All this will help improve the investment market, provide investment and development of Ukrainian companies and the economy as a whole. An analysis of researchon the use of methods for computational intelligence, including neural networks to model the behavior of stock market participants and solve the problem of forecasting. A study was conducted based on using neural networks of different architecture to predict the market value of shares in the stock markets of Ukraine, which are in the process of formation and development. The following models of neural networks were chosen for experimental research: Long short-term memory; Convolutional neural network; a hybrid model that combines two neural network architectures CNN and LSTM; a hybrid model consisting of a variational mode decomposition algorithm and a long-term memory neural network (VMD-LSTM). Four shares of the Ukrainian Stock Exchange were selected forforecasting: Tsentrenergo (CEEN); Ukrtelecom (UTLM); Kriukivs’kyi Vahonobudivnyi Zavod PAT (KVBZ); Raiff Bank Aval (BAVL). Estimates of forecast quality are calculated. It was concluded that it is advisable to use the LSTM network to forecast stocks that are on the stock exchanges of countries with unstable economies.