Abstract The problem of "correct" description of the dynamics of stochastic processes is the subject of numerous studies. The article deals with financial and economic indicators as the objects of research. Understanding the dynamics of financial indicators of the world market is becoming more and more relevant due to the growth of complex financial instruments, as well as due to political and economic events in the world. Observations of recent months indicate that a number of events can significantly affect the dynamics of financial indicators and, therefore, new approaches and methods are needed to quickly assess the impact of such events on financial indicators. The article investigates the possibility of predicting the dollar exchange rate against the ruble based on news reports using NLP and machine learning algorithms [1]. It is shown that the model constructed using text analysis is able to assess the trend of movement of the financial indicator. Fractal and statistical analysis of the selected indicators was performed. Time series clustering was performed using topological analysis methods. Neural network models of the indicator of interest are constructed on the basis of all the conducted studies [2]. The analysis of the obtained solutions is carried out.