The paper investigates methods of artificial intelligence in the prognostication and analysis of financial data time series. The usage of well-known methods of artificial intelligence in forecasting and analysis of time series is investigated. Financial time series are inherently highly dispersed, complex, dynamic, nonlinear, nonparametric, and chaotic nature, so large-scale and soft data mining techniques should be used to predict future values. As the scientific literature superficially describes the numerous artificial intelligence algorithms to be used in forecasting financial time series, a detailed analysis of the relevant scientific literature was conducted in scientometric databases Scopus, Science Direct, Google Scholar, IEEExplore, and Springer. It is revealed that the existing scientific publications do not contain a comprehensive analysis of literature sources devoted to the use of artificial intelligence methods in forecasting stock indices. Besides, the analyzed works, which are related in detail to the object of our study, have a limited scope because they focus on only one family of artificial intelligence algorithms, namely artificial neural networks. It was found that the analysis of the use of artificial intelligence systems should be based on two well-known approaches to predicting the behavior of financial markets: fundamental and technical analysis. The first approach is based on the study of economic factors that have a possible impact on market dynamics and more common in long-term planning. Representatives of technical analysis, on the other hand, argue that the price already contains all the fundamental factors that affect it. In this regard, technical analysis involves forecasting the dynamics of price changes based on the analysis of their change in the past, ie time series. Although today there are many developed models for forecasting stock indices using artificial intelligence algorithms, in the scientific literature there is no established methodology that defines the main elements and stages of the algorithm for forecasting financial time series. Therefore, this study has improved the methodology for forecasting financial time series.
展开▼
