Neural Network Approaches for Forecasting Stock Market Dynamics Based on News Flow

The news flow contains a large array of information that can be used to predict stock prices and other assets prices. However, private investors are not able to analyze a lot of textual information in a short period of time. In turn, progress in the field of neural network natural language processing allows creating models capable of processing large amounts of data quickly enough. The aim of this study is to increase the accuracy of forecasting the dynamics of the stock market through the implementation of a model for processing news flow using neural networks. The work includes the tasks of reviewing the subject area and existing approaches to solving the problem, implementing and testing a number of neural network models for predicting changes in the stock index, as well as analyzing the results. As a result of the experiments, a number of models were identified that have the greatest potential in solving the problem and can be used by private investors as a tool to evaluate the effectiveness of the securities portfolio.