SENTIMENT ANALYSIS OF THE TEXT FOR FORECASTING DATA IN FINANCIAL MARKETS

Abstract

Data forecasting in financial markets is an urgent task in today's world, as the ability to predict the direction of market movement helps investors avoid obvious risks and get rid of additional financial costs. Many different trading platforms have been developed to quickly gain access to large amounts of historical data, which allows you to analyze the financial market from anywhere on the planet and in real time, using only a laptop or a stationary personal computer. Such platforms allow you to develop special strategies and approaches based on fundamental or technical analysis, which take into account news about a particular company, its income, capitalization and the amount of dividends it should pay on time, as well as various technical indicators. News about this or that company helps a potential investor identify certain risks, in particular, personnel, production or, most often in modern realities, reputational. Therefore, the analysis of news texts plays an important role in the formation of fundamental analysis, and that is why it can be done most effectively using neural networks. Over the last decade, thanks to technological innovations and developments, neural networks have taken an important place in the analysis of large data sets, in particular text. Since every piece of news about one or another company, which is an object for a potential investor or trader, has a certain emotional color, for example, positive or negative, it can be determined with the help of a specifically trained neural network, which will help make correct predictions on financial markets and develop effective strategies. In combination with technical analysis, the development and research of such a forecasting approach can produce accurate results. That is why scientific research in this direction is relevant. This scientific article substantiates the relevance of developing a data forecasting system based on the analysis of the emotional coloring of texts. The software was developed based on the Python programming language. The results of the research are described, as well as the output data of the developed program with the obtained accuracy of the analysis are presented. Based on the results of the scientific research, conclusions were drawn.