АСИМЕТРИЧНИЙ АНАЛІЗ ТОНАЛЬНОСТІ ДІЛОВИХ НОВИН В СИСТЕМАХ ПРОГНОЗУВАННЯ ДАНИХ НА ФІНАНСОВИХ РИНКАХ

Abstract

Forecasting data in financial markets is a pertinent task in the modern world. The ability to predict market movements helps investors avoid obvious risks and spare themselves additional expenses. Numerous trading platforms have been developed to quickly access extensive historical data, enabling real-time analysis of the financial market from any corner of the planet using only a laptop or personal computer. Such platforms allow the development of unique strategies and approaches based on fundamental or technical analysis, taking into account news about a particular company, its earnings, capitalization, and the amount of dividends it is expected to pay on time. Business news is a crucial source of information about the state of the economy and markets. They can be used for forecasting future events. One method of forecasting based on business news is sentiment analysis. Sentiment analysis allows assessing the positivity or negativity of business news.Traditional sentiment analysis methods employ a symmetric approach. This means that positive and negative news are equally considered in forecasting. However, in the real world, positive news may have a greater impact on markets than negative news. This is because positive news can stimulate economic activity, while negative news may hinder it. The article explores the application of asymmetric sentiment analysis of business news in financial data forecasting systems. Various methods of sentiment analysis of business news, their advantages, and disadvantages are analyzed. A new approach to sentiment analysis of business news is proposed, which comprehensively utilizes artificial neural networks and principal component analysis.