Modeling the third-tier stock prices based on the information environment dynamics

Abstract

Subject. The article examines the influence of the tonality of news on the tire 3 stock quotes. Objectives. The purpose is to reflect the specifics of the influence of news sentiment on the prices of third-tier shares traded on the Moscow Exchange. Methods. We chose Telegram messenger as a source of news. Using the Python programming language, we wrote scripts. On their basis, we obtained stock quotes. Results. We used binarization of the difference in closing and opening prices. We also used tokenization to divide one text into separate words, lemmatization to bring a word into its initial form, taking into account contextual information, clustering of the result obtained, and selection of the most prominent topics. After that, we built classification models based on the methods of the naive Bayesian classifier, tree ensemble, and random forest techniques. Conclusions. We combined the above-mentioned models with the corresponding methods, and presented conclusions on the findings. The paper mentions the subsequent stages to study the impact of news on stock quotes.