| Abstract [eng] |
The relevance of major pharmaceuticals companies’ stock prices is noticeable due to major rise in stock prices in several previous years. From 2014 in almost a year stock prices in this sector has risen about 3 times. Before buying stocks, investors make a thorough analysis about the companies, but main part of this analysis consist of analyzing textual information. Therefore, in this master thesis, with the appropriate methods for text mining, is intended to define whether there exists news impact to stock prices. For this research timeframe of 2014 – 2016 years has been chosen due to noticeable rise in stock prices, major pharmaceuticals companies have been chosen by the amount of revenue in 2016 and by impact in the market. Information signals were evaluated by their positivity and negativity – it was done by evaluating each word in an article by its sentiment and polarity. After the research, it has been noticed, that only in “AstraZeneca” and “Bayer” companies’ cases news impact on stock prices exists. It has been ascertained, that for the rest of the companies’ information signals’ impact on stock prices is not statistically significant. It has been also noticed, that in “Bayer” company’s case news impact statistical significance differs due to evaluation of news positivity and negativity: when information signals are evaluated by sentiments it has been estimated, that news impact for “Bayer” stock prices is statistically significant, also it has been identified that one day news’ lag impact is statistically significant. However, evaluating information signals by polarity, news and their lagged values are not statistically significant. It was also confirmed with this research, that major pharmaceuticals companies’, that have been analyzed, stock prices alteration is a random walk process. |
