Improving of stock market prices forecasting by means of Twitter users messages analyzing

Theme of this article is very actual. Investors want to improve stock market forecast algorithms, for the instance, they try to use information about users emotions for forecasting needs. Social network Twitter gives good possibilities to analyze users emotions.The main target of this research is to study dependence between emotions and stock market prices and possibility to use this information for trading.In the first part of research, we concentrate on methods of emotions analyzing. We review theories of behavior finance. In the second part of research, we create methodology to evaluate users emotions by means of Twitter. The main instrument to solve this problem is to use information about emotional word frequency in Twitter messages. Besides, we review the most interesting researches in this sphere. In the third part of research, we use this information for forecasting needs. We arrive at conclusion that we can not use information about emotions in order to improve forecast quality. The research object is feelings and emotions of Twitter users. The research subject is Twitter messages, financial prices, other information.Methodology includes dictionary of emotional words creation. If word is in dictionary, we will account this word in frequency evaluation. We use support vector machine to learn algorithm. We use R project software for our evaluations and use Twitter API to download messages from Twitter. We check assumption about dependence between users twits and stock market prices. The novelty of work is linked with new means of dictionary creation.