Financial Sentiment on Twitter's Community and it's Connection to Polish Stock Market Movements in Context of Behavior Modelling*
Purpose: The paper analyzes the relationship between tweets published on social media related to the stock market and stock market movements. The research conducted aims to find patterns that may account for behavioral modelling patterns between social media and the stock market. Design/Methodology/Approach: To conduct the research, 452,776 tweets containing a hashtag with the name of at least one company included in the WIG20 of the Polish stock market were analyzed. The data obtained also included information about the reach of the tweet and the popularity of the author. The analyzed text also contained a timestamp allowing the tweets to be linked to the behavior of the stock market. Additionally, the analysis implemented developed Polish financial sentiment analysis vocabularies allowing for better pattern retrieval. Findings: The obtained results indicate that among the analyzed data, three patterns related to the sentiment of statements accompanying a given company in social media can be distinguished. A consistent sentiment present in many statements with a wide range is decisive. When different sentiments are present, the presence of a pattern is not identifiable. The implementation of Polish financial dictionaries allowed to distinguish also individual words characterizing positive and negative sentiment. Practical Implications: The results obtained can form the basis for further, more in-depth analyses between sentiment published in social media and stock market movements. Originality value: Studies analyzing the relationship between the analysis of tweets and stock markets are conducted but analyses related to the Polish stock market are not a popular field of analysis. Due to the unique nature of each stock exchange, it is indicated to be innovative and to be able to emerge some patterns, useful in the context of investing.