Correlating Financial Time Series with Micro-blogging Activity

The paper is devoted to the study the behavior of financial time series, as well as the search for their dependence on the series in the Twitter system. At the beginning of the work, a theoretical account is given of the theory of chaos and modern methods of predicting financial series. In the paper analysis of financial time series for their nature was done in order to determine whether regularity is possible or financial time series are random variables. The analysis was carried out on the basis of the Takens method or as it is called the immersion method. Then the Grassberger-Procaccia algorithm was used. Using this algorithm, the correlation integral was calculated. Further in the work were used two financial series: the euro-dollar and quotes Toyota Motor. For each series, a series of ranks in the Twitter system was made. Work with the Twitter information system was carried out using the analytics system IBM Watson Analytics. In addition to IBM Watson Analytics, we will perform a comparative analysis of such systems for working with time series of the Twitter system. In Twitter, we searched for hashtags. For the euro-dollar series, two sets of data have been created with hashtags #trump and #brexit. For Toyota Motor, a series was selected for the #toyota hashtag. With these series, analytical processing was carried out. In particular, each series was subject to sentimental analysis. For the stock price chart, some problems that arise in the course of market analysis will be identified. We will offer our classification of news coming to the market which was not previously used in researches. The work will analyze the dependence of sentimental charts and financial time series. Next, an algorithm will be given on how to work with the data obtained in order to predict the behavior of financial charts.