Year of Graduation
Detection of Patterns Emerging in Financial Data in Response to External Events
This paper improves a class of models for the prediction of stock price direction, triggered by certain external event types, such as earnings calls and news about a company. The analysis includes financial features along with features extracted and processed from texts. Performance is evaluated on the constituents of S&P 500 and MOEX indices. Various methods of text representation are assessed in terms of theoretical suitability and relative performance.