Leveraging temporal properties of news events for stock market prediction

Abstract

Investors make decisions based on various factors, including consumer price index, price-earnings ratio, and also miscellaneous events reported by newspapers. In order to assist their decisions in a timely manner, many studies have been conducted to automatically analyze those information sources in the last decades.  However, the majority of the efforts was made for utilizing  numerical information, partly due to the difficulty to process  natural language texts and to make sense of their temporal  properties.  This study sheds light on this problem by using deep  learning, which has been attracting much attention in various areas  of research including pattern mining and machine learning for its  ability to automatically construct useful features from a large  amount of data.  Specifically, this study proposes an approach to  market trend prediction based on a recurrent deep neural network to  model temporal effects of past events.  The validity of the proposed  approach is demonstrated on the real-world data for ten Nikkei  companies.