Empirical Analysis: News Impact on Stock Prices Based on News Density

Analyzing the latent relationship between parallel news articles and stock prices has become an important research issue which attracts more and more researchers' attention. It is believed that news articles have impact on prices. Many approaches address this issue either from the documents' sentiment point of view or from the word frequency point of view. In this paper, we propose a new model which captures the density of news articles and mines the latent relationship by employing information entropy to explore the news impact on the market. An empirical study is conducted to analyze market news articles' impact on stock prices. We compare our results with the traditional model which is based on support vector machine (baseline). Experimental results show that our proposed news density model has a better performance on predicting relatively long term news impact.