Sentiment in German-language News and Blogs, and the DAX

An analysis of a diachronically organised corpus of Germanlanguage newspaper articles and blog posts on economy and finance is presented using a prototype dictionary of affect in German. The changes in the frequency of occurrence of positive and negative polarity words are rendered as return time series and the properties of this time series are described. The returns and the variance of returns show that the time series of affect words share a number of properties with that of financial time series in general and in particular with a contemporaneous time series of the DAX – the German stock exchange index. Both the series have outliers not predicted by a normal distribution. Our analysis is a starting point for studying the impact of news and blogs on financial instruments and vice versa.