Capital markets and sentiment analysis in different text structures

The project will involve the analysis of how information affects asset prices over high frequencies. The innovation is that the raw information will be in the form of text contained in a high number of Twitter feeds. This text will be analysed using computational methods and categorised according to whether it is market sensitive (i.e. good or bad news) or not. The relationship between this clean information and asset returns will then be assessed in an out-of-sample forecasting exercise.