Exchange rate modelling using news articles and economic data

This paper provides a framework of using news articles andeconomic data to model the exchange rate changes between Euro andUS dollars. Many studies have conducted on the approach of regressingexchange rate movement using numerical data such as macroeconomicindicators. However, this approach is effective in studying the long termtrend of the movement but not so accurate in short to middle termbehaviour. Recent research suggests that the market daily movement isthe result of the market reaction to the daily news. In this paper, it isproposed to use text mining methods to incorporate the daily economicnews as well as economic and political events into the prediction model.While this type of news is not included in most of existing models due toits non-quantitative nature, it has important influence in short to middleterms of market behaviour. It is expected that this approach will lead toan exchange rate model with improved accuracy.