In this work we investigate how news events can be used to predict the financial markets. Namely we built a time series model that includes features obtained from the news and investigated whether the changes in volume of traded shares can be predicted more accurately with this information. The time series model that was built is of an ARMA-GARCH type, because we wanted to account for any clustering of the volatility that is normal for the financial markets. The models were evaluated with the Akaike and Bayesian Information Criterion, while also being compared to the base-line model that did not include any features from the news. Overall our results show that there is an improvement in the model when the information from the news is used and hence show a promising avenue for future research work.