An application of SVMs to predict financial exchange rate by using sentiment indicators

This paper analyses, through the use of Support Vector Machines (SVM), the impact that the economic sentiment indicator variable known as ESI has on a model's forecasting accuracy when it comes to currency exchange rate prediction. The study has been carried out either using exponential or simple moving averages. Weekly currency exchange rate between the European euro and some major currencies have been considered. The results obtained show that the proposed indicator can significantly impact the model's forecasting performance compared to traditional models where no qualitative information is incorporated.