Modeling and Forecasting High Frequency Financial Data: Statistical and Neural Approach

Forecasting systems are applied which are based on the latest statistical theory and artificial neural networks. The impact of these methods to risk reduction is judged in managerial decision-making. The fundamental question arises whether non-linear methods like neural networks can help modeling any non-linearities being inherent within the estimated statistical model and outperforms statistical modeling approach. The proposed novel neural modeling approach is applied to high frequency time series of USD/CAD exchange rates. Our results show that the proposed neural approach achieves better forecast accuracy on the validation dataset than most available statistical techniques. We also show how the proposed information technology contributes for the people who will make and at the sometime use the forecast in financial institutions, companies, medium and small enterprises.