Much ado about nothing? Exchange rate forecasting: Neural networks vs. Linear models using monthly and weekly data

Creators: Hann, Tae Horn and Steurer, Elmar
Title: Much ado about nothing? Exchange rate forecasting: Neural networks vs. Linear models using monthly and weekly data
Item Type: Article
Projects: ILR
Journal or Publication Title: Neurocomputing
Page Range: pp. 323-339
Date: April 1996
Divisions: Wirtschaftswissenschaften
Abstract: One of the most popular applications of neural networks is the prediction of financial data, in particular the prediction of the US-$/DM-exchange rate. Though promising results have been achieved, the performance of Neural Nets is hardly ever related to the performance of conventional econometric methodologies. The question arises whether nonlinear methods such as Neural Nets really do outperform linear models. This study compares both approaches in the framework of error-correlation models using monthly and weekly data. Finally the performance with unseen data is compared.
Citation:

Hann, Tae Horn and Steurer, Elmar (1996) Much ado about nothing? Exchange rate forecasting: Neural networks vs. Linear models using monthly and weekly data. Neurocomputing, 10 (4). pp. 323-339. ISSN 0925-2312

Actions (login required)

View Item View Item