search for


A study on machine learning-based KRW/USD exchange rate prediction model using swap point determinants: Focused on optimal structure finding in multi layer perceptron
Journal of the Korean Data & Information Science Society 2018;29:203-16
Published online January 31, 2018
© 2018 Korean Data and Information Science Society.

Young Churl Kim1 · Hyun Jun Lee2 · Ji Woo Kim3 · Jae Joon Ahn4

1Department of Investment Information Engineering, Yonsei University
23Department of Industrial Engineering, Yonsei University
4Department of Information & Statistics, Yonsei University
Correspondence to: Assistant professor, Department of Information & Statistics, Yonsei University, Wonju, 26493, Korea. E-mail:
Received October 29, 2017; Revised November 23, 2017; Accepted November 26, 2017.
This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License ( which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited.
With the globalization of the economy and the increasing interconnection of countries, the influence of the dollar, which is the main key currency of international settlement and financial transactions, is increasing, and the need for precise exchange rate prediction is emerging. In this study, the exchange rate between the korean won and the dollar is predicted by using multi layer perceptron model. The input variables are the spot exchange rate, the korean won interest rate, the dollar interest rate, and the dollar procurement risk premium, which are the determinants of the swap point. Empirical study is carried out by varying the number of hidden layers and nodes in the model to find the optimized model structure for USD/KRW exchange rate prediction. The performance of the prediction is measured by the difference from the actual exchange rate, and the superiority of the suggested model is verified by comparing the performance with the random walk process and forward exchange rate. The results of this study confirm the exchange rate forecasting ability of the multi layer perceptron, which is a representative artificial neural network. Also, the market participants will be supported for the foreign exchange investment by using proposed model in this study.
Keywords : Artificial neural network, exchange rate prediction, hidden layer, multi layer perceptron, swap point.