Should Deep Learning Models be in High Demand, or Should They Simply be a Very Hot Topic? A Comprehensive Study for Exch

  • PDF / 1,643,098 Bytes
  • 29 Pages / 439.37 x 666.142 pts Page_size
  • 56 Downloads / 162 Views

DOWNLOAD

REPORT


Should Deep Learning Models be in High Demand, or Should They Simply be a Very Hot Topic? A Comprehensive Study for Exchange Rate Forecasting Firat Melih Yilmaz1   · Ozer Arabaci2  Accepted: 11 September 2020 © Springer Science+Business Media, LLC, part of Springer Nature 2020

Abstract Exchange rate movements can significantly impact not only foreign trade, capital flows, and asset portfolio management, but also real economic activity. Therefore, the forecast of exchange rates has always been of great interest among academics, economic agents, and institutions. However, exchange rate series are essentially dynamic and nonlinear in nature and thus, forecasting exchange rates is a difficult task. On the other hand, deep learning models in solving time series forecasting tasks have been proposed in the last half-decade. But the number of formal comparative study in terms of exchange rate forecasting with deep learning models is quite limited. For this purpose, this study applies ten different models (Random Walk, Autoregressive Moving Average, Threshold Autoregression, Autoregressive Fractionally Integrated Moving Average, Support Vector Regression, Multilayer Perceptron, Recurrent Neural Network, Long Short Term Memory, Gated Recurrent Unit and Autoregressive Moving Average-Long Short Term Memory Hybrid Models) and two forecasting modes (recursive and rolling window) to predict three major exchange rate returnsnamely, the Canadian dollar, Australian dollar and British pound against the US Dollar in monthly terms. To evaluate the forecasting performances of the models, we used Model Confidence Set procedure as an advanced test. According to our results, the proposed hybrid model produced the best out-ofsample forecast performance in all samples, without exception. Keywords  Deep learning · Forecasting · Exchange rates · Hybrid model

* Firat Melih Yilmaz [email protected] Ozer Arabaci [email protected] 1

The Graduate School of Natural and Applied Sciences, Dokuz Eylul University, Izmir, Turkey

2

Department of Econometrics, Uludag University, Bursa, Turkey



13

Vol.:(0123456789)



F. M. Yilmaz, O. Arabaci

1 Introduction Deep learning (DL) models have started to emerge in various economics fields; time-series forecasting is one of the fields exhibiting the most enthusiasm. In particular, when one considers its evolution, it is clear that exchange rate forecasting is at the forefront of this movement. Of course, a question arises: “Should DL models be in high demand, or should they simply be a ‘very hot topic’?”. The answer to this question can be derived by increasing the number of formal comparative studies undertaken and explicitly revealing the statistical gains derived from using DL models in these fields. Foreign exchange rates are one of the important financial and macroeconomic variables for all economies. Exchange rate movements can significantly impact not only foreign trade, capital flows, and asset portfolio management, but also real economic activity. It is clear that given the r