Premium
Forecasting exchange rates using feedforward and recurrent neural networks
Author(s) -
Kuan ChungMing,
Liu Tung
Publication year - 1995
Publication title -
journal of applied econometrics
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 2.878
H-Index - 99
eISSN - 1099-1255
pISSN - 0883-7252
DOI - 10.1002/jae.3950100403
Subject(s) - computer science , foreign exchange market , feedforward neural network , mean squared error , feed forward , artificial neural network , random walk , mean squared prediction error , recurrent neural network , series (stratigraphy) , exchange rate , nonlinear system , econometrics , sample (material) , artificial intelligence , machine learning , mathematics , statistics , economics , paleontology , physics , chemistry , chromatography , quantum mechanics , control engineering , biology , engineering , macroeconomics
In this paper we investigate the out‐of‐sample forecasting ability of feedforward and recurrent neural networks based on empirical foreign exchange rate data. A two‐step procedure is proposed to construct suitable networks, in which networks are selected based on the predictive stochastic complexity (PSC) criterion, and the selected networks are estimated using both recursive Newton algorithms and the method of nonlinear least squares. Our results show that PSC is a sensible criterion for selecting networks and for certain exchange rate series, some selected network models have significant market timing ability and/or significantly lower out‐of‐sample mean squared prediction error relative to the random walk model.