A cross-validation analysis of neural network out-of-sample performance inexchange rate forecasting

Econometric methods used in foreign exchange rate forecasting have produced inferior out-of-sample results compared to a random walk model. Applicatio ns of neural networks have shown mixed findings. In this paper, we investig ate the potentials of neural network models by employing two cross-validati on schemes. The effects of different in sample time periods and sample size s are examined. Out-of-sample performance evaluated with four criteria acro ss three forecasting horizons shows that neural networks are a more robust forecasting method than the random walk model. Moreover, neural network pre dictions are quite accurate even when the sample size is relatively small.