The Application of Artificial Neural Networks to Exchange Rate Forecasting: The Role of Market Microstructure Variables

Artificial neural networks (ANN) are employed for high-frequency Canada/U.S. dollar exchange rate forecasting. ANN outperform random walk and linear models in a number of recursive out-of- sample forecasts. The inclusion of a microstructure variable, order flow, substantially improves the predictive power of both the linear and non-linear models. Two criteria are applied to evaluate model performance: root-mean squared error (RMSE) and the ability to predict the direction of exchange rate moves. ANN is consistently better in RMSE than random walk and linear models for the various out-of-sample set sizes. Moreover, ANN performs better than other models in terms of percentage of correctly predicted exchange rate changes (PERC). The empirical results suggest that optimal ANN architecture is superior to random walk and any linear competing model for high-frequency exchange rate forecasting.