Several Problems of Exchange Rate Forecasting Using Neural Network

In this paper, the authors discuss several controversial issues about exchange rate forecasting by using neural network, which include whether the type of data sample and the size of it will affect the prediction accuracy, whether time-series model or neural network model is more suitable for exchange rate forecasting, and whether neural network suitable for long-term prediction or short-term prediction. The techniques that the authors applied are BP-neural network model in Matlab and time-series model in Excel. Some beneficial results are indicated from this study, for example, the type of data sample has little effect on the prediction accuracy in exchange rate forecasting by neural network.