Evaluation of Wavelet Neural Network for Predicting Financial Market Crisis

In this paper, we examined the forecasting effect of the wavelet neural network for the currency market crisis. The back-propagation neural network (BPNN) model and the wavelet neural network (WNN) model were compared by the crisis forecasting accuracy and in-sample and out-of-sample test. The dataset consisted of the quarterly data with the time span of Q1/1971-Q2/2006 of eight emerging market countries. The results showed that WNN model could be applied to the currency crises could effectively capture the economic variables associated with the currency crises, and might be to provide a more powerful tool for macroeconomic time series data.