Optimal signal multi-resolution by genetic algorithms to support artificial neural network models for financial forecasting

Detecting the features of significant patterns from their own historical data is crucial in obtaining optimal performance, especially in time series forecasting. Wavelet analysis, which processes information effectively at different scales, can be very useful in accomplishing this. One of the most critical issues to be solved in the application of wavelet analysis is to choose the correct filter types and filter parameters. If the threshold is too small or too large, the wavelet shrinkage estimator will tend to overfit or underfit the data. The threshold is often selected arbitrarily or by adopting certain theoretical or statistical criteria. Recently, new and versatile techniques have been introduced to solve that problem. In this study, we propose an integrated thresholding design of the optimal wavelet transform by genetic algorithms (GAs) to represent a significant signal that is most suitable in neural network models, especially for use in chaotic financial markets. The results show that a hybrid system using GAs has better performance than any other method.