An application on stock price prediction with neural network by using wavelet transform

Wavelet transform is a method based on time and frequency analysis used in many areas. It is very useful in noise extraction, image processing, frequency analysis, and financial forecasting methods. Therefore, this method is widely used in literature with its beneficial qualifications. Stocks are an investment tool generally preferred by risk lovers. Investors use various methods to reduce risk and to predict the direction of the stocks right. Artificial neural networks are one of the widely used methods in stock market predictions. But, high volatile data sets may cause to memorize rather than learning. Therefore, in order to clear daily fluctuating movements wavelet transform is used to smooth the highly volatile data and afterwards neural network analysis is performed. In this study, an application of wavelet based artificial neural networks on BIST is studied. In the application, 6 of the BIST30 stocks are selected randomly. The primary aim of this study is to determine whether stock price movements are predictable with wavelet transform and artificial neural networks and to contribute to the literature related to this area.