Forecasting Stock Index with Neural Network and Wavelet Transform

Stock index as time series are non-stationary and highly noisy due to the fact that stock markets are affected by a variety of factors. It is regarded as one of the most challenging application of time series forecasting. Predicting stock index with the noisy data directly is usually subject to large errors. In this paper we compare forecasting the stock index via Wavelet De-noising-based Neural Network (WDNN) with forecasting stock index via single neural network. The daily Tehran Stock index from April 2006 to June 2013 are used to compare the application of the WDNN in predicting the stock index. Experimental results show that de-noising with wavelet transform outperforms the single neural network.