Forecasting Stock Price using Hybrid Model based on Wavelet Transform in Tehran and New York Stock Market

Forecasting financial markets is an important issue in finance area and research studies. On one hand, the importance of prediction, and on the other hand, its complexity, have led to huge number of researches which have proposed many forecasting methods in this area. In this study, we propose a hybrid model including Wavelet Transform, ARMA-GARCH and Artificial Neural Network (ANN) for single-period and multi-period forecasting of stock market price in different markets. At first, we decompose time series into detail and approximate series with wavelet transform, and then we used ARMA-GARCH and ANN models to forecast detail and approximate series, respectively. In addition to the approximate series, we use some technical indexes in this model to improve our ANN model. To evaluate the proposed model in forecasting stock price, we compare our model with ANN, ARIMA-GARCH and ARIMA-ANN models on Tehran and New York Stock Exchange (NYSE) historical prices. The results of study show that the proposed model has better performance in single-period forecasting on Tehran and New York market rather than other models.