Forecasting stock markets using wavelet transforms and recurrent neural networks: An integrated system based on artificial bee colony algorithm

摘要: This study presents an integrated system where wavelet transforms and recurrent neural network (RNN) based on artificial bee colony (abc) algorithm (called ABC-RNN) are combined for stock price forecasting. The comprises three stages. First, the transform using Haar is applied to decompose time series thus eliminate noise. Second, RNN, which has a simple architecture uses numerous fundamental technical indicators, construct input features chosen via Stepwise Regression-Correlation Selection (SRCS). Third, Artificial Bee Colony (ABC) utilized optimize RNN weights biases under parameter space design. For illustration evaluation purposes, this refers simulation results of several international markets, including Dow Jones Industrial Average Index (DJIA), London FTSE-100 (FTSE), Tokyo Nikkei-225 (Nikkei), Taiwan Stock Exchange Capitalization Weighted (TAIEX). As these demonstrate, proposed highly promising can be implemented in real-time trading forecasting prices maximizing profits.