An intelligent trend prediction and reversal recognition system using dual-module neural networks

Short-term trends of price movement for common stocks traded on the Taiwan stock market have been modelled and predicted using the dual-module neural networks (dual net) proposed. Both neural network modules of the dual net learn the correlations between the trends of price movement and the retrospective technical indices. An adaptive reversal recognition mechanism which can self-tune the threshold to identify the buying or selling signals is developed in the system. Due to the features of acceptable returns, high hit ratio and low risks shown in the performance evaluation, an intelligent stock trend prediction and reversal recognition system can be realized using the dual-module neural networks