NeuroSolutions applied to Taiwan stock price prediction

摘要

This article uses neuron network software, named NeuroSolutions 5, to forecast the stock price of the next day. Two methods, Time-lag recurrent network and New-function approximation network, are selected for suiting the stock price forecast by using Taiwan stock market historic information. The stock price prediction of HON HAI PRECISION IND.CO.LTD for next day are analyzed and discussed with its serviceability. The test results are shown that Newfunction approximation network method achievement is nice fitted with the field data, i.e. better than the time series related method. This result is opposite of the original cognition.