Time series neural network systems in stock index forecasting
COMPUTER MODELLING & NEW TECHNOLOGIES 2015 19(1B) 57-61
School of Management, Harbin Institute of Technology, Harbin, China
This paper adopts artificial neural network (ANN) and two varieties of time series neural network to forecast the stock index of Chinese market. Daily close prices between 1999 and 2011 are tested. The ANN works as a benchmark. Its inputs include delayed price and technical indicators. Time series neural network with external input (NARX) outperforms the Time series neural network (NAR), and it works best when the delay is 8. Moreover, NARX has the best ability of the three. This is mainly resulted from the fact that it contains external data and the technical indicators while NAR does not. As a whole, the ANN and NARX models achieved satisfying results. They can be employed by practitioners to assist trading and by regulators for monitoring. The NARX will be improved when more external data imported.