Output details

Financial time series prediction using Polynomial Pipelined Neural Networks

<24> This work introduces a uniquely effective method for predicting financial time series. The predictor structure is based on the engineering concept of divide and conquers in which highly difficult problems such as the prediction of financial time series can be solved by using a number of simpler pipelined neural network architectures. It is the very first approach to indicate that transforming financial signals into stationary signals is not necessary for future prediction.