Short term forecasting of stock market based on R/S analysis and fuzzy neural networks
Yiwen Yang, Yang Chao-jun · 2004
Input space of nonlinear model is partitioned into several fuzzy subspaces. Within each subspace, a local linear model is used to model the local features of nonlinear system, and then the global model output is obtained by interpolating the outputs of local models. Adaptive network fuzzy inference system, based on Sugeno fuzzy inference model, is one way of neural network realization of the fuzzy modeling based on the idea of local linear modeling above. The results of R/S analysis show that Shanghai stock market has long-term memory, thus possible to predict. This paper combines adaptive network fuzzy inference system and fractal market hypothesis to implement multi-step prediction of Shanghai Stock Composite Index. The final result shows that the prediction can benefit from human's intuitive knowledge on the market, even if it is little and simple.