Neural network methods in financial time series forcasting
WU Chun-guo · Journal of Changchun Post and Telecommunication Institute · 2004
Some problems in applications of neural network methods to financial time series forecasting are introduced. Same methods to solve these problems are proposed. Numerical simulations and analysis are performed based on some relevant models and algorithms. Some useful conclusions are obtained. Several neural network models, such as multilayer feadforward models, radial basis function neural networks and support vector machines, used mainly in financial time series forecasting are discussed and analyzed. Recent advances and results of studies on some modified methods are summarized. Some possible topics for the future study applying neural networks to financial time series forecasting are pointed out.