Neural Network Methods Applied Research In The Securities Market Forecast

In the financial securities market, traditional linear methods have long seen their wide use as the major analysis methods, and they still play a dominant role at present. To some extent, linear analysis methods are successful and effective, since they provide a way to explain many directly perceptible phenomena and concrete problems in the financial securities market. However, the securities market is so complex that pure linear analysis methods cannot generate satisfactory explanations to more complex phenomena and difficult problems. Upon that,it is becoming a new trend to model securities market as a complicated nonlinear system.Firstly, this paper investigate the property of financial and economic systems base on studying the securities market which prove that Chinese securities market is a complex nonlinear system. On one hand, it is spontaneous and disordered. On the other hand, it is self-organized, self-adjusted and approaching a certain rule. The purpose of the research in this paper is to combine neural network analytic method, one of the most important nonlinear methods, with financial theories in terms of the nonlinear characteristics of the financial market, and base the research on a more intelligent foundation as well as theories that come closer to the reality by constructing models, extracting features, and analyzing the time series data of the financial market.This paper introduces basic of neural network and its application in securities market prediction, investigates some typical neural networks, includes BP network, fuzzy network,??and their capability of predicting time sequence properties of China securities market. The author gives a detail comparison and analysis of results and also proposes a new type of neural network called Combined Neural Network, which including BP network for tracking training errors. A better result is achieved by using Combined Neural Network. The applications of the research results to real project cases in the securities market are also presented in the paper. It can be regarded as a useful effort to apply nonlinear theories and nonlinear analysis methods to the securities market.