The Application Of The Wavelet Neural Networks In The Equity Warrant Market Of China

China re-established the warrants market in August 22th 2005,due to lack of a sound system,investors are in need of a detailed understanding of the warrants' theory,which leads to the confusion in this market.Blind speculation increased the risk of investment. If we are able to simulate the future trend of prices within a certain period of time,the investors will be provided a basis for decision-making,which can help to reduce risk and enhance the market stability.As the Black-Scholes model assumptions do not match with the actual situation,the theory-price set through the Black-Scholes model often falls far away from the actual market price.Therefore we need to establish a model with less hypothesis,to simulate the price trend of the warrants.Neural network with its characteristics of Non-linear,self-organization,self-learning and self-regulating has provided us a powerful simulation tool.This article is based on the recently developed "wavelet neural network theory",to use the Morlet wavelet function as activation function in hidden layer,LMS(Least Mean Square) energy function as the error function. We adjust the coefficients of gradient and momentum items step by step when training the network using additional momentum method.Then a Wavelet Neural Network of in the Equity Warrant market of China is built.A case on the SAIC MOTOR CWB1 is studied by comparing the differences between warrant price and market price using Wavelet Neural Network and the conventional Black-Scholes models.It shows that Wavelet Neural Network model is better than the conventional Black-Scholes model in the simulation of Warrant Price.