Multi-resolution Wavelet Neural Network In The Application Of The Stock Market Forecast

The stock market is the national macro-control, important areas of direct financing, which is proportional to the benefits and risks. Since the stock price random walk characteristics, by domestic and international economic and political changes, many factors influence investors and other psychological characteristics, it is difficult for scientific evaluation. Factors Affecting Share Prices complex, is highly non-linear, using traditional statistical model has many characteristics of high noise and nonlinearity of the stock price prediction difficult to achieve the desired results. Therefore, the establishment of a high accuracy, streamlined and practical evaluation model has great practical significance for the national macro-control and investments in marketable securities.For the current stock price prediction, based on stock trends, graphic shapes, the vast majority of investor sentiment indicator used in technical analysis, since a number of analytical methods, the lack of scientific support for the theory of the system, and the independence of each index is strong, so the prediction accuracy It is not high. Artificial intelligence is a master of science, covering computers, psychology, image processing and other knowledge in recent years has made a breakthrough and a wide range of applications, the neural network is a branch of artificial intelligence, wavelet neural network is based on neural network based on the introduction of wavelet analysis of its transformation, both non-linear neural network approach, self-organized learning, simplicity, structure, etc., while both black box discerning wavelet analysis can greatly enhance the accuracy of forecasting stock prices degree.2015, the stock market experienced a lever mad cow, thousand shares daily limit, limit, suspend, the government hastily introduced measures to rescue the market, which is destined to be an extraordinary year. This paper introduces in detail after the stock market crash of 2015, from a macroeconomic point of view of the causes of the stock market crash, and then describes the relevant background knowledge, including the basics of stock market forecasting methods, the share price at this stage, the neural network and wavelet analysis of relevant concept, the basic characteristics of wavelet neural network and the specific classification and wavelet neural network system described.In the empirical part of this paper, first, the data pre-processing, and the establishment of multi-resolution wavelet neural network model, according to the characteristics of the sample data on the number of nodes in the network layers, the training parameters set in 2014 to 339 in 2015 Shanghai Composite index trading day for the study, with 311 data before the network is trained,28 data after use as a test sample, the error rate established evaluation criteria-based model of the 2015 stock market crash fluctuation situation analysis and forecast. The results show that China’s Shanghai Composite Index was not haphazard, but predictable, certain operating rules; multi-resolution wavelet neural network for the stock price prediction small error rate of the data, forecast good results, the promotion of high value.