The Research On The Correlation Of China Stock Market Based On Wavelet Analysis

With the continuous development of the Chinese stock market, the market segmentation is becoming more and more serious, domestic shares and foreign shares, tradable shares and non-tradable shares have been coexisting for long-term. The segmention is linked to a series of the policy direction in China’s reform and opening, such as the focus on attracting foreign investment, the restructing state-owned to list and so on. And this artificial market segmentation is more and more becoming the obstacle to market price discovery mechanism and securities’rational pricing. Then it is a big negative impact to the development of the entire stock market. Therefore, the accurate mastery of the dynamics correlation between the various markets, is not only provide important basis for the capital asset pricing, risk management, the optimal portfolio and so on, and provide a theoretical basis for the investors to make cross-market investment, besides, it will also help to improve China’s financial market efficiency of resource allocation, and for the integration process of China’s securities market, for the healthy development of China’s securities market has great practical significance.This paper uses the maximal overlap discrete wavelet transform method, and chooses the weekly closing price of Shanghai A index, Shanghai B index, Shenzhen A index, Shenzhen B index and Hengsheng China Enterprise index as the research objects. Meanwhile, uses LA((least asymmetric) wavelet filter and scale filter to make wavelet transform, in order to study the time-varying correlation between A shares, B shares and H shares markets. The empirical results show that(1) Wavelet variance can be a good measure of the risk of each index’s return series in different holding periods, besides, the different time scales corresponding to different holding period for the index return series of wavelet variance is different, that is to say, the risk of holding the same stock is different in different holding periods.(2) Wavelet correlation can be a good measure of the time-varying of the correlation coefficients between different index series. Besides, the different time scales corresponding to different holding period for the index return series of wavelet correlation coefficient is different, and then show the different segmentation of different markets. As a result, it provides a very important theoretical basis for the investors’investment portfolio allocation and optimization. Finally, this paper points out the problems and future directions for further research.