Research On The Methods Of Measuring The Correlation Between Different Stocks In The Stock Market

The stock market of China was initially built in 1986, and since then it has grown to a market where the trading volume per day exceeds two trillion CNY. The progress in the recent less than 30 years is visible to numerous observers from all the worlds of our planet. As the development and perfection, especially the coming of the bull market since last year, the Chinese people are addicted to invest in the stock market, and thus the need for stock analysis technology is significantly increased. As an important metric to evaluate the macro general market risk and the efficiency of the micro property portfolio, the correlation between different stocks is meaningful to investors. At the same time, the instantaneous increasing or decreasing of different stocks indicates that the correlation may be affected by non-causal factors. Due to this, analysis of the correlation between different stocks, to optimize the currently existing methods and to discover the real correlation between different stocks, is useful.This thesis systematically investigates currently existing correlation analysis methods, including the Granger causal relationship test, linear correlation coefficient method and the method based on Copula functions. Furthermore, this thesis studies the relationship between stocks and the broader market in the stock market and the factors that affect stocks, and uses positive analysis results to verify the existence of the non-causal association. Finally, after fully discovering the shortcoming of exiting correlation analysis methods, this thesis proposes a time series mining method that can as much as possible avoid the non-causal relationship between different stocks by theoretical analysis and positive computation and a method that can eliminate the market impact basing on CAPM model. Based on this method, the method of measuring the correlation has been optimized to make the measured correlation coefficient better reflect the real correlation relationship, and to provide an effective reference for investors on the evaluation of the general market risk and the investment portfolio.