The Research And Empirical Analysis Of Co-intergartion Arbitrage Strategy

With the advent of the shorting time, quantitative investment occupies a more and more important role in financial markets. Statistical arbitrage as effective means of quantitative investment attracted much attention in China. The study of statistical arbitrage strategy has important practical significance.This paper through study and analysis of application of strategy of co-integration arbitrage in the domestic market in a detailed analysis and empirical research. Put forward effective set during the process to pick stocks and identifying of two kinds of errors. In the case of cannot avoid errors. Avoiding the first kind of error and reducing the second category of error by statistical method. Using PCA and constructing factor model, the pick of stocks will be available to all industry. AS sample space to expand, the alternative asset pool is closer to the real market. According to the model of the common trend, the stock factor model is established. When the associated factors of the stock returns is strong, and special share price points sequence is stationary series, the relationship turns the co-integration. In addition, the paper considers model cycles. Through real-time adjustment in the process of empirical model parameter adjustment cycle of look back and the days of look back to select the optimal parameter.Empirical shows that although there is no stock industry classification by human will. But through the principal component analysis of quasi matching stock still has a strong industry clusters. This also from the mathematical level explains the direct use of the applicability of the classifying stock industry. Through principal component analysis can also get the same pair of stock has a very high yield. But there are also matched fewer problems. After expanding t the sample space, the cumulative yield and the average yield are improved, and the number of transactions fell.