Empirical Research On A Stock Market Based On GARCH Model Statistics Arbitrage

Statistical arbitrage strategy is not the latest financial research findings of the monetary sector, this investment strategy has been well applied in foreign mature capital markets such as hedge fundsinves、investment Banks, it is a market neutral strategy without relying on the economic environment, by using quantity means to construct portfolio, thus to immune of market risk, so as to obtain a stable, low risk of excess returns. With the launch of margin on March 31,2010 and Shanghai and shenzhen 300 index futures officially listed trading on April 16,2010, it also makes the securities market in our country has the short hedge mechanism, all this provides a good platform for the implementation of statistical arbitrage strategy in the stock market of our country. In China, current statistical arbitrage strategy is still in its infancy, the study of statistical arbitrage strategy are often confined to the futures market, this article will study its use in the stock market,2010 years ago, our country’s tock market is a typical single market, can only make more but can’t be empty, with the launch of margin, A-share market can also be shorted, bilateral trade can both make a profit. In the past two years, with the increasingly rise of high frequency data, it reduces the cost of data records and storage, making it possible to study on high frequency data. Low-frequency data has gradually can’t meet the demand of the broad masses of investors for high returns, the use of low frequency data statistics has also encounter a substantial decline, therefore, how to make use of statistical arbitrage on high frequency data has become the future development direction.In this paper, the high frequency data of the stock price from Industrial and Commercial Bank of China and the Agricultural Bank of China in 30 minutes, as the subject matter of margin trading and securities lending, are selected as the research object, firstly, the two time series of stock price correlation analysis has been made, the results show that the correlation coefficient between the two reached 0.993, while the high correlation between the two can’t guarantee co-integration, but at least it can be able to explain the possibility of co-integration is high, secondly, to test co-integration, the results show that the two time series of stock price is the first order list of the whole and a co-integration relationship, so we have the condition of statistical arbitrage. Since the spread volatility financial time is not fixed, but changes with time, Granger proved that if there is a co-integration relationship between the variables, it must have the corresponding error correction expression, this paper selects the GARCH model and ECM model to depict the dynamic variance of two time series spread, then give the arbitrage buying point and exit point, to establish arbitrage trading strategy with the help of MATLAB programming. Finally, the paper respectively used the sample data and external data to inspect statistical arbitrage trading performance of the model, the results show that, without considering the premise of transaction costs, based on the GARCH model to depict the dynamic variance, within the sample data the cumulative yield is 12.81%, with the external data May’s cumulative yield is 3.51%, June’s cumulative yield is 1.4%, July’s cumulative yield is 3.64%, August’s cumulative yield is 2.97%. to sum up, good benefits have been achieved, that proves in our country’s stock market statistical arbitrage strategy is effective.