Research On Statistical Arbitrage Theory And Application For Strategy Developement

Last century, statistical arbitrage had had a period of brilliance and a brilliant investment performance in the capital market of U.S..Due to lack of shorting mechanism, the domestic market had a lot of difficulties to make a difference with many quantitative investment strategies, for example, statistical arbitrage. After Margin Trading and Short Selling and Stock Index Futures had started, this situation eased. Each big brokers and other asset management companies launched deep research on statistical arbitrage, and statistical arbitrage flourished in domestic market.This article makes efforts to research the application of statistical arbitrage strategy in domestic market from the perspective of positivism.On the basis of theoretical knowledge, the article shows the basic principle of statistical arbitrage and the investment results of usual arbitrage strategy by introducing the general situation of statistical arbitrage models and making empirical work.This main body of the article is divided into three sections:The first part mainly expounds the basic principle and usual strategies of statistical arbitrage, including the standard definition of statistical arbitrage, the relationship between statistical arbitrage and no risk arbitrage, arbitrage oppor-tunities inspection, the relationship between statistical arbitrage and effective market inspection and the overview of statistical arbitrage models.The second part for statistics introduces several statistical arbitrage strategies based on statistical method:mean reversion strategy, the multiple factor model, cointegration arbitrage, forecasts based on granger causality test and principal component arbitrage. The paper discusses the principle and method of cointe-gration arbitrage in details, and a group of shares in domestic market are used to prove the above. The results shows that the yield of statistical arbitrage is good, and the application in the domestic market is completely feasible.The third part for data mining introduces the basic principles and processes of chaos time series prediction, involving phase space reconstruction, wavelet analysis and data mining, etc. This paper forecasts the price of stock index futures using support vector machine based on phase space reconstruction and wavelet denoising with a high forecast precision. This paper also built up a statistical arbitrage strategy based on the forecast.