Research On The Application Of Statistical Arbitrage Based On Cointegration In Stock Market

For the whole year of2011, China's stock market remained in the doldrums with the market index repeatedly falling below the lowest point, and the total market value has suffered heavy losses. In the turbulent stock market, investors always want to be able to get a stable income in the case of low-risk. The statistical arbitrage discussed in this paper is a market neutral investment strategy, and it can get a stable return whether in a bear market or bull market. In early2010, China has introduced the stock index futures and the pilot margin trading, which means the exiting of short-selling mechanism in China's market and creates conditions for the implementation of statistical arbitrage.The first half of this paper focuses on the theories and methods of statistical arbitrage. The second chapter describes the definition of statistical arbitrage, the difference between risk-free arbitrage and statistical arbitrage, statistical arbitrage strategy, as well as domestic and international research status about statistical arbitrage. The third chapter describes the measurement methods related to the implementation of. statistical arbitrage, including smooth, unit root tests, co-integration and the Ornstein-Uhlenbeck process.In empirical study, we select banking stocks as the object of study which occupies a relatively large share in the A-share market. When constructing the trading signals, we also consider the case that the spread satisfies the Ornstein-Uhlenbeck process in addition to using non-parametric method. In the case of non-parametric method, there are three arbitrage opportunities with the total yield of11.49%in a simulated period. While in the case of spread satisfying the Ornstein-Uhlenbeck process, there are four successful arbitrage opportunities with the total yield of15.42%. Results of two kinds of arbitrage strategy are also superior compared with the Shanghai Composite Index, the. Sharpe ratio are0.78and1.26, and the beta coefficients are0.1252and0.0669which are much smaller than1. All of these fully reflect the market neutral characteristics of statistical arbitrage.As we can see from the theoretical and empirical analysis, statistical arbitrage is feasible in China's stock market. Statistical arbitrage has been widely used in foreign markets, mainly due to lax supervision and flexible operations. With the launch of short-selling mechanism and the improvement of the China's market system, statistical arbitrage will surely be favored by domestic investors.