Algorithm Research And System Design For Spread Arbitrage Process In The CSI-300 Futures Market

stable, I use window-moving prediction methods to handle the data out of sample and then predict the fluctuations of price spreading sequences. The Matlab software can timely deal with data, doing the judgment, and then get the empirical results of data out of sample. During using the window-moving prediction method, the size of window and moving frequency together determine the arbitrage opportunities and accumulated return rate. Meanwhile, good arbitrage results could be gained after optimizing the two coefficients above. The final average return rate is slightly higher than the one of data in sample.The stock index futures can be defined as a futures contract where the underlying asset is a stock index. With stock index futures,investors can effectively avoid systematic risks as well as put forward numerous investment strategies. For instance, two stock assets as well as series of stock index future contracts can be used for hedging. Until now, arbitrage function of stock index futures has aroused extensive concernamong investors all over the world. Thus, this paper focuses on the study of algorithm research and system design for spread arbitrage process in the CSI-300 Futures Market based on the statistical arbitrage thoughts.In this paper we first introduce thecalendar spread strategies and models of stock index futures. Then we construct a programming trading algorithm and conduct an empirical analysis by using the one minute high frequency data of near-month contract and far-month contract of CSI-300 stock index futures. Finally we use Matlab to accomplish the target of achieve automatically judging whether the arbitrage opportunities exist and dynamically calculating the return rate.As to the programming trading algorithm we design two strategies. Later as to empirical study,we test the co-integration relationships among the data in sample and prove that the differences between contracts’prices have mean regression. By using the coefficients resulting from co-integration test regression as the ratio of trading positions and choosing the residual sequences of regression to judge whether arbitrage exists, I get the opening position, closing out position and cutting losses position based on the normal distribution. After calculating the transaction times and accumulated return rate, I put forward the arbitrage trading strategies. As to system implementation by Matlab, because the futures price sequences are not...