Empirical Research Of Arbitrage Between CSI300Index Futures And300ETF

Huataiborui300ETF fund lists successfully in May2012, which provide the newimpetus for the market of spot-futures arbitrage. Before it listed, Shangzheng50ETF,the Shangzheng180ETF,Shenzheng100ETF always serve as the spot for the arbitragebetween CSI300index futures for their strong correlation to Hushen300. However,there is tracking error problem. Although a little error rate, less than1%, but the stockindex futures trading comes to more than600,000one hand, which could cause tensof thousands of gap.In this paper, we select112trading days of data with the interval of day and5minutes, from May30to November9in2012. Select80days of data as the modelingdata, the rest32days’ data as the model test data. Using the co-integration model first,and then ECM model correction. Finally we get ideal model and the test for detectingdata is successful. The result showed that arbitrage in5minutes is much higher thanthat of in day. Investors are increasingly mature, high frequency arbitrage deal hasbegun to take shape.Here is the detail conclusion. ECM model number of arbitrage, in32days testingdata, the possibility of a arbitrage exists only3. We select one of3as a example forarbitrage, we can get the rate of return amount to138%one year. While the samemodel for5min data the rate could amount to462%.what’s more, the possibility ofarbitrage as high as116out of1156data. We can get the almost same result from thesecond model——spread of simple statistics arbitrage model. The possibility of aarbitrage exists only three (although the same number as ECM’s, but the data is notthe same).We select the one that has maximum spread for the variation of arbitrage.The result is that we fail in the arbitrage, which indicates that the model is not fit forarbitrage by day. On the contrary for5min data, we get124out of1156that has the possibility of arbitrage, one third of these data is same as the ECM model. We chooseone which is not in the ECM model for the verification of arbitrage. We could get thereturn rate as high as570%per year.The conclusion indicates that any price distortion is immediately found byinvestors. The reason is not difficult to understand. The empirical findings suggestthat cost for the arbitrage in ETF transaction is high. It is a bottleneck fordevelopment of stock index futures. Author also gives some advice on this.The model will not fit for every condition. We must pay attention to the transferof risk in the stock index futures arbitrage. Finally we briefly discuss the risk controlproblems.