The Analysis And The Optimization Of Tail Statistical Arbitrage Strategy Based On The Copula Model

In today's financial markets,stable excess return has been the targets to all kinds of investors.However,Today's financial markets are changing so fast that it is so hard to get long-term,stable excess returns.Under this kind of background,based on the characteristic of stability of the investment performance,the concept of quantitative investment has gradually emerged,which has become a major investment method for the investors in the financial markets.Statistical arbitrate,as a mature strategy of quantitative investment,is widely used in the investment in developed capital market.Statistical arbitrage refers to an arbitrage model which is established by building a quantification model through the knowledge of mathematical statistics,analyzing the inherent law between two kinds of asset price,building a quantitative portfolio based on this,then making profits according to the temporary deviation of the assets' price.Statistical arbitrage has a series of characteristics such as market neutrality and systemic risk aversion,so it's adopted widely by large hedge funds,asset management companies,investment bank and other kinds of institutional investors.Since most of the financial markets in China have been lacking of short selling mechanism,there have been few studies on statistical arbitrage in Chinese market.Along with the constantly development and improvement of Chinese financial market system,the launch of the short selling methods such as securities margin trading and the financial derivatives which have short selling mechanism,the statistical arbitrage strategy already has the room for development.Therefore,in this context,we construct a tail statistical arbitrage strategy based on the Copula model,carry out empirical analysis on the high frequency data of stock index futures,to verify the feasibility of arbitrage strategy.Firstly,this article sorted out the domestic and foreign research literature related to statistical arbitrage and found that most scholars are using the cointegration arbitrage model to do their research.But the cointegration model describes the linear correlations between the variables,which limits the selection of the transaction objects.However,one of the characteristic of the Copula model is that it is able to match the nonlinear correlation between the variables,which makes up for the deficiency of the statistical arbitrage model.Therefore,based on the Copula model,we used it to do the statistical arbitrage research.After reviewing the theory of statistical arbitrage and Copula model,we used the tail correlation coefficient of the Copula model as an index,designed a Copula statistical arbitrage strategy,and then made the empirical test through using 5-minute data of the stock index future,from January 3rd,2017 to December 29 th,2017.By empirical research,the results obtained are as follows:Firstly,based on cointegration model,do the simple statistical arbitrage using the data of stock index future,it showed that the simple tail cointegration strategy turns out to be a loss in almost all kinds of threshold;the simple tail cointegration strategy was not able to get a good arbitrage result.The cointegration strategy based on rolling interval acted better than the simple cointegration strategy,the highest return achieves 0.90%.The Copula tail arbitrage strategy designed by us earned 3.91%,which means the strategy is effective to some extent.Meanwhile,the length of modeling interval showed a great impact on the return of the arbitrage strategy.The return of the Copula strategy decreased gradually,with the increase of the length of rolling interval.After that,we used robustness test in order to verify the usefulness of the strategy.The result of the strategy is 3.21%,which verify the effectiveness of the Copula tail statistical arbitrage strategy.In the fifth chapter of this essay,we used enumeration method to optimize the tail cointegration strategy and Copula strategy.Meanwhile,this paper designed other three kinds of statistical arbitrage strategies based on the tail dependence and made another empirical test.After optimization of the 5 strategies,we found that the yield of the Copula tail arbitrage strategy based on 5-minute data achieves 7.39%;the return of Copula tail arbitrage strategy based on 3-minute data and the one-way Copula tail arbitrage based on 5-minute data are 5.25% and 4.51%;the yield of the Copula arbitrage strategy based on 1-minute data is 3.61%;while the yield of tail cointegration strategy is lower relative to the other three strategies,it's yield is only 1.15%.Then,we did the robustness test again,based on the five optimized arbitrage strategies,using out-of-sample data.The result shows that the return of the arbitrage strategy,based on 1-minute data,although it's only 4.59%,it's 0.98% higher than 3.61%;The return of the one-way Copula arbitrage strategy based on 5-minute data and the Copula arbitrage strategy based on 3-minute data have a slightly decrease,it keeps stable relatively;The other two strategies have a certain decline.Therefore,we make a conclusion that the Copula arbitrage strategies,based on 1-minute data,has a relatively stable performance,it can be used as an arbitrage strategy in the reality;The one-way Copula arbitrage strategies based on 5-minute data and the Copula arbitrage strategies based on 3-minute data still need to be improved in order to acquire a higher return;However,the Copula arbitrage strategies based on 5-minute data and the tail cointegration strategy has a larger decline of return rate,so it may not be suitable for actual arbitrage.The research of text means a lot to the microcosmic aspect such as the arbitrage of the financial assets and the pricing of the stock index future;and it also means a lot to the macroscopic aspect,the system of the stock index futures,for example.However,this essay has some shortcomings,such as the sample data is not big enough,the arbitrage transaction objects do not change and so on,and it needs to be improved in the future.