Research On Index Futures Arbitrage In A Nonlinear Framework

The official launch of the first stock index futures in the China Financial Futures Exchange, along with margin trading and short selling in2010have greatly improved the arbitrage mechanisms and flourished the derivatives market in mainland China. Since then researchers have been studying the characteristics and arbitrage techniques in the stock index futures market. Based on existing literature and theories, this paper aims to explore two important issues in stock futures arbitrage in a nonlinear framework.Index tracking is an important procedure in both stock futures arbitrage and index fund management. This paper employs a sampling and optimization method to replicate the HS300stock index. Borrowing the insight of structural risk minimization from Support Vector Machines (SVM), it builds up the optimization function, makes several constraints,and presumes that the weights of the each stock in the portfolio are held constant. Then it solves the problem by using genetic algorithm and SVM to choose the stocks, weights, and parameters of the model. Compared with traditional quadratic programming method, SVM has shown better robustness in predicting test data, which is proof to its good nature of structural risk minimization.Cash-futures basis prediction is another essential part of stock futures arbitrage. Correct prediction of the movement of basis could help investors build the right holdings of portfolios and decide whether to perform cash and carry arbitrage or reverse cash and carry arbitrage. This paper builds a prediction model using stock futures trading data from the previous21trading days and predict the basis on the next day using today’s trading data. In order to compare the performance of linear and nonlinear models, SVM models based on Gaussian RBF kernel function, Laplace kernel function, ANOVA kernel function and Bessel kernel function are created, along with a normal linear regression model.The result shows that nonlinear models could capture the trends of the basis better and predict the movement of the basis more precisely.