Support Vector Machine In The Exploration And Application Of The CTA Strategy

Recent 30 years, thanks to the statistics, data mining, artificial intelligence and the rapid developed of other advanced technology, the securities market prediction research develops a large number of nonlinear classification prediction method.There are a lot of methods for classification and regression.Chaos dynamics , fractal theory, neural network and wavelet analysis are typical of it. Although these methods hold dominant position in the practical application.However people still found that there are many shortcomings.Compared with traditional method ,support vector machine(SVM) overcomes the large sample requirement of traditional method and has strong extension and generalization ability.This thesis, by using SVM's strength in tackling small sample problem, trys to set up a model based on support vector machine(SVM) theory to predict the main commodity fu-tures contract price change trend in the future. And joining support vector machine(SVM)to form a new CTA strategy, which verify the prediction effect by making comparison with the original strategy, finally, we try to establish effective futures investment strategy on this model.The empirical results show that SVM's forecast effect is better in the futures data.And the improved CTA strategy combined with SVM perform better in investmen-t.At the same time,through the construction of the portfolio,fully dispersing the risk of the original strategy,while improving the return.