Research On Quantitative Timing Based On Improved MFA-SVM

In recent years,Chinese stock market has been developing and growing,and the degree of public participation and social discussion has been significantly improved.At present,the research of stock market investment has been widely concerned by scholars.What's more,the research of stock timing is one of the important contents in this field.Based on the study of the existing literature,this paper finds the aspects that need to be improved in the stock market timing and puts forward the quantitative timing model based on the improved firefly algorithm and support vector machine(MFA-SVM),aiming to improve the accuracy of stock timing and the return on investment.It is found that the Firefly Algorithm has two defects in dealing with high-dimensional optimization tasks.One is that the low attraction between firefly individuals causes the algorithm to fall into local optimization easily in the early stage of the iteration.The other is that the accuracy of the algorithm is low in the post iteration stage.In view of the above defects,a modified firefly algorithm(MFA)is proposed.Firstly,the concept of minimum attraction is added to increase the possibility of information exchange between fireflies in the early stage of optimization.Then,using the idea of dynamic search,according to the information of the optimal value of the objective function.The firefly adaptively adjusts the iteration step.The empirical results show that the efficiency and stability of MFA are significantly improved.At the same time,MFA is adopted to optimize the penalty parameter of support vector machine in the timing model,which makes the general SVM become a more realistic weighted SVM.At the same time,the MFA also optimizes the parameter ? in the radial basis function,which effectively reduces the complexity of the model.In this paper,MFA-SVM model is constructed by improving firefly algorithm,selecting stock index and optimizing parameters of SVM.Finally,in different stock and stock index situations,the timing accuracy and yield of MFA-SVM model are compared with other strategic models.MFA-SVM timing model performs better,which proves that the quantitative timing model can better improve the stock investment returns.At present,although this paper failed to combine the technical indicators with the macro data,the comprehensive interpretation ability of the data also needs to be improved.However,this paper applies MFA and SVM to the model construction of timing strategy,which has a certain reference value to further supplement the diversity of timing model and the practice of stock timing.