Exchange Rate Prediction Based On Variational Mode Decomposition And Hybrid Grey Wolf Algorithm

Since October 1,2016,the RMB has been formally incorporated into the International Monetary Fund(IMF)special drawing right(SDR)currency basket,and the function of the RMB international use has been recognized which is an important milestone for the internationalization of the RMB.In order to improve the accuracy of exchange rate prediction,a decomposition-integration model named VMD-PSR-DE-GWO-SVR is established in this paper.First,the latest variational mode decomposition(VMD)algorithm is used to decompose the original sequence into multiple components(modes);the second step,taking into account the difference between each component after decomposition,we select all components automatically using the phase space reconstruction(PSR)based on the C-C method to determine the appropriate input-output form for the SVR.In the third step,the SVR model is used to predict all the components,and the parameters in the SVR model are optimized by the recently proposed hybrid gray wolf(DE-GWO)algorithm.Finally,the prediction results of all components are integrated in order to get the final prediction results.The comparative results of prediction error and income analysis of the model show that the VMD algorithm is more effective than the common CEEMD and WD algorithms;the feature selection is automatically selected by PSR,and the accuracy of the model prediction is improved significantly;in addition,the DE-GWO hybrid optimization has a more robust global optimization ability compared with other single optimization algorithms.The empirical results show that the VMD-PSR-DE-GWO-SVR model presented in this paper has good prediction performance on the three groups of exchange rate data of USD/CNY,EUR/CNY and CNY/JPY,so it can be applied to the practical work of designing trading strategy,hedging the exchange rate risk and making the policy of the central bank.