Research On The Prediction Ability Of PMI To Stock Return

PMI index is widely used in the prediction of macro-economy because of its leading and comprehensive characteristics,and PMI has been called the "barometer"of the stock market.Therefore,this paper studies the prediction ability of PMI on the stock return rate,and uses the empirical results to explain whether PMI has the prediction effect on the stock return rate.As the stock market is extremely complex and highly volatile,it is essentially a complex and diverse dynamic non-linear system.It is affected by various factors,which increases the difficulty of prediction.The traditional linear prediction model has certain limitations.Therefore,this paper uses SVM model in machine learning to predict stock return,which is very effective in solving nonlinear,nonparametric and high noise problems.In this paper,firstly,the theory and development process of PMI are described,and the influence mechanism of PMI on stock return is discussed.Then,the SVM model based on PCA is established.The official PMI and Caixin PMI are added to the model respectively to predict the monthly stock index return of CSI 300,and the conclusion is:when the historical transaction information is not processed by PCA,the accuracy rate of historical transaction information prediction is 8 5.5%,86.8%of the official PMI and 88.1%of the Caixin PMI;when the PCA method is used to build the composite index,the prediction accuracy of the direct composite index is 84.1%,85.1%of the official PMI and 89.1%of the Caixin PMI.The empirical results show that both the official PMI and Caixin PMI can predict the stock return,and Caixin PMI is better than the official PMI.Finally,according to the empirical results,this paper puts forward the corresponding policy recommendations,and expounds the shortcomings and prospects of this paper.