| Since 1987,when fund products first appeared in my country’s capital market,they have gradually become an investment hotspot for Chinese residents because of their moderate returns and risk characteristics.Fund performance which mainly includes holding returns and exposure to risks,has greatly affected the investment decisions of fund buyers.However,scholars have done amount of research on the influencing factors of fund performance,which couldn’t form a unified standard index system.Nevertheless,confronting with the frequently volatile capital market in recent years,the past research models of fund performance have also exposed certain defects.A new predictive model,which is in line with the actual fund performance,needs to be established urgently,and providing policy advice for multi-party participants in fund transactions will also contribute to the construction and improvement of Chinese financial market.With sorting out some predictive models of fund performance,and concerning on the availability,integrality,objectivity and accuracy of data,the predictive model finally selects20 index from four aspects,adopts the principal component factor analysis method to extract six independent principal component factors of different emphasis.The predictive model also adopts the Logistic binary regression model,and sets a dummy variable with a value of 0-1 distribution,which bases on the event of "whether the sample fund can generate more than the average return ".Next,the change of the possibility of the dummy variable is used as the dependent variable,and the HL test is used to determine the goodness of the model fitness.After employing classification prediction and the ROC test to determine the model’s prediction accuracy,the prediction model of fund performance can be finalized.Combining the modeling conclusion with reality,policy suggestions on fund transactions for the main participants are provided,which might help the development and maturity of Chinese fund market. |
PDF Full Text Request