The Investment Value Analysis Of The Mixed Open-end Funds In Our Country Based On Ensemble SVM Model

Since 16 years ago,the open-end funds come into being.No matter in quantity or scale,they are growing rapidly.Today,hybrid open-end funds have become the main force of open-end funds.Therefore,how to evaluate the fund performance of hybrid open-end funds and take advantage of the performance evaluation to find the investment value of the fund is very important.There are many indicators of fund performance evaluation,such as return indicators,risk indicators,risk-adjusted return indicators and indicators to measure the fund manager's stock picking and timing ability,forming a fund evaluation system.At the same time,because VaR is a better indicator for the downside risk,to get an accurate VaR is very important.So we calculate the VaR of different funds based on different GARCH models.Besides,because SVM model can solve non-linear and non-Gaussian problems well and has high prediction accuracy,based on the formation of the indicators system of performance evaluation,we introduce SVM model to predict the investment value of the fund.In this paper,different models are selected among the GARCH,EGARCH and GJRGARCH models with different distributions after back-testing according to the Kupiec and the AIC statistics.Then we take advantage of the fund performance evaluation indicators to determine whether the fund has investment value the next season?half year and year.This article selects 70 hybrid open-end fund data from October 1,2014 to December 31,2017,and uses the three-year performance evaluation indicators to predict whether it has investment value in the next time period using SVM model and then use AdaBoost algorithm to optimize the predictive effect of SVM model.And the results show that the predictive power on the yearly investment value is the best.