| Securities investment funds as a type of investment instruments, has been widelyaccepted by investors. The scale of global mutual fund reached$25.92trillion at the end of2011. Securities investment funds in China made dramatically rapid development in recentyears. There was more than800funds at the end of2011, and the asset value was increasedrapidly. However, with the rapid development of China’s fund industy,some problems hadexposed. After2008, due to the shock of the stock market, the performance of the fund wasunsatisfactory. At the end of2011, the net asset management of fund in China has shrunk15.16%. So when the fund industy is downturn, it’s necessary to proposed a method that canbe applied to choose a valuable fund.Based on the previous problem, a new mixture kernels function of SVM has beenproposed for fund investment decision. First, in the paper we constructed the linearcombination of kernel function combined with polynomial kernel and RBF kernel functions.Then a new kernel function called Multiple Widths RBF kernel function has been proposed toimprove the accuracy of the RBF. Second, in order to solve the problem of unbalance data, amulti-param eter selection criterion based of F-m easure has been given. Third, based on theresearch on domestic and foreign mutual funds performance evaluation, we establish indexsystem, which is suitable for our present market condition. The index system is composed offive aspects: return capalibity, risk level, security selecting and timing ability, investmentportfolios characteristic.The method proposed in this paper are used to evaluate the performance of the39stockopen-ended funds from2008-2011. The result are follows:(1) For unbalanced data problem,F-m easure can improve the accuracy of SVM;(2) The effect of the fund classify based onmixture kernels function SVM is better than others;(3) We revised guidelines in order tochoose a fund with less loss, and put it into pratice;(4) We present several advices aboutinvestment. |
PDF Full Text Request