The prediction of applying Smooth Support Vector Regression and Back Propagation Network in mutual fund performance

This study used smooth support vector regression and back propagation network as the basic theory in study of the mutual fund performance prediction. This paper used return on performance and return on market to make a comparison, and through the risk values, explored each modelpsilas advantages and disadvantages. This study used Taiwanpsilas equity fund as the prediction target, the validation study period was January 2004 to December 2004. The empirical results showed that the SSVR application and BPN application can both increase return on investment, and will receive an even better return in the bull market. In addition, applying SSVR prediction model, in the bear market, will also result in excess return, and reduction of the investorspsila loss. This study thinks that with smooth support vector regression model and back propagation networking model respectively, according to different risk preferences of investors, investors can, based on their personal risk preferences, choose a suitable prediction model in order to create the excess return in line with the expectations.