Research On The Optimal Portfolio Decision Of Open-end Funds From The Perspective Of Investors

In today's society,more and more investors to the fund as an important investment tool,the open-end funds have obtained more and more recognition as a kind of fund management.This thesis analyzed the optimal size of the fund portfolio,the optimal number of the investment style of the fund portfolio and holding time,and combining the three factors to the integrated interaction analysis,obtained the fund portfolio of higher returns and lower,finally modeled the evaluation modeling on the fund portfolio to market standards.First,through the data collection and data preprocessing,aiming at the existence of more than two years and screening with the benefits comparison benchmark of the open-end fund;Then segmentation processing to the holding time,and the holding time and the investment style was interactive processed and then sieving the part of the worst interactive effect based on average standard deviation of analysis of variance;Then combined with the modern portfolio theory,using the sampling without replacement to study the optimal size of the fund portfolio,the number of samples found within 10 of the fund portfolio has stable and excellent effect,as the the optimal size;Then,we analyzed the standard deviation of returns and Sharpe index of the optimal investment style of the fund portfolio,and found that the optimal numbers of the investment style are seven in this case.The combination of investment style,scale and position at the same time into consideration,introducing portfolio diversification variable V,structure and analysis was carried out on the fund portfolio on the basis of modern portfolio theory and by using the sampling without replacement,get as V = 0.5556 that is the funds portfolio investment style for five and the number of funds for nine,revenue performance high and low risk.After a series of analyses,some preferred fund portfolios were obtained.According to the fund rating data in the market,we use support vector machine,random forest and the neural network algorithm to construct different fund portfolio evaluation models respectively.After compared the three,random forest is selected as the final evaluation model.On the basis of the model,we selected some fund portfolios which were previously selected to grade,and got 6 fund which were selected as the final best investment portfolios.