| How to fully control risks in investing in FOF and obtain long-term and stable income?In this paper,we construct the support vector machine(SVR),long-term and short-term memory network(LSTM)and SVR&LSTM three prediction models to complete the dynamic screening of the sub-funds,and combine the mean variance model to configure the sub-funds.This will provide investors with the best investment solutions for the pursuit of stable returns.The study time interval for this paper is from January 2016 to March 2019.First,we combine the maximum retracement rate indicator and the industry evaluation index to score the factor risk for each type of assets,select the medium and low risk funds,and construct the fund primary selection pool.Then,we use the income class,risk class,risk adjustment class and fund manager as factors,based on the sliding time window support vector machine model,long-term and short-term memory network model and SVR&LSTM model to predict the fund's predicted excess return rate.In the back-test interval from July 2017 to March 2019,we sorted the predicted values of excess returns in each period to complete the construction of each phase of the fund pool.After completing the dynamic screening of the sub-funds in each period,we configured the mean variance for them,not only the large-scale assets,but also the secondary allocation of the funds in each fund pool.The back-test results show that the FOF products based on the SVR&LSTM model combined with the mean variance model are the best.Regardless of the market's uptrend phase,the highest point to the down phase,and the lowest point to the recovery phase,the cumulative yield of FOF investment clearly outperformed the benchmark cumulative rate of return,with an average annualized rate of return of 21.93%and annualized volatility controlled below 13.9%.The average annualized excess return rate is 17.29%.It is a good choice for investors who are pursuing stable returns,and it also proves the feasibility of product design. |
PDF Full Text Request