Research On Momentum Effect Investment Strategy Of FOF Funds Based On Hidden Markov Model

As a special fund,the fund in the fund(FOF)plays an important role in the financial market.Although FOF funds started late in China market,with the rapid growth of the domestic fund market and the official landing of China Securities Regulatory Commission's "Operational Guidelines for Public Offered Securities Investment Funds No.2-Fund Guidelines in Funds" in 2016,public offered FOF is expected to develop into an important type of investment,with broad market prospects.In this context,how to effectively select high-quality investment targets from many funds and how to carry out corresponding asset switching in different market cycles has become an urgent problem to be solved in the process of expanding FOF fund business and improving investment performance.Based on Hidden Markov Model(HMM)timing method,this paper proposes a scheme for momentum investment strategy of FOF funds: using momentum effect selection strategy combined with HMM-based timing judgment of market trends to form a complete base selection+timing asset allocation scheme,in order to achieve better investment returns under the condition of small investment withdrawal.For this reason,the plan first takes momentum effect theory as the guiding basis for the selection strategy,and then ranks the funds according to the performance of different selection indicators in the sample period,and chooses the funds with the best performance of each indicator according to the ratio of the first 2.5% to form the portfolio.At the same time,it arranges the holding periods of monthly,quarterly and semi-annual to follow regularly in the whole measurement interval.The above-mentioned methods change the portfolio,and observe the momentum effect performance of different types of funds in different base selection indicators and different positions periods from the perspective of absolute and relative returns,so as to select the best base selection strategy of parameter portfolio.Secondly,the hidden Markov model is used for market timing.In this paper,the best combination of model parameters is selected by comparing the absolute and relative returns of sample period length and the number of hidden states.Then theparameters of the hidden Markov model are trained by selecting the optimal sample period length of the day logarithmic return difference of the Shanghai stock index,the five-day logarithmic return difference of the Shanghai stock index,the logarithmic high-low price difference of the Shanghai stock index,the daily trading volume of the Shanghai stock index and the logarithmic financing balance difference of the two cities as the observation vectors of the hidden Markov model,and the parameters are updated periodically according to the monthly interval.The market status of each trading day in the forecasting period of the model is forecasted day by day.Finally,based on the above ideas,the paper establishes fund categories with good momentum effect as the main positions,a small number of funds with weak correlation with the market trend,and funds with little momentum effect as the secondary positions,and chooses the optimal base selection index and the holding period of two types of positions as the base selection strategy in the Hidden Markov Model for predicting the state of market rise.According to the relatively poor performance of momentum effect,the fund category which is insensitive to market fluctuation is the main position,and the fund category which has small momentum effect and relatively small fluctuation is the secondary position strategy.The optimal base selection index and position period of two types of position assets are selected to form the hidden Markov model's base selection strategy to reduce risk in forecasting the decline of the market.Under the above-mentioned investment scheme,in the period from January 1,2012 to June 1,2018,under the strategy of 80% mixed-base + 20% bond-base(up)/80% bond-base + 20% mixed-base(down),the strategy of "IR(Bund)+XR(Hybrid)" which is composed of the monthly information ratio basis selection index of bond funds and the monthly X ratio basis selection index of Mixed-fund is a six-and-half-year forecast.In the interval,the average annual return is 15%,and the excess return is close to 11%.Meanwhile,the maximum withdrawal is controlled in the interval of 10%.According to the scheme design proposed in this paper,first of all,we think that momentum effect does exist in the securities investment market,but it lasts for a shorttime.We can use this effect as a reference for short-term investment.In the specific investment process,we can use momentum effect to show different types of funds in different market cycles to carry out asset switching in order to achieve stable growth of returns and smaller returns.Withdrawal level.Secondly,as a pattern recognition method,hidden Markov model can achieve better results in predicting market trends.In the specific operation process,due to the volatility of the securities investment market,the sample cycle time span of training set should not be too large,and the model parameters need to be adjusted regularly,so that the dynamic parameters of hidden Markov model can be used to trend the market.Potential prediction can achieve better results.