An Enhanced HS300 Index Tracking Model Based On Random Forest Algorithm

Enhanced index tracking strategy is to control the tracking error of the benchmark index under the premise of pursuing performance beyond the benchmark index,taking the benefits of both Beta and Alpha into account.From the perspective of investment concept,index enhancement fund is an organic combination of active investment and passive investment.In recent years,Index-Enhanced funds have attracted the attention of institutional and individual investors because of their low rate,high transparency and excellent long-term returns.At the same time,the increase in available data and technological progress provide a broad space for investment in R&D.In this paper,Random-forest Algorithm in data science field is introduced into stock selection framework,and its high-quality classification ability is used to screen high-quality stocks,and the low volatility anomaly in the stock market is applied to allocate stock positions in the portfolio in order to obtain stable and considerable returns.The first part of this paper studies the development process,current situation and trend of domestic public offering Index-Enhanced funds.Judging from the quantity and scale,the domestic index enhancement fund is in the ascendant channel.By the end of 2018,the number of public index enhancement fund reached 106,and the scale reached 57.5 billion yuan.In terms of performance,most of them beat the tracking index,and the annual tracking error remained within 8%.It has good development momentum and configuration value.The ways of strengthening Index-Enhanced funds can be divided from top to bottom into position control(mid-long term timing),industry rotation,style rotation,stock selection and so on.The results of performance attribution show that stock selection contributes most to the excess return of the fund.Quantitative stock selection is an effective way to enhance the size of public funds with higher requirements for wind control.The relative maturity of China's financial market still needs some time,which makes it possible for Index-Enhanced funds to obtain alpha returns.In the future,the investors of A-share market will be institutionalized,and the market beta will have strategic allocation value when the current market is undervalued.The index enhancement fund with ‘alpha + beta' investment target has met with good opportunities and has research value.The second part of this paper constructs the basic framework of the Random-forest stock selection model,including the establishment of factor bank,the use of R-F to get factor importance score,the combination of factor importance score to score the stock and industry screening.The factor bank includes 46 factors,such as fundamentals,technical indicators,valuation categories and market sentiment.Through the factor importance score of the output of the R-F classification algorithm,we can see that the evolution of the income factor has changed from momentum,quality,dividend factor to value,valuation,volatility,liquidity and other factors.Among them,the basic factors such as net interest rate and revenue growth rate are becoming active,which indicates that domestic investors are becoming more mature and begin to pay attention to value investment.The factor importance score is used to score the stocks in the stock pool,and the number of stocks selected for each industry is set according to the industry distribution characteristics of Shanghai and Shenzhen 300 constituent stocks.Thirty stocks with higher score are selected by industry.For the selected stocks,the author made a simple equal weight position performance test,which lasted from April 5,2012 to September 28,2018.Regarding the stability of strategic earnings,the annual cycle of the strategy portfolio is 100% higher than that of Shanghai and Shenzhen 300 in the past seven years.Based on the seasonal cycle,the strategy portfolio rose in 16 quarters in 26 quarters,with a quarterly winning rate of 61.54%.The total return of the strategy portfolio is 71%,and the annual return is 7.3%.Over the same period,the total return of HS300 was 40.81%,and the relative return of strategy combination was 30.91%.Stock selection yields are considerable.The last part of this paper,based on the low volatility anomaly in behavioral finance,adjusts the portfolio positions of stocks selected by R-F Algorithm according to the low allocation proportion and the large excess proportion of stocks with small historical volatility.From April 2012 to September 2018,the total rate of return was 117%,the excess rate of return was 77%,and the annual rate of return was 11.34%.Compared with the equal weight portfolio,the return of low volatility portfolio is better.It can be seen that the low volatility factor itself is an alpha factor and can obtain excess return.For portfolio risk indicators,the annual volatility of low volatility portfolio is 22.28%,which is lower than that of equal weight portfolio.On the Beta coefficient,the equal weight portfolio is 1.01,while the low volatility portfolio is less than 1 and 0.93.It effectively controls the volatility of the fund curve and meets the low risk preference of Index-Enhanced funds.Generally speaking,the enhanced HS300 strategy of low-volatility-Random-Forest in this paper provides a new idea for investment management.