Research On Quantitative Fundamental Combination Based On A-share Market

The research of fundamental investment portfolio has been popular for a long time.There have been a lot of researches on the configuration of fundamental data in foreign countries in the last century.There have also been follow-up studies in China at the beginning of the century,many scholars have tried to run foreign configuration methods in the A-share market.This article uses traditional methods and machine learning methods,and conducts relevant research based on the Shanghai Stock Exchange 50 Index of the A-share market.The research window is from 2010 to the end of 2018,and explores the feasibility of deploying fundamentals in the A-share market in recent years.The first part is based on the traditional fundamental weighting method,making corresponding improvements based on the research of relevant literature at home and abroad,and redistributing the weights based on the company's net assets,cash flow,gross sale and monetary funds to construct a fundamental investment portfolio.A variety of construction forms are used,on the one hand to verify the relevant domestic literature,on the other hand to seek a better configuration method,and at the same time to carry out in-depth research on the nature of the combination;the second part uses a non-parametric model,based on 30 fundamental feature attributes,using machines Learning methods to select stocks to build a portfolio,and implementing detailed analysis of multiple periods and the theoretical construction of long and short portfolios.Our long port,folio is based on the fundamental characteristics of the previous period to select low-volatility component stocks in the current positive rating,expecting its continued profitability,and the short portfolio is based on the fundamental characteristics of the previous period to select high-volatility component stocks in the current positive rating,it is expected that,its income will reverse,and the results show that such hedging income far exceeds the algorithm's long combination,and each period of income is more stable and efficient.The overall empirical research results show that all portfolios have excess returns in the long run.The former portfolios shows low volatility.,The value of of the Jensen Alpha test is generally less than 1.In general,in the upward market,the fundamental portfolio income tendency is not as good as the market,and it's losses in t.he downward market are lower;the algorithm portfolio is characterized by high and high volatility,the trend charact.eristics relative to capitalization-weighted indexes are opposite to the fundamental portfolio.This article attempts to reproduce the fundamental research methods abroad in the A-share market.Compared with the research of domestic scholars,one is to carry out a more comprehensive and in-depth research of portfolio,using a variety of methods,and at the same time will pass the time to recent years,during the period of more complete disclosure,the fundamental data has higher reliability;the second is to combine the machine learning method directly with the underlying fundamental data instead of the integrated fundamental factor data,which reflects the the value of fundamentals data more intuitively.Proving the he construction of the fundamental portfolio has certain feasibility,which has many benefits for the A-share market.On the one hand,for listed companies,it will pay more attention to the company's fundamentals and improve the company's operating quality.On the other hand,for investors,they can better screen quality stocks,form good investment habits,reduce the behavior of chasing up and down,and take the initiative to long-term sustainable investment.Furthermore,for large-scale management,funds,it can also obtain better investment management methods.