Research On Stock Investment Strategy Based On Morphological Clustering

According to the annual survey of the World Bank database,the number of listed companies in China have reached to 3,485 in 2017,and it is increasing year by year.The stock investment methods affect the investment income of investors directly.Quantitative investment has the characteristics of high profitability and low risk volatility.It has been decades years in abroad,but it started relatively late in China.In recent years,with the rise of methods such as machine learning,quantitative investment has been growing rapidly.In addition,data mining technology has become a hot issue in recent years.Its application in quantitative investment is very extensive,and it can be used for stock price forecasting,pattern recognition,stock picking,etc.Combining quantitative investment method and data mining technology,this paper studies the problem of quantitative investment stock selection: select the 15-minute closing price data of food and beverage industry,the hierarchical clustering method is used to cluster according to different forms presented by the closing price time series.Subsequently,using the Granger analysis we found the leading and lagging relationship between different types of stock clusters and their block index.Two experimental groups were constructed,in which experimental group 1 did not eliminate the lag group,and experimental group 2 eliminated the lag group.After that,through the factor analysis method,12 financial indicators that can represent the performance of the stock's fundamentals were classified,and the stock selection method was used to separate stocks in the two experimental groups.Finally,select the top 5,top 10 and top 20 stocks to form different stock portfolios and compare their excess returns and risk with the food and beverage industry index and market benchmark index.Through the research we found that using the quantitative investment strategy selected in this paper,the investment portfolio selected by the two experimental groups:(1)Revenue: the stock portfolios constructed by the two experimental groups all outperformed the market and obtained excess revenue,in which the yield of the stock portfolio selected in the experimental group 2 is generally higher than that of the experimental group 1,(2)The risk aspect: the Sharpe ratio value indicates that each unit risk of the selected stock combination can obtain an excess return higher than the market benchmark;The beta coefficient value indicates that the volatility of the selected portfolio is lower than the corresponding industry index standard.Compared with the market benchmark,except for the portfolio with the highest return,the other portfolio volatility is lower than the market benchmark.In addition,the study also found the "best bending window" when measuring stock financial data by using DTW method.Considering comprehensively,under the background of complex data and large amount of information in China's stock market,the quantitative investment strategy constructed in this paper can effectively construct a stock portfolio that outperforms the market and has certain practical value.