Research On Fund Characteristics Based On Apriori Algorithm And Gibbs Sampling Methods

The purpose of this study is to construct a comprehensive index system of funds from three perspectives of "fund itself,fund managers and fund companies,from the perspective of big data,and to identify the importance of the indicators in the characteristic system,so as to find out the important characteristic indicators and eliminate the irrelevant(or weak-related)indicators,which are used to provide feature-based reference for the foundation selection of FOF(=Fund of Funds).Based on the initial 62 association rules from three dimensions for funds5 performance,this paper extracts 21 characteristic indicators that are strongly related to fund performance.Based on 16l9 real sample data of bond funds with 62 preliminary basic features.Using Gibbs-sampling-induced stochastic search procedure as a tool,62 preliminary features are used to establish corresponding basic association rules.Under the framework of Markov Chain Monte Carlo(MCMC),300 new random samples are generated by Gibbs Sampling method.Based on these new samples,we extract the correlation characteristic indicators with fiund performance(strong,correlated,and weak correlated).Our findings are summarized as follows:1,Based on 62 primary indicators,we find that there are 21 characteristic indicators strongly related to "good" fund performance,and 18 characteristic indicators strongly related to "poor" fund performance.2.There are three characteristic indicators which are weakly correlated with fund performance indicators.3.In addition,there are 6 characteristic indicators related to the performance of"good" funds and 9 characteristic indicators related to the performance of "bad" funds.Through the research and analysis of fund characteristic indicators,we find that we have the following general rules:Firstly,for the fund itself,the financial indicators of the fund have a strong correlation with the performance of the fund;Secondly,for fund managers,there are strong correlations between individual managers and fund performance;&Thirdly,for fund companies,their operating ability,basic information characteristics,and the characteristics of supervisors have a strong correlation with fund performance.Through the analysis of the indicators,it is found that for the characteristics of the fund itself,only the financial indicators of the fund have strong correlation with the performance of the fund.For the characteristics of the fund manager,the indicators of the individual characteristics and performance characteristics of the fund manager have strong correlation with the performance of the fund.For the characteristics of the fund company,only its operating ability characteristics have a strong correlation between gold performance.