基于股市网络中心性的投资组合构建与分析  

Construction and Analysis of Portfolio Based on Stock Market Network Centrality

作　　者：李晨辉 舒子宸 陈俣睿 李双宏 肖雅雯 LI Chenhui;SHU Zichen;CHEN Yurui;LI Shuanghong;XIAO Yawen(Department of Research and Development,Orient Securities Company Limited,Shanghai 200010;Melman School of Public Health,Columbia University,New York 10032;School of Electronic Information and Electrical Engineering,Shanghai Jiaotong University,Shanghai 200240)

机构地区：[1]东方证券股份有限公司系统研发总部,上海200010 [2]哥伦比亚大学梅尔曼公共卫生学院,纽约10032 [3]上海交通大学电子信息与电气工程学院,上海200240

出　　处：《计算机与数字工程》2021年第12期2420-2424,2475,共6页Computer & Digital Engineering

摘　　要：分别采用归一化互信息和Pearson相关系数两种指标衡量A股市场股票价格波动相关性,并利用阈值法和PM⁃FG算法来构建股票市场网络。通过对所构建网络模型的比较和分析,发现基于Pearson相关系数的阈值筛选法表现较优。基于所构建的A股市场时序动态网络,对度中心性进行研究,通过设计度中心性策略并进行实证性分析,得到了市场表现更好的基于社团分析的度中心性策略。In this paper,the normalized mutual information and Pearson coefficient are used to indicate the relevance of the fluctuation of A-share market stocks'price.As for constructing stock market networks,the threshold approach and PMFG algorithm are introduced separately.Through the comparison and analysis of the constructed network models,it is found that the threshold screening method based on Pearson correlation coefficient performs better.Based on the constructed time-series dynamic network of the A-share market,the degree centrality is studied.Through the design of degree centrality strategy and empirical analysis,a de⁃gree centrality strategy based on community analysis with better market performance is obtained.

关 键 词：股票市场 复杂网络 度中心性 

分 类 号：F83[经济管理—金融学] TP301[自动化与计算机技术—计算机系统结构]