| 857 | 21 | 172 |
| 下载次数 | 被引频次 | 阅读次数 |
股票收益波动具有典型的连续函数特征,将其纳入连续动态函数范畴分析,能够挖掘现有离散分析方法不能揭示的深层次信息。本文基于连续动态函数视角研究上证50指数样本股票收益波动的类别模式和时段特征:首先由实际离散观测数据信息自行驱动,重构隐含在其中的本征收益波动函数;进一步,利用函数型主成分正交分解收益函数波动的主趋势,在无核心信息损失的主成分降维基础上,引入自适应权重聚类分析客观划分股票收益函数波动的模式类别;最后,利用函数型方差分析检验不同类别收益函数之间波动差异的显著性和稳健性,并基于波动函数周期性时段划分、图形展示和可视化剖析每一类别收益函数在不同时段波动的势能转化规律。研究发现:上证综指股票收益波动的主导趋势可以分解为四个子模式,50只股票存在五类显著的波动模式类别,并且五类波动模式的特征差异主要体现在本次研究区间的初始阶段。本文拓展了股票收益波动模式分类和差异因素分析的研究视角,能够为金融监管部门管理策略的制定和证券市场的投资组合配置提供实证支持。
Abstract：There is a typical character of continuous function existing in the volatility of stock shares. If analyzing its trajectory under continuous dynamic function domain,we can excavate more in-depth information,which cannot be revealed by the existing discrete analysis. The paper exploits the category patterns and timevarying characteristics of returns volatility of sample shares in Shanghai stock exchange 50 index( SSE50)based on the continuous dynamic functional perspective. Firstly,to construct the intrinsic volatility function driven by information implied in actual discrete observation. Further,to decompose the dominant mode of returns volatility function orthogonally using functional principal components. Thirdly,to perform adaptive weight clustering analysis on the fluctuation mode of stock returns based on the dimensionality reduction of principal components without losing core information. Lastly, to test the robustness and significance of difference among different volatility patterns via functional analysis of variance,then display graphically and analyze visually groups' potential energy transformation rule in different intervals based on periodic divisions.The Empirical results show: 1) the dominant mode of returns volatility function of SSE50 can be divided into four sub-models. 2) 50 sample stock shares have five kinds of significant volatility patterns. 3) The characteristic differences of five volatility patterns are all reflected at the initial stage of this research interval.This paper expands the research perspective by classifying returns volatility modes of stock shares and analyzing difference factors,which can be the empirical support for the financial regulatory department to formulate management strategy and for the stock market to allocate portfolio.
[1]Ramsay J O,Ramsey J B.Functional Data Analysis of the Dynamics of the Monthly Index of Nondurable Goods Production[J].Journal of Econometrics,2002,107(1/2):327-344.
[2]严明义.网上拍卖竞买者出价水平的动态演变模式研究[J].统计研究,2010,27(3):59-65.
[3]Tsay R S.Some Methods for Analyzing Big Dependent Data[J].Journal of Business&Economic Statistics,2016,34(4):1-47.
[4]严明义.函数性数据的统计分析:思想、方法和应用[J].统计研究,2007,24(2):87-94.
[5]Hong M.Potential Applications of Function Data Analysis in High-frequency Financial Research[J].Journal of Business&Financial Affairs,2013(2):23-25.
[6]Müller H G,Sen R,Stadtmüller U.Functional Data Analysis for Volatility[J].Journal of Econometrics,2011,165(2):233-245.
[7]Liebl D.Modeling and Forecasting Electricity Spot Prices:A Functional Data Perspective[J].Annals of Applied Statistics,2013,7(3):1562-1592.
[8]Kokoszka P,Reimherr M.Predictability of Shapes of Intraday Price Curves[J].Econometrics Journal,2013,16(3):285-308.
[9]Shang H L.Functional Time Series Approach for Forecasting very Short-term Electricity Demand[J].Journal of Applied Statistics,2013,40(1):152-168.
[10]Shang H L.Forecasting Intraday S&P500 Index Returns:A Functional Time Series Approach[J].Journal of Forecasting,2017,36(7):741-755.
[11]Kokoszka P,Young G.KPSS Test for Functional time Series[J].Statistics A Journal of Theoretical&Applied Statistics,2016,50(5):957-973.
[12]Kokoszka P,Miao H,Zhang X.Functional Dynamic Factor Model for Intraday Price Curves[J].Journal of Financial Econometrics,2015,13(2):456-477.
[13]岳敏,朱建平.基于函数型主成分的中国股市波动研究[J].统计与信息论坛,2009,24(3):52-56.
[14]郭均鹏,孙钦堂,李汶华.Shibor市场中各期限利率波动模式分析-基于FPCA方法[J].系统工程,2012(12):88-92.
[15]Wang Z,Sun Y,Li P.Functional Principal Components Analysis of Shanghai Stock Exchange 50 Index[J].Discrete Dynamics in Nature and Society,2014(3):1-7.
[16]王德青,等.函数型数据聚类分析研究综述与展望[J].数理统计与管理,2018,37(1):51-63.
[17]Wang J L,Chiou J M,Mueller H G.Review of Functional Data Analysis[J].Statistics,2015(1):1-41.
[18]吴晓求.大国金融中的中国资本市场[J].金融论坛,2015(5):28-35.
[19]Kokoszka P,Reimherr M.Introduction to Functional Data Analysis[M].London:Chapman and Hall/CRC Press,2017.
[20]Ramsay J,Hooker G,Graves S.Functional Data Analysis with R and MATLAB[M].NewYork:Springer Press,2009.
[21]Chiou J M,Li P L.Functional Clustering and Identifying Substructures of Longitudinal Data[J].Journal of the Royal Statistical Society,2007,69(4):679-699.
[22]Jacques J,Preda C.Functional Data Clustering:A Survey[J].Advances in Data Analysis&Classification,2014,8(3):231-255.
[23]王德青,刘晓葳,朱建平.基于自适应迭代更新的函数型数据聚类方法研究[J].统计研究,2015,32(4):91-96.
[24]Cuevas A,Febrero M,Fraiman R.An Anova Test for Functional Data[J].Computational Statistics&Data Analysis,2004,47(1):111-122.
[25]Zhang J T,Liang X.One-Way Anova for Functional Data via Globalizing the Point-wise F-test[J].Scandinavian Journal of Statistics,2014,41(1):51-71.
[26]吴晓求.股市危机:结构缺陷与规制改革[J].财贸经济,2016,37(1):22-32.
(1)本文后续通过图形对比发现,粗糙惩罚的收益函数均值曲线与上海证券交易所发布的月度K线在快速上涨、断崖式下跌、政府救市、市场稳定等多个趋势转变的点位极为吻合。事实上,自2015年5月A股市场结束“二八分化”格局后迎来普涨行情,受金融、地产、煤炭、石油石化等权重股发力,2015年5月19日、2015年5月22日逾200只个股连续涨停。然而,2015年6月始受美联储的加息预期导致的资产价格重新定位影响,全球股市集体大跌,A股市场出现多次暴跌事件,特别是2015年8月24日大盘几近跌停。之后受政府救市政策影响,市场短期稳定。后期受救市反向操作行为影响,市场出现第二波幅度较小的下跌,之后政府和监管部门加大打击违规行为力度,并清理场外配资等行为,股市危机逐渐淡出。上述波动过程反映在惩罚收益函数曲线中则是前期明显的两个峰谷波动和后期逐渐平稳的轨迹。
(1)由于第2类股票仅包含“东方证券”(600958)一只股票,不存在类别内部不同股票之间的收益波动差异,因此图9仅展示另外4类股票在不同时段的收益波动信息。
基本信息:
DOI：10.19343/j.cnki.11-1302/c.2018.09.007
中图分类号:F832.51
引用信息:
[1]王德青,何凌云,朱建平.基于函数型自适应聚类的股票收益波动模式比较[J].统计研究,2018,35(09):79-91.DOI:10.19343/j.cnki.11-1302/c.2018.09.007.
基金信息:
国家自然科学基金“函数型数据的自适应分类预测方法及其在金融高频预测中的应用”(71701201);; 教育部人文社会科学基金“金融市场的函数型数据挖掘方法与应用研究”(15YJCZH162);; 2016年度全国统计科学研究项目“函数型数据自适应聚类分析的方法与应用研究”(2016LY13)的阶段性研究成果;; 江苏省自然科学基金青年项目“金融高频数据的函数型自适应分类预测方法研究”(BK20170268);; 中国博士后科学基金项目“函数型数据挖掘理论、方法与应用”(2015M571839);; 中央高校基本科研业务费专项基金“基于函数型数据分析的大数据挖掘方法及其经济管理应用”(2015WA01)的资助
2018-09-25
2018-09-25