基于知识约简与人工神经网络技术的中国证券市场研究
|
摘要
随着计算机技术，特别是人工智能技术的飞速发展，对金融市场的定量仿真与实验成为了可能，于是一门新的分支学科—实验金融学—产生了。实验金融学是指应用计算机技术来模拟实际金融市场(如股票市场)，在既定的市场结构条件下通过研究市场微观层次投资者的行为来揭示市场宏观特性的一门金融学分支。
本文的研究就是从实验金融学的角度着手，把人工智能的重要技术—粗糙集理论与人工神经网络技术应用于我国证券市场的研究当中，以此方法为基础对我国证券市场的有效性及相关问题进行简单的探讨。为此，本文首先对证券市场有效性理论的作了简单介绍；随后对粗糙集理论进行了简单的介绍，并对其用于知识约简的计算方法进行了讨论；然后介绍了人工神经网络的技术原理及其改进算法。在对上述基础理论进行论述的基础上，本文提出了一种粗糙集与人工神经网络相结合的技术方法，并把这一方法应用于我国证券市场的研究当中。
本文利用以上方法对我国证券市场进行了模拟并获得了大量数据，并进行了大量的统计分析，然后结合国内相关的研究结果得出结论；首先，我国证券市场目前尚不满足弱式有效性；其次，我国证券市场股票价格不能反映上市公司的年度财务状况信息。本文认为，我国证券市场股票价格不能反映上市公司年度财务信息的主要原因在于企业及注册会计师行业的诚信危机。为此本文从内部体制、外部监管等多个方面应用现代经济管理理论对可能导致企业及注册会计师行业诚信危机的原因进行了详细探讨，并提出了一系列的改进措施，以进一步提高国内上市公司信息批露环境及信息质量。
The development of computer technology, especially the development of artificial intelligence technology made it possible to simulate and test the financial market quantitatively. So a new branch of finance -experimental finance formed. In experimental finance, computer technology should be used to simulate the real financial market (e.g. Stock market). It aims to disclose the macro characteristic about the market by researching the behavior of the investor in micro-level under certain market structure.
In this thesis, begin with experimental finance, some significant technology-rough sets theory and artificial neural network technology have been applied to the research on financial market. Based on this method, the efficiency of Chinese security market has been discussed simply. Wherefore, the theory about efficient market has been discussed in this thesis firstly; then the Rough Set theory has been introduced, and at the same time, some algorithm used to knowledge reduction have been discussed; in this thesis, the principal of Artificial Neural Network and its improved algorithm are introduced too. After introduce of the basic theory, we posed a method which combine the rough sets theory and artificial neural network and applied it to the research of Chinese security market.
We carried out a lot of static analysis on the data got from simulate about Chinese security market by the method mention pretext. Combine with correlation research, we have get the conclusion: First, the Chinese security market is not a weak efficient market; Second, the information of corporation financial state has not been reflect in the market price of corporation stock. According to this thesis, the reason why financial information has not been reflect in the stock price is that the credit crisis of enterprise and the certified public accountant industry. Aim at this, this article using economic and managerial theory discussed in detail the probable reason for the credit crisis in corporation and CPA industry from many aspects such as the inside structure and the outside supervision. Moreover, it presents a series of ameliorative measure to further promote the information exposure environment and information quality of Listed Company in domestic.
本文的研究就是从实验金融学的角度着手，把人工智能的重要技术—粗糙集理论与人工神经网络技术应用于我国证券市场的研究当中，以此方法为基础对我国证券市场的有效性及相关问题进行简单的探讨。为此，本文首先对证券市场有效性理论的作了简单介绍；随后对粗糙集理论进行了简单的介绍，并对其用于知识约简的计算方法进行了讨论；然后介绍了人工神经网络的技术原理及其改进算法。在对上述基础理论进行论述的基础上，本文提出了一种粗糙集与人工神经网络相结合的技术方法，并把这一方法应用于我国证券市场的研究当中。
本文利用以上方法对我国证券市场进行了模拟并获得了大量数据，并进行了大量的统计分析，然后结合国内相关的研究结果得出结论；首先，我国证券市场目前尚不满足弱式有效性；其次，我国证券市场股票价格不能反映上市公司的年度财务状况信息。本文认为，我国证券市场股票价格不能反映上市公司年度财务信息的主要原因在于企业及注册会计师行业的诚信危机。为此本文从内部体制、外部监管等多个方面应用现代经济管理理论对可能导致企业及注册会计师行业诚信危机的原因进行了详细探讨，并提出了一系列的改进措施，以进一步提高国内上市公司信息批露环境及信息质量。
The development of computer technology, especially the development of artificial intelligence technology made it possible to simulate and test the financial market quantitatively. So a new branch of finance -experimental finance formed. In experimental finance, computer technology should be used to simulate the real financial market (e.g. Stock market). It aims to disclose the macro characteristic about the market by researching the behavior of the investor in micro-level under certain market structure.
In this thesis, begin with experimental finance, some significant technology-rough sets theory and artificial neural network technology have been applied to the research on financial market. Based on this method, the efficiency of Chinese security market has been discussed simply. Wherefore, the theory about efficient market has been discussed in this thesis firstly; then the Rough Set theory has been introduced, and at the same time, some algorithm used to knowledge reduction have been discussed; in this thesis, the principal of Artificial Neural Network and its improved algorithm are introduced too. After introduce of the basic theory, we posed a method which combine the rough sets theory and artificial neural network and applied it to the research of Chinese security market.
We carried out a lot of static analysis on the data got from simulate about Chinese security market by the method mention pretext. Combine with correlation research, we have get the conclusion: First, the Chinese security market is not a weak efficient market; Second, the information of corporation financial state has not been reflect in the market price of corporation stock. According to this thesis, the reason why financial information has not been reflect in the stock price is that the credit crisis of enterprise and the certified public accountant industry. Aim at this, this article using economic and managerial theory discussed in detail the probable reason for the credit crisis in corporation and CPA industry from many aspects such as the inside structure and the outside supervision. Moreover, it presents a series of ameliorative measure to further promote the information exposure environment and information quality of Listed Company in domestic.
引文
[1]．中国证券监督管理委员会：http://www.csrc.gov.cn/cn/statinfo
[2]．张亦春，周颖刚．中国股市弱式有效研究综述[J]．当代财经．2001，8：30～32．
[3]．黄兴，苏卫东．当代证券市场研究的发展趋势[J]．国际金融研究．2003，6：9～14．
[4]. E.Fama. The Behavior of Stock Market Prices[J]. Journal of Business. 1965,1.
[5]. E.Fama. Efficient Capital Markets: A Review of Theory and Empirical Work[J]. Journal of Finance. 1970,5.
[6]. E.Fama. Efficient Capital Markets:Ⅱ, Fiftieth Anniversary Invited Paper. Journal of Finance. 1991,6.
[7]. D.Bond,Theler. Is There Overreaction in Stock Market?[J].Journal of Finance. 1985,7.
[8]. E.Fama,L.Fisher, M.Jenson,et al. The Adjustment of Stock Price to New Information[J].International Economic Review. 1969,2.
[9]．俞乔．市场有效性、周期异常与股价波动—对上海、深圳股票市场的实证分析[J]．经济研究．1994，9：43～50．
[10]．宋颂兴，金伟根．上海股市市场有效性实证研究[J]．经济学家．1995，4：107～113．
[11]．陈小悦，陈晓，顾斌．中国股市弱型效率的实证研究[J]．会计研究．1997，9：13～17．
[12]．沈艺峰，吴世农．我国证券市场过度反应了吗?[J]．经济研究．1999，2：21～26．
[13]．宋冬林．市场有效性理论评析[J]．经济学动态．1995，10：65～73．
[14]．陆金海．市场有效性理论研究综述[J]．经济学动态．2000，6：63～66．
[15]. Beaver.W.H.The Information Content of Annual Earnings Announcements[J]. Journal of Accounting Research. 1968,Sup.
[16]. Ball.R,Brown.P.An Emperical Evaluation of Accounting Income Numbers[J]. Journal of Accounting Research. 1968,Sup.
[17]．杨松．西方学者对证券市场信息有效性的研究[J]．外国经济与管理；1997，(1)：22～26
[18]．梁东，汪静静．知识经济下的企业经营策略探索[J]．商业研究．2001，2：51～53．
[19]．冯俊文．企业知识管理[J]．改革与发展．2000，1：53～56．
[20]．王永庆．人工智能原理与方法[M]．西安：西安交通大学出版社，1998年5月第1版．
[21]．曾黄麟．粗集理论及其应用——关于数据推理的新方法[M]．重庆：重庆大学出版社，1996年9月第1版．
[22]．陈世福，陈兆乾等．人工智能与知识工程[M]．南京：南京大学出版社，1997年12月第1版．
[23]．刘清．Rough集及Rough推理[M]．北京：科学出版社，2001年8月第1版．
[24]. Z Pawlak. Rough Set[J]. International Journal of Computer and Information Science. 1982, 11 (5): 341～356.
[25]. Z Pawlak, J Grzymala-Busse, R Slowinski et al. Rough Sets[A]. Communications of the ACM. 1995, 38 (11): 89～95.
[26]. Z Pawlak. Rough set, Theoretical Aspects of Reasoning about Data[M]. Amsterdam: Kluwer Academic Publishers, 1991.
[27]．王珏，苗夺谦，周育健．关于Rough Set理论与应用的综述[J]．模式识别与人工智能．1996，9(4)：337～343．
[28]．徐德友，胡寿松．一种基于粗糙集的近似质量求取属性约简的决策算法[J]．控制与决策．2003，18(3)：313～316．
[29]. W Ziarko. The discovery, Analysis and Representation of Data Dependencies in Database[A]. Knowledge Discovery in Database[C]. Combrige: AAAI/MIT Press,1990:213～228.
[30]. A Skowron. The Discerniblilty Matrices and Functions in Information System[A]. Intelligent Decisin Support-Handbook of Advances and Application of the Rough Set Theory. Amsterdam: Kluwer Academic Publishers, 1992:311～369.
[31]．常犁云，王国胤，吴渝．一种基于Rough Set理论的属性约简及规则提取方法[J]．软件学报．1999，10(11)：1206～1211．
[32]．唐建国，谭明术．粗糙集理论中的求核与约简[J]．控制与决策．2003，7(4)：449～452．
[33]. A φhm. ROSETTA Technical Reference Manual[H]. Knowledge Systems Group at NTNU, Norway and the Logic Group at Warsaw University, Poland. 2001.
[34]. D.S. Johnson. Approximation Algorithms for Combinatorial Problems[J]. Journal of Computer and System Sciences. 1974, 9: 256～278.
[35]．丛爽．面向MATLAB工具箱的神经网络理论与应用[M]．合肥：中国科学技术大学出版社，2003年5月第2版．
[36]．F.Rosenblatt．Principles of Neuron Dynamics[M]．Washington D．C：Spartan Books， 1962．
[37]．闻新，周露，王丹力等．MATLAB神经网络应用设计[M]．北京：科学出版社，2000年9月第1版．
[38]. M.T. Hagan, H.B. Demuth, M.H. Beale. Neural Network Design [M]. Boston: PWS Publishing Company, 1996.
[39]．《运筹学》教材编写组．运筹学(修订版)[M]．北京：清华大学出版社．1990年1月第2版．
[40]．杨建刚．人工神经网络实用教程[M]．杭州：浙江大学出版社，2001年1月第1版．
[41]. D.E Rumelhart, G.E. Hinton, R.J. Williams. Learning Representation by Back-Propagation Errors [J]. Nature. 1986, 323: 533～536.
[42]．刘向阳，王如云，李未．基于相对误差平方和的BP算法[J]．计算机仿真．2003，20(6)：32～35．
[43]. Math works Inc. Neural Network Toolbox[H].2001.
[44]. Y.Hirose, et al. Back-Propagation Algorithm Which Varies the Number of Hidden Units. Neural Networks. 1991,4(1):61～66.
[45]．李晓峰，刘光中．人工神经网络BP算法的改进及其应用[J]．四川大学学报(工程科学版)．2000，3(2)：105～109．
[46]．焦李成．神经网络系统理论[M]．西安：西安电子科技大学出版社，1990年12月第1版．
[47]. Math Works. Inc. MATLAB documentation[H]. 2001.
[48]. N.W.Charence, E.G.Witting. A Study of the Parameters of a Back propagation Stock Price Prediction Model. The first New Zealand International Two-stream Conf on Artificial Neural Networks and Expert System. 1993:288～291.
[49]．韩力群．人工神经网络理论、设计与应用[M]．北京：化学化工出版社，2002年1月第1版．
[50]. J.Jelonek, K.Krawiec, R.Slowinski. Rough Set Reduction of Attributes and Their Domains for Neural Networks[J]. Computational Intelligence. 1995,11(2):339～347.
[51]．陈遵德．Rough Set神经网络智能系统及其应用[J]．模式识别与人工智能．1999，
12(1): 1～5.
[52]. R.Hashemi, L.A.Blanc, C.T.Rucks, et al. A Hybrid Intelligent System for Predicting Bank Holding Structures [J]. European Journal of Operational Research. 1998,109(2):390～402.
[53]. B.S.Ahn, S.S.Cho, C.Y.Kim, The Integrated Methodology of Rough Set Theory and Artificial Neural Network for Business Failure Prediction[J]. Expert System with Applications. 2000,17(18):65～74.
[54]．朱林，何建敏，常松．粗集与神经网络相结合的股票价格预测模型[J]．中国管理科学．2002，10(4)：7～12．
[55]．刘国良，强义文，麻亮，陈兴林．基于粗神经网络的仿人智能机器人的语音融合算法研究[J]．控制与决策．2003，18(3)：364～366．
[56]．陈可，张琴舜，陈培培等．BP神经网络在证券分析预测中应用[J]．计算机工程．2001，27(11)：95～97．
[57]．侯木舟，韩旭里．基于MATLAB的神经网络在股市预测中的应用[J]．系统工程．2003，21(2)：112～115．
[58]．汪卫霞．对虚假会计信息批露的制度分析[J]．经济问题探索．2003，10：100～104．
[59]．王玉英．会计造假的原因及其治理对策[J]．企业经济，2003，10：184～185
[2]．张亦春，周颖刚．中国股市弱式有效研究综述[J]．当代财经．2001，8：30～32．
[3]．黄兴，苏卫东．当代证券市场研究的发展趋势[J]．国际金融研究．2003，6：9～14．
[4]. E.Fama. The Behavior of Stock Market Prices[J]. Journal of Business. 1965,1.
[5]. E.Fama. Efficient Capital Markets: A Review of Theory and Empirical Work[J]. Journal of Finance. 1970,5.
[6]. E.Fama. Efficient Capital Markets:Ⅱ, Fiftieth Anniversary Invited Paper. Journal of Finance. 1991,6.
[7]. D.Bond,Theler. Is There Overreaction in Stock Market?[J].Journal of Finance. 1985,7.
[8]. E.Fama,L.Fisher, M.Jenson,et al. The Adjustment of Stock Price to New Information[J].International Economic Review. 1969,2.
[9]．俞乔．市场有效性、周期异常与股价波动—对上海、深圳股票市场的实证分析[J]．经济研究．1994，9：43～50．
[10]．宋颂兴，金伟根．上海股市市场有效性实证研究[J]．经济学家．1995，4：107～113．
[11]．陈小悦，陈晓，顾斌．中国股市弱型效率的实证研究[J]．会计研究．1997，9：13～17．
[12]．沈艺峰，吴世农．我国证券市场过度反应了吗?[J]．经济研究．1999，2：21～26．
[13]．宋冬林．市场有效性理论评析[J]．经济学动态．1995，10：65～73．
[14]．陆金海．市场有效性理论研究综述[J]．经济学动态．2000，6：63～66．
[15]. Beaver.W.H.The Information Content of Annual Earnings Announcements[J]. Journal of Accounting Research. 1968,Sup.
[16]. Ball.R,Brown.P.An Emperical Evaluation of Accounting Income Numbers[J]. Journal of Accounting Research. 1968,Sup.
[17]．杨松．西方学者对证券市场信息有效性的研究[J]．外国经济与管理；1997，(1)：22～26
[18]．梁东，汪静静．知识经济下的企业经营策略探索[J]．商业研究．2001，2：51～53．
[19]．冯俊文．企业知识管理[J]．改革与发展．2000，1：53～56．
[20]．王永庆．人工智能原理与方法[M]．西安：西安交通大学出版社，1998年5月第1版．
[21]．曾黄麟．粗集理论及其应用——关于数据推理的新方法[M]．重庆：重庆大学出版社，1996年9月第1版．
[22]．陈世福，陈兆乾等．人工智能与知识工程[M]．南京：南京大学出版社，1997年12月第1版．
[23]．刘清．Rough集及Rough推理[M]．北京：科学出版社，2001年8月第1版．
[24]. Z Pawlak. Rough Set[J]. International Journal of Computer and Information Science. 1982, 11 (5): 341～356.
[25]. Z Pawlak, J Grzymala-Busse, R Slowinski et al. Rough Sets[A]. Communications of the ACM. 1995, 38 (11): 89～95.
[26]. Z Pawlak. Rough set, Theoretical Aspects of Reasoning about Data[M]. Amsterdam: Kluwer Academic Publishers, 1991.
[27]．王珏，苗夺谦，周育健．关于Rough Set理论与应用的综述[J]．模式识别与人工智能．1996，9(4)：337～343．
[28]．徐德友，胡寿松．一种基于粗糙集的近似质量求取属性约简的决策算法[J]．控制与决策．2003，18(3)：313～316．
[29]. W Ziarko. The discovery, Analysis and Representation of Data Dependencies in Database[A]. Knowledge Discovery in Database[C]. Combrige: AAAI/MIT Press,1990:213～228.
[30]. A Skowron. The Discerniblilty Matrices and Functions in Information System[A]. Intelligent Decisin Support-Handbook of Advances and Application of the Rough Set Theory. Amsterdam: Kluwer Academic Publishers, 1992:311～369.
[31]．常犁云，王国胤，吴渝．一种基于Rough Set理论的属性约简及规则提取方法[J]．软件学报．1999，10(11)：1206～1211．
[32]．唐建国，谭明术．粗糙集理论中的求核与约简[J]．控制与决策．2003，7(4)：449～452．
[33]. A φhm. ROSETTA Technical Reference Manual[H]. Knowledge Systems Group at NTNU, Norway and the Logic Group at Warsaw University, Poland. 2001.
[34]. D.S. Johnson. Approximation Algorithms for Combinatorial Problems[J]. Journal of Computer and System Sciences. 1974, 9: 256～278.
[35]．丛爽．面向MATLAB工具箱的神经网络理论与应用[M]．合肥：中国科学技术大学出版社，2003年5月第2版．
[36]．F.Rosenblatt．Principles of Neuron Dynamics[M]．Washington D．C：Spartan Books， 1962．
[37]．闻新，周露，王丹力等．MATLAB神经网络应用设计[M]．北京：科学出版社，2000年9月第1版．
[38]. M.T. Hagan, H.B. Demuth, M.H. Beale. Neural Network Design [M]. Boston: PWS Publishing Company, 1996.
[39]．《运筹学》教材编写组．运筹学(修订版)[M]．北京：清华大学出版社．1990年1月第2版．
[40]．杨建刚．人工神经网络实用教程[M]．杭州：浙江大学出版社，2001年1月第1版．
[41]. D.E Rumelhart, G.E. Hinton, R.J. Williams. Learning Representation by Back-Propagation Errors [J]. Nature. 1986, 323: 533～536.
[42]．刘向阳，王如云，李未．基于相对误差平方和的BP算法[J]．计算机仿真．2003，20(6)：32～35．
[43]. Math works Inc. Neural Network Toolbox[H].2001.
[44]. Y.Hirose, et al. Back-Propagation Algorithm Which Varies the Number of Hidden Units. Neural Networks. 1991,4(1):61～66.
[45]．李晓峰，刘光中．人工神经网络BP算法的改进及其应用[J]．四川大学学报(工程科学版)．2000，3(2)：105～109．
[46]．焦李成．神经网络系统理论[M]．西安：西安电子科技大学出版社，1990年12月第1版．
[47]. Math Works. Inc. MATLAB documentation[H]. 2001.
[48]. N.W.Charence, E.G.Witting. A Study of the Parameters of a Back propagation Stock Price Prediction Model. The first New Zealand International Two-stream Conf on Artificial Neural Networks and Expert System. 1993:288～291.
[49]．韩力群．人工神经网络理论、设计与应用[M]．北京：化学化工出版社，2002年1月第1版．
[50]. J.Jelonek, K.Krawiec, R.Slowinski. Rough Set Reduction of Attributes and Their Domains for Neural Networks[J]. Computational Intelligence. 1995,11(2):339～347.
[51]．陈遵德．Rough Set神经网络智能系统及其应用[J]．模式识别与人工智能．1999，
12(1): 1～5.
[52]. R.Hashemi, L.A.Blanc, C.T.Rucks, et al. A Hybrid Intelligent System for Predicting Bank Holding Structures [J]. European Journal of Operational Research. 1998,109(2):390～402.
[53]. B.S.Ahn, S.S.Cho, C.Y.Kim, The Integrated Methodology of Rough Set Theory and Artificial Neural Network for Business Failure Prediction[J]. Expert System with Applications. 2000,17(18):65～74.
[54]．朱林，何建敏，常松．粗集与神经网络相结合的股票价格预测模型[J]．中国管理科学．2002，10(4)：7～12．
[55]．刘国良，强义文，麻亮，陈兴林．基于粗神经网络的仿人智能机器人的语音融合算法研究[J]．控制与决策．2003，18(3)：364～366．
[56]．陈可，张琴舜，陈培培等．BP神经网络在证券分析预测中应用[J]．计算机工程．2001，27(11)：95～97．
[57]．侯木舟，韩旭里．基于MATLAB的神经网络在股市预测中的应用[J]．系统工程．2003，21(2)：112～115．
[58]．汪卫霞．对虚假会计信息批露的制度分析[J]．经济问题探索．2003，10：100～104．
[59]．王玉英．会计造假的原因及其治理对策[J]．企业经济，2003，10：184～185