發(fā)布時(shí)間：2018-01-15 23:25

  本文關(guān)鍵詞：信息及其擴(kuò)散對(duì)證券市場(chǎng)的影響　出處：《天津大學(xué)》2014年博士論文　論文類型：學(xué)位論文

  更多相關(guān)文章： 證券市場(chǎng)信息 Ward權(quán)熵指標(biāo) 連續(xù)滲流 厚尾現(xiàn)象 Lévy過程

【摘要】：信息在證券市場(chǎng)中具有舉足輕重的作用。信息的內(nèi)容及價(jià)值影響市場(chǎng)中投資者的決策;信息的擴(kuò)散導(dǎo)致市場(chǎng)波動(dòng)等,都是證券市場(chǎng)的研究熱點(diǎn)。本文基于證券市場(chǎng)中靜態(tài)和動(dòng)態(tài)兩類信息,使用不同的信息處理技術(shù),分別從兩個(gè)角度討論了信息對(duì)投資者的效用和信息擴(kuò)散對(duì)市場(chǎng)波動(dòng)產(chǎn)生的影響。首先,針對(duì)靜態(tài)信息,主要是上市公司的財(cái)務(wù)數(shù)據(jù),從財(cái)務(wù)聚類的角度進(jìn)行研究。由于對(duì)股票聚類方法的研究較多,但對(duì)聚類結(jié)果的優(yōu)劣、取舍卻較難評(píng)定,本文提出了一個(gè)評(píng)價(jià)聚類結(jié)果優(yōu)劣的Ward權(quán)熵指標(biāo),以幫助投資者對(duì)股票進(jìn)行聚類和篩選。該指標(biāo)兼具股票聚類所要求的準(zhǔn)確性和面向投資者的實(shí)用性特點(diǎn),從基于距離度量的偏差損失最小和基于信息熵度量的信息量損失最小兩個(gè)角度衡量聚類結(jié)果的優(yōu)劣。Ward權(quán)熵指標(biāo)適用于不同的聚類方法、相似性度量、以及指標(biāo)加權(quán)等狀態(tài),具有廣泛的適應(yīng)性。文中驗(yàn)證了在聚合聚類下,指標(biāo)關(guān)于聚類數(shù)K單調(diào)不降。通過實(shí)證,分析了指標(biāo)的特性,并使用該指標(biāo)對(duì)不同聚類方法和聚類結(jié)果進(jìn)行了較為有效的評(píng)價(jià)。其次,針對(duì)動(dòng)態(tài)信息,本文討論了證券市場(chǎng)信息擴(kuò)散的羊群效應(yīng)及其對(duì)市場(chǎng)波動(dòng)的影響。借鑒與信息擴(kuò)散結(jié)構(gòu)相似的連續(xù)滲流理論,本文詳細(xì)討論了利用連續(xù)滲流模擬價(jià)格或指標(biāo)波動(dòng)的模型構(gòu)造、理論分析和實(shí)證。主要介紹了滲流及連續(xù)滲流的理論和應(yīng)用背景。給出一個(gè)基本的RCM波動(dòng)模型,進(jìn)而重點(diǎn)討論單串和多串的連續(xù)滲流模型及其改進(jìn),使所構(gòu)造的模型不斷貼近市場(chǎng)的真實(shí)狀態(tài)。在不同的模型下均驗(yàn)證出波動(dòng)率收斂于L′evy過程,而非有效市場(chǎng)假說(shuō)條件下得到的Wiener過程。通過程序?qū)崿F(xiàn),形象地展示了在連續(xù)滲流模型下,信息擴(kuò)散的機(jī)制及上、下臨界狀態(tài)的差異,并通過模擬波動(dòng)過程,驗(yàn)證了收益率的厚尾現(xiàn)象。最后根據(jù)不同參數(shù)的變化導(dǎo)致收益率變化的狀態(tài),解釋了參數(shù)的作用。
[Abstract]:Information plays an important role in the securities market. The diffusion of information leads to the fluctuation of the market, which is the research hotspot of the securities market. Based on the static and dynamic information in the securities market, different information processing techniques are used in this paper. This paper discusses the effect of information on investors and the influence of information diffusion on market volatility from two angles. Firstly, aiming at static information, it mainly focuses on the financial data of listed companies. From the perspective of financial clustering, because of the more research on the stock clustering method, but the advantages and disadvantages of the clustering results, it is difficult to evaluate. In this paper, a Ward weight entropy index is proposed to evaluate the clustering results. In order to help investors to cluster and screen the stock. This index has both the accuracy required by stock clustering and the practicability of investor oriented. From the two aspects of minimum deviation loss based on distance measurement and minimum information loss based on information entropy measure, Ward weight entropy index is suitable for different clustering methods, similarity measurement. In this paper, it is verified that the cluster number K of the index is not monotone decreasing under the aggregation clustering. Through the empirical analysis, the characteristics of the index are analyzed. This index is used to evaluate the different clustering methods and clustering results. Secondly, the dynamic information is analyzed. In this paper, the herding effect of information diffusion in securities market and its influence on market fluctuation are discussed, and the continuous seepage theory similar to information diffusion structure is used for reference. This paper discusses in detail the construction of a model for simulating price or index fluctuation by continuous seepage. This paper mainly introduces the theory and application background of seepage flow and continuous seepage, gives a basic RCM wave model, and then discusses the continuous seepage model with single string and multiple strings and its improvement. The constructed model is kept close to the real state of the market and the volatility converges to the Levy process under different models. The Wiener process obtained under the condition of the non-efficient market hypothesis is realized by the program, which vividly shows the mechanism of information diffusion and the difference of the upper and lower critical states under the continuous seepage model. By simulating the fluctuation process, the thick tail phenomenon of the return rate is verified. Finally, the effect of the parameter is explained according to the state of the change of the return rate caused by the change of the different parameters.
【學(xué)位授予單位】：天津大學(xué)
【學(xué)位級(jí)別】：博士
【學(xué)位授予年份】：2014
【分類號(hào)】：F832.51

【參考文獻(xiàn)】

中國(guó)期刊全文數(shù)據(jù)庫(kù) 前10條

1 DONG Yanfang;WANG Jun;;COMPLEX SYSTEM ANALYSIS OF MARKET RETURN PERCOLATION MODEL ON SIERPINSKI CARPET LATTICE FRACTAL[J];Journal of Systems Science & Complexity;2014年04期

2 葛顥;;非平衡態(tài)統(tǒng)計(jì)物理的隨機(jī)數(shù)學(xué)理論[J];數(shù)學(xué)進(jìn)展;2014年02期

3 張敞;王園園;趙裕嘯;伍章俊;;一種基于信息熵的聚類結(jié)果評(píng)價(jià)方法[J];合肥工業(yè)大學(xué)學(xué)報(bào)(自然科學(xué)版);2011年08期

4 周漩;張鳳鳴;惠曉濱;李克武;;基于信息熵的專家聚類賦權(quán)方法[J];控制與決策;2011年01期

5 余樂安;汪壽陽(yáng);;基于核主元聚類的股票分類[J];系統(tǒng)工程理論與實(shí)踐;2009年12期

6 李云飛;李鵬雁;;基于模糊聚類技術(shù)的股票投資價(jià)值評(píng)價(jià)指標(biāo)選擇[J];燕山大學(xué)學(xué)報(bào);2008年06期

7 孫吉貴;劉杰;趙連宇;;聚類算法研究[J];軟件學(xué)報(bào);2008年01期

8 李敏;何理;;聚類分析在證券投資基本分析中的應(yīng)用[J];遼寧師范大學(xué)學(xué)報(bào)(自然科學(xué)版);2006年02期

9 朱惠倩;聚類分析的一種改進(jìn)方法[J];湖南文理學(xué)院學(xué)報(bào)(自然科學(xué)版);2005年03期

10 王寧,王軍;利用Poisson Boolean模型建立股票價(jià)格波動(dòng)過程及數(shù)據(jù)模擬[J];北京交通大學(xué)學(xué)報(bào);2005年03期

中國(guó)博士學(xué)位論文全文數(shù)據(jù)庫(kù) 前2條

1 呂宗磊;對(duì)聚類及聚類評(píng)價(jià)若干問題的研究[D];南京航空航天大學(xué);2009年

2 章融;證券市場(chǎng)中的羊群行為研究[D];浙江大學(xué);2004年

中國(guó)碩士學(xué)位論文全文數(shù)據(jù)庫(kù) 前2條

1 王園園;基于決策樹的模糊聚類評(píng)價(jià)算法及其在證券領(lǐng)域的應(yīng)用[D];合肥工業(yè)大學(xué);2010年

2 李瑩;聚類結(jié)果評(píng)價(jià)方法與聚類知識(shí)提取技術(shù)的研究[D];南京航空航天大學(xué);2008年

，

本文編號(hào)：1430557