基于网络舆论与文本分析的公募基金主观评级研究

发布时间：2018-01-05 19:08

本文关键词：基于网络舆论与文本分析的公募基金主观评级研究　出处：《西南财经大学》2012年硕士论文　论文类型：学位论文

【摘要】：基金业与基金评级业的发展是相辅相成的。无论是对于金融市场发达的欧美还是金融市场刚刚起步的发展中国家来说,基金评级对于基金业都是非常重要的。一方面,基金业的发展,需要基金评级为其保驾护航；另外一方面,基金评级的发展使得评级结果更加真实可靠。进一步来说,投资者对于基金评级的精准度的要求也越来越高,也是促进基金评级业循序前进的主要动力。我国的基金业发展至今已经有30年的历史了,2005年之后,我国资本是市场发展迅猛,截止2011年底,国内一共有66家基金管理公司,发行并管理了914公募基金,基金资产份额总规模为26464.65亿份,管理的基金资产净值总规模为21879.79亿元,但与国外成熟的市场比较来说仍处于起步阶段,运作过程中还存在着许多不规范的地方。随着我国基金业的持续发展,基金经理的能力高低不等,基金产品良莠不齐的状况愈发明显,基金的业绩有较大的差距,但是投资者却无法从披露信息、去判断基金的好坏。随着我国资本市场的发展和投资理念的普及,基金投资者的数量出现了飞速增长的态势。在选择基金进行投资的时候,大多数散户投资者依据基金的评价指标来选择基金进行投资,如根据各个中介机构对基金的评级排名,该基金的历史收益率等来选择合适投资对象。而这些选择的方法都是属于客观评级的范畴。但是,当前的客观评级研究的状况并不能很好的保护投资者,相当多的投资者在投资证券投资基金的时候依然无所适从,个体风险很大。因此,从保护投资者利益和防范基金投资风险的角度出发,我们急需研究出一个和客观评级相对应的主观评级方法来指导投资者投资。并且基于一个更好的评级方法可以更好的为投资者的理性投资提供参考,更好的激励基金经理提高业绩、约束基金经理不端行为,更好的促进整个是基金市场稳定有序发展,更好的提升资本市场的运作效率这四个理由。在本文中,笔者尝试利用文本分析的现代方法,从文本数据和主观评价数据最活跃的论坛数据入手,利用主观评价的方法对基金进行主观业绩评价,从而得到一个主观评级的方法。本文主要采用理论分析的方法,辅之以实证分析,考察互联网视角下公募基金的实际年收益与主观评价之间相互影响的关系,从理论和实证上对公募基金主观评价体系进行了研究。本文将沿着文献综述、理论分析到假设、实证检验再到结论与政策建议的技术路线来搭建研究的框架。通过提出主观评级可以真实反映基金的业绩实际状况、公募基金主观评级对收益率的敏感度超过客观评级、公募基金的主观评级和客观评级相互独立的关系三个假设,并通过实证证实了这三个假设的真实性。本文共分为七个部分,具体如下：第一部分为绪论,作为本文研究的起点,主要介绍选题的相关背景、研究的目的及意义、思路与方法、本文的创新点。第二部分是国内外相关研究文献的综述与分析。介绍国内外对公募基金业绩评价的研究现状及成果、国内外对基金评级的研究与应用,最后对已有文献进行了总结并引出本文的研究。第三部分是基于文本挖掘的主观业绩评价理论分析。本部分主要介绍了行为金融学中关于主观业绩评价的理论,信息不充分条件下投资者所采取的羊群行为,并提出主观评价理论和公募基金主观业绩评价的三个假设。第四部分是主观评价模型中所用文本挖掘的技术介绍。介绍了数据挖掘和文本挖掘的理论基础、文本挖掘的方法和步骤,接着按照文本挖掘的理论和方法构建主观评价模型第五部分是主观评价模型的运行。本文选取天天基金网中的数据,截取了2011年30只公募基金的子论坛所发帖子的文本数据作为基础样本,选择合适的控制变量,构建自己的理论模型进行分析检验。第六部分是对上一部分所提取的数据进行多角度的分析,从统计角度来说,通过引进评错率,波动率和准确性对主客观评级进行分析,随后利用统计模型验证主观评级、客观评级和实际收益率之间的相关关系第七部分是本文研究的结论。根据理论分析、现实分析及实证分析得出结论,由此提出关于建立我国公募基金主观评价体系的政策性建议,并总结了本文研究的局限性和不足,提出了后续研究的方向。本文的创新之处体现在以下三个方面： (1)理论创新。首次在理论界提出以主观评价的方式来判断基金业绩,并建立起主观评价的模型。在主观评价理论的基础上,结合实际提出主观评级的三大假设。 (2)方法创新。把数据挖掘和文本挖掘相关方法实际运用在金融领域,把主观评价的指标纳入基金评价体系,在实证分析的基础上构建主观评价方法,提出一套完整的公募基金主观评级模型。 (3)应用创新。把成熟的客观评级模型应用在主观评价领域,把信息交流最为密切的论坛数据作为研究的对象,把海量的文本数据转化为主观评价的特征,最后应用于金融领域的研究。这和大多数基金评级研究的思路是不一致的。最后,由于时间和个人能力有限,本文的研究还存在着许多不足和有待改进的地方,具体表现在：第一、对于构建公募基金的主观评级系统来说,30只基金的803个帖子作为样本偏小,如果想要构建真实可靠的主观评级系统需要更多相关论坛的帖子为数据源,必须以互联网为研究基础。以海量的文本数据所得出的主观评价是真实可靠的,可以使公募基金的主观评级会更有实践意义。第二、本文最终所选取5个具有普遍性的主观指标是通过文本挖掘的方法得到的。如果对5大主观指标进行进一步的细分,加入一些二级指标,那基金的主观评级结果是否会发生改变呢?这个还需要后来人对该问题进行更深层次的研究。第三、本文仅证实了主观评级与基金实际收益率的相关性,验证了主观评价作为基金评级必不可少的一部分,而对于投资者依靠主观评级系统来进行投资的结果本文并未进行涉及,而这方面的研究还需要进行近一步的实证检验。
[Abstract]:The development of the fund industry and fund rating industry are complementary to each other. Whether it is for the developed financial market and financial market has just started developing countries, the fund rating is very important for the fund industry. On the one hand, the development of the fund industry, fund rating required for the escort; on the other hand, the development of the fund rating the rating results are more reliable. Furthermore, investors have become for the fund rating accuracy of the higher requirements, is the main driving force to promote the fund rating industry step by step.
The development of the fund industry in China has a history of 30 years, after 2005, the rapid development of Chinese capital market, by the end of 2011, China has a total of 66 fund management companies, issuance and management of 914 public fund, the total size of the fund assets accounted for 2 trillion and 646 billion 465 million, the total size of the net assets of the fund management for 2 trillion and 187 billion 979 million yuan, but compared with the foreign mature market is still in its infancy, there are still in the process of many non-standard place. With the continuous development of China's fund industry, fund manager ability ranging from the level of fund products, the status of uneven in quality is becoming increasingly apparent, the performance of the fund has a large gap, but investors but not from the disclosure of information to judge the fund directly. With the popularization and development of China's capital market and investment philosophy, the number of fund investors in the trend of rapid growth in the choice of base. When the gold investment evaluation index, most retail investors on the basis of the fund to fund investment, such as by various intermediaries to rank the fund rating, the fund's historical return to choose the appropriate investment. And these choices are the objective rating category. However, investor protection objective rating study on the current situation is not very good, many investors are still at a loss, the individual risk greatly. Therefore, the investment risk prevention fund and protect the interests of investors from the point of view, we need to research the subjective rating method with an objective rating corresponding to guide investors and a better rating methods can better based on rational investment for investors to provide a reference, better incentive fund managers to improve performance, Bound fund manager misconduct, better promote the stable and orderly development of fund market, the four reason to promote capital market better operational efficiency. In this paper, the author tries to use modern methods of textual analysis, starting from the text data and subjective data of the live data to the forum, of subjective performance evaluation of the fund the use of subjective evaluation, resulting in a subjective rating methods.
This paper mainly adopts the method of theoretical analysis, supplemented by empirical analysis, relationship between the study from the perspective of Internet public offering of the fund's actual income and subjective evaluation, conducted the research on the subjective evaluation system of public fund from the theoretical and empirical analysis. This article will follow the literature review, theoretical analysis to the hypothesis, empirical framework test the technical route to the conclusions and policy recommendations to build the research. Through the performance of the actual situation to give subjective ratings may reflect the fund, public fund subjective rating for yield sensitivity over objective rating, relationship between the public offering of the fund's subjective rating and objective rating of three independent hypotheses, and the authenticity of this three hypotheses were verified through empirical research.
This article is divided into seven parts, which are as follows:
The first part is the introduction, which is the starting point of this study. It mainly introduces the background of the topic, the purpose and significance of the study, the ideas and methods, and the innovation of this article.
The second part is the summary and analysis of related research literature at home and abroad. It introduces the research status and achievements of public fund performance evaluation at home and abroad, and the research and application of fund rating at home and abroad. Finally, it summarizes the existing literature and leads to the research in this paper.
The third part is the analysis of subjective performance evaluation based on the theory of text mining. This part mainly introduces Behavioral Finance on subjective performance evaluation theory, the herd behavior of investors to take under the condition of insufficient information, and put forward three hypotheses of subjective evaluation theory and subjective performance evaluation of public fund.
The fourth part is the introduction of text mining technology in the subjective evaluation model. It introduces the theoretical basis of data mining and text mining, the methods and steps of text mining, and then constructs the subjective evaluation model according to the theory and method of text mining.
The fifth part is the operation of the subjective evaluation model. This paper selects the data from the daily fund net, intercepts the textual data of the 30 public offering sub forums in 2011 as the basic sample, selects the appropriate control variables, and builds their own theoretical models for analysis and verification.
The sixth part is the analysis of the extracted data on the part of the multi angle, from a statistical point of view, through the introduction of the assessment error rate, fluctuation rate and accuracy of objective and subjective ratings were analyzed, then verify the subjective rating using statistical model, objective rating and the actual rate of return correlation between
The seventh part is the conclusion of this study. According to the theoretical analysis, practical analysis and empirical analysis conclusion, puts forward policy suggestions on the establishment of subjective evaluation system of raised funds in China, and summarizes the limitations and shortcomings of this study, put forward the direction for further research.
The innovation of this article is embodied in the following three aspects:
(1) theoretical innovation. For the first time, in the theoretical circle, we propose to evaluate the fund performance by subjective evaluation and set up a subjective evaluation model. Based on the subjective evaluation theory, we put forward three hypotheses of subjective rating based on the reality.
(2) method innovation. Data mining and text mining methods are applied in the financial field, and the subjective evaluation indicators are included in the fund evaluation system. Based on the empirical analysis, a subjective evaluation method is constructed, and a complete set of subjective rating models for public offering funds is put forward.
(3) the application of innovation. The mature objective rating model used in the field of subjective evaluation, the most significant forum data as the object of study for the exchange of information, the massive text data into the subjective evaluation of the characteristics, and finally used in the financial sector. The fund rating and most of research is not consistent.
Finally, because of the limited time and personal ability, there are many deficiencies and areas to be improved in this paper.
First, the subjective rating system construction of public fund, 803 post 30 fund as the sample is small, if you want to build a subjective rating system more reliable forum posts as the data source, must be on the Internet. On the basis of the subjective assessment the massive text data is true and reliable that can make the public offering of the fund's subjective ratings will be more practical significance.
Second, this paper finally selected 5 subjective index universal is obtained by text mining. If further divided the 5 subjective indicators, adding some two indicators, the fund's subjective rating results will change? This need later for a deeper study of the the problem.
Third, this article only confirmed the correlation between subjective rating and the actual yield of the fund, as part of the validation of subjective evaluation is essential for the fund rating, investors rely on the subjective rating system to invest the results in this paper are not involved, and this research also need to empirically test the next step.

【学位授予单位】：西南财经大学
【学位级别】：硕士
【学位授予年份】：2012
【分类号】：F832.51;F224

【参考文献】