基于蚁群聚类算法的股票板块分类研究

发布时间：2018-02-06 00:01

本文关键词： 股票板块股票分类聚类分析蚁群算法　出处：《复旦大学》2012年硕士论文　论文类型：学位论文

【摘要】：随着中国股票市场不断发展,正确对股票进行分类,以构建投资组合降低投资风险的重要性也不断提高。根据现代投资组合理论,通过构建投资组合,可以起到分散非系统性风险的作用。投资组合的风险程度与组合内各股票之间的相关性有关,各股票之间的相关性越小,组合起到的风险分散效应越明显。现阶段投资者常按照行业对股票进行分类。因此,如果同一行业内的股票收益率之间的相关性高于不同行业的股票收益率间的相关性,不同行业间资产的搭配也应该能起到更好的效果。但是,本文通过实证研究证明了中国股市行业之间股票价格波动具有很高的相关性,按行业分类构建投资组合以降低风险的效果较差。因此,提出一种行业之外的有效分类股票的方法就显得非常重要。本文提出使用优化的蚁群聚类算法,对中国A股市场上所有的两千多支股票进行聚类分析。分别采用财务指标和个股收益率波动对股票进行聚类,通过对聚类结果的分析验证了使用蚁群聚类算法对大样本量数据进行聚类分析的可行性和良好效果。为在中国市场进行股票分类提供了新的思路,为投资决策和风险控制提供了理论和数据基础。本文共分为五章：第一章为研究背景、文献综述及论文框架介绍；第二章介绍了中国股票市场行业分类及其存在问题：第三章提出了基于优化的蚁群聚类算法的股票分类方法；第四章就第三章提出的方法进行了实证研究；最后提出了结论和展望。
[Abstract]:With the continuous development of China's stock market, the importance of correctly classifying stocks in order to build a portfolio to reduce investment risk is also increasing. According to modern portfolio theory, through the construction of portfolio. The degree of risk in the portfolio is related to the correlation between the stocks in the portfolio, and the smaller the correlation among the stocks. Portfolio plays a more obvious risk dispersion effect. At this stage, investors often classify stocks according to the industry. Therefore. If the correlation between stock returns in the same industry is higher than the correlation between stock returns in different industries, asset matching among different industries should also play a better role. However, through empirical research, this paper proves that stock price volatility among Chinese stock market industries has a high correlation, according to the industry classification of investment portfolio to reduce the effect of risk is poor. It is very important to propose an effective method of classifying stocks outside the industry. In this paper, we use the optimized ant colony clustering algorithm to cluster all the more than 2,000 stocks in the A-share market of China. We use the financial index and the volatility of individual stock return to cluster the stocks. The analysis of clustering results verifies the feasibility and good effect of using ant colony clustering algorithm to cluster large sample data, which provides a new idea for stock classification in Chinese market. It provides theoretical and data basis for investment decision and risk control. This paper is divided into five chapters: the first chapter is the research background, literature review and the introduction of the paper framework; The second chapter introduces the industry classification of Chinese stock market and its existing problems. In chapter 3, we propose a stock classification method based on ant colony clustering algorithm. Chapter 4th makes an empirical study on the methods proposed in the third chapter. Finally, the conclusion and prospect are put forward.
【学位授予单位】：复旦大学
【学位级别】：硕士
【学位授予年份】：2012
【分类号】：F832.51;F224

【参考文献】