基于影响力计算模型的股市系统趋势预测研究

发布时间：2018-01-20 12:28

本文关键词： 股票网络影响力计算模型 BCNHC算法社团演化趋势预测　出处：《合肥工业大学》2014年硕士论文　论文类型：学位论文

【摘要】：复杂网络是复杂系统的结构化研究工具,是以全局的角度研究整个系统的内部能量的相互作用关系。在股市系统中,利用历史重演性假设,建立可以描述股市系统要素之间关联性的复杂网络,再从系统的角度出发,分析网络的结构、功能特性,是股市系统研究的主要方向之一。论文主要研究两方面内容：一是,从应用出发,研究社团内部影响力关系；二是,分析社团间的演化特性,研究社团与社团间的影响力关系,进而实现对股市系统的趋势预测。具体如下：第一,介绍以相关系数为权值的股票网络的研究现状,分析社团划分算法存在的不足,进而引入偏相关系数的概念,再从应用的角度加入活跃性条件定义影响力计算模型,以此模型得到的影响力为权重值,建立股票网络,并提出BCNHC算法。应用个股数据与板块数据对BCNHC算法进行实验,表明影响力计算模型和算法设计的合理性和有效性。第二,根据社团演化理论研究社团及社团间的影响力关系,进而利用社团间影响力计算模型预测股市趋势。思想是：首先,利用技术指标建立指标间的贝叶斯网络获得当前分时线下的股市趋势,然后,在贝叶斯网络中加入短期分时线的股价趋势建立社团影响力计算模型,用影响力衡量当前分时线的股市趋势,最后,建立各分时线子系统的传动模型预测股市趋势。通过实际数据,对论文提出的算法与模型进行评估。实验表明,只要能够建立比较合理的影响力计算模型,就能较准确的预测走势。
[Abstract]:Complex network is a structured research tool for complex systems, which studies the interaction of internal energy of the whole system from a global perspective. In the stock market system, historical replay hypothesis is used. To establish a complex network which can describe the correlation between the elements of the stock market system, and then from the point of view of the system, analyze the structure and functional characteristics of the network. It is one of the main research directions of the stock market system. The thesis mainly studies two aspects: first, from the application, studies the influence relation within the community; The second is to analyze the evolution characteristics of the community, study the influence relationship between the community and the community, and then realize the trend prediction of the stock market system. The details are as follows: first. This paper introduces the research status of stock network with correlation coefficient as weight, analyzes the shortcomings of community partition algorithm, and then introduces the concept of partial correlation coefficient. Then from the perspective of application, we add the active conditions to define the influence calculation model, and then establish the stock network based on the weight of the influence obtained by the model. BCNHC algorithm is proposed. The experiment of BCNHC algorithm with individual stock data and block data shows the rationality and effectiveness of the influence calculation model and algorithm design. Second. According to the theory of community evolution, this paper studies the influence relationship between the community and the community, and then predicts the stock market trend by using the inter-community influence calculation model. The idea is as follows: first of all. Using the technical indicators to establish the Bayesian network between the indicators to obtain the current stock market trend under the time-sharing line, then, the stock price trend of the short-term time-sharing line is added to the Bayesian network to establish a model for calculating the influence of the community. Finally, the transmission model of each time-sharing subsystem is established to predict the stock market trend. Through the actual data, the algorithm and model proposed in this paper are evaluated. As long as we can establish a more reasonable influence calculation model, we can predict the trend more accurately.
【学位授予单位】：合肥工业大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：F830.91;O157.5

【相似文献】