基于投资者交互作用的股市异常波动形成机理研究

The formation mechanism of abnormal fluctuations in the stock market has been one of the most challenging issues of widespread concern by domestic and foreign experts in the financial field. The current research suggests that stock market is a complicated network made up of a number of investors. The interaction among investors will spur collective composition effect which is different from individual behavior, and will influence the overall stability of the market. However, the current issues in research are the complicated model parameters, a lack of time and space dimensions consideration, etc. This project firstly introduces q-states Potts model in the Condensed Matter Physics to the studies on the individual behavioral choices in the stock market and replicates the interaction mechanism among the investors. And then based on macro-investor interaction, stock price index time series is developed and applied to the weighted fractal network in the stock market, which connects the individual macro behavior in the capital market to micro emergency phenomenon so as to reveal the forming mechanism of the abnormal fluctuation in the stock market from the perspective of investors’ interaction. Finally, the project takes the reasonable and effective control of irrational behavior in the stock market as the goal, analyzes the effect of media intercourse and the investors themselves on abnormal volatility of stock market, and find out the key factors affecting the stock abnormal fluctuations. The expected results will benefit security market supervisors to gain a deeper understanding of the forming mechanism of the abnormal fluctuation in the market, early warn abnormal fluctuation, and provide theory and policy support to take effective measures to evade market risk.

股票市场的异常波动形成机理一直是金融领域最具挑战性的课题之一，受到国内外专家的广泛关注。目前较多研究认为股市是一个由众多投资者组成的复杂网络，投资者间的交互作用能够激发出异于个体行为的群体性结构效应，并会影响市场整体的稳定。但存在模型参数过于复杂、缺乏时空维度的结合等问题。本项目将凝聚态物理中的3态Potts模型引入到股票市场个体行为选择的研究中，复刻投资者之间的交互机理；并建立一个基于微观投资者交互作用的股指时间序列，应用于股票市场的加权分形网络中，将资本市场个体微观行为与宏观涌现现象有机地联系在一起，从投资者交互作用的视角揭示股市异常波动的形成机理；并以股票市场的非理性行为的合理与有效控制为目标，分析媒体外场和投资者自身对股市异常波动的影响，寻找影响股市异常波动的关键因素。预期研究成果将为监管层深入了解市场异常波动机理，预警股市异常波动，并采取有效措施防范市场风险提供理论和政策支持。

股票市场的异常波动形成机理一直是金融领域最具挑战性的课题之一，受到国内外专家的广泛关注。本项目将资本市场个体微观行为与宏观涌现现象有机地联系在一起。从股票市场中微观投资者的交互作用的视角，致力于从理论方面揭示股市异常波动的形成机理，从实证方面预警股票市场的异常波动。本项目为采取有效措施防范市场风险提供一定的理论支撑与参考价值。主要取得的成果如下：.（1）基于异构主体的两态（买和卖）社会物理学模型，构建了多数人投票模型（MVM2），研究欧式空间规则网络中投资者（反向交易者和噪声交易者）的交互作用。通过蒙特卡洛仿真显示，提出的MVM2模型展示了诸如波动率聚集，收益率幂律分布以及绝对收益与指数衰减的长程相关性等真实金融市场的特征事实。.（2）扩展MVM2模型，提出三态MVM模型（MVM3）。反向交易者和噪声交易者有买、卖和持有三种状态，研究欧式空间规则网络中投资者的交互作用。发现了真实金融市场中的特征事实，并且定量地揭示了描述系统复杂性的非线性统计耦合方面的尖峰、中峰和低峰状态的转换。.（3）基于Barabási-Albert网络，通过提出的MVM3模型，研究股票市场中投资者在无标度网络中的交互作用。前人的研究显示，许多不同复杂网络的有效维度等于1。我们认为，此结果是标度（scaling）应用的结果，并且这不是那些网络的真实维度。.（4）研究了媒体外场和投资者自身因素对股市异常波动的影响。基于网络关注度和媒体关注度，研究股票市场中的信息不对称。基于热最优路径模型，从微观视角考察股票的网络关注度变化与收益率、波动率、流动性以及信息不对称程度等市场微观结构指标之间的动态领先滞后关系，预测股票市场价格变化。