基于双GARCH的股票风险预测.docVIP

海 南 大 学 毕 业 论 文（设计） 题 目：基于双GARCH的股票风险预测 学 号：20060444175 姓 名：李文辉 年 级：2006级 学 院：经济与管理学院 系 别：金融学 专 业：金融学 指导教师：徐 艳 完成日期：2010年5月1日 摘要 1982年，Engle教授提出ARCH模型，给计量经济学带来了新的建模方法。自那以后，一系列以ARCH为基础的模型相继被建立。资本市场的波动被看成是资产面临的风险。投资者最为关心资产在未来面临的风险大小，从而几乎每个投资者都想尽办法去预测资产的价格走势和资产未来的风险。本文旨在建立一个能够帮助投资者预测股票波动变化的模型。本文中的模型以被广泛应用于金融领域的ARCH-M模型为基础。独特之处在于引入了市场波动作为资产波动方程的一个解释变量，从而将单项资产波动变化过程与市场波动变化过程联系起来。在模型中，舍弃了正态分布假设，以更一般化的“广义误差分布”取而代之。另一个创新之处是，本文提出了一种全新的密集计算法用以估计方程参数，将复杂的优化过程转换成高密度的计算机运算。本文最后还选择了一个实例用以对新模型的可行性和预测能力进行验证。 关键词：ARCH模型 广义误差分布 密集计算法 MonteCarlo模拟 Abstract Since professor Engle put forward the ARCH model in 1982, which refreshed the modeling method in econometric, a series of models based on the ARCH model has been established. The volatility in capital market has always been considered to be the risk that the asset might take, which is what the investors concern most. Therefore, hardly any investors do not make effort to predicate the price trend of their asset and the future risk their asset might take. This thesis aims to build such a model so as to assist the investors to predicate the volatility of their assets. Based on the widely used ARCH model in financial area, the new model is characterized by having introduced the market variable as an explanatory variable in the asset volatility formula, thus connecting the changing process of single asset volatility with that of the market volatility. In addition, the new model has substituted the hypothesis of the normal distribution for the more generally used Generalized Error Distribution. Another creation in this thesis lies in that a brand new computationally intensive methods has been brought up to evaluate the formula parameters, transforming the complicated optimization into intensive computerization. At the end of the thesis, an e