基于遗传算法的沪深300指数HAR模型结构优化

发布时间：2018-04-05 09:43

本文选题：已实现波动　切入点：HAR模型　出处：《南京大学》2013年硕士论文

【摘要】：波动率研究一直是金融研究的重要课题之一。波动率是金融资产价格变化的主要特征,选择适合的估计量以及对波动率进行建模分析并对其未来变化进行预测,能够为金融资产的定价、投资组合确定以及风险管理提供有力支持。随着信息储存成本的降低,金融研究逐渐从低频领域转向高频领域,高频数据中包含了更加丰富的信息,使得波动率的预测变的更加复杂。已实现波动是在高频数据发展基础上提出的对波动率进行描述的方法,异质自回归模型(HAR模型)是对已实现波动估计的模型,以计算简便,估计准确而在波动率的研究领域逐渐发展起来。本文对已实现波动率HAR以及模型的发展做了详细介绍,并对HAR模型在中国市场的适用性作出了研究,利用HAR模型对能够代表中国市场特性的沪深300指数进行回归分析,讨论其在(1,5,22)的时间尺度结构下,对沪深300指数的预测能力,而后利用遗传算法对模型的时间尺度结构进行优化,得到了更能够反映中国市场投资者行为特点的新的时间尺度结构,为HAR模型在中国市场的发展以及沪深300指数的波动率预测提供了参考和建议。
[Abstract]:Volatility research has been one of the most important issues in financial research.Volatility is the main characteristic of the price change of financial assets. Choosing the appropriate estimator, modeling and analyzing the volatility and predicting its future changes can price the financial assets.Portfolio determination and risk management support.With the decrease of the cost of information storage, the financial research has gradually shifted from the low-frequency domain to the high-frequency field. The high-frequency data contain more information, which makes the volatility prediction more complicated.The realized fluctuation is a method to describe the volatility based on the development of high frequency data. The heterogeneous autoregressive model (Har) is the model of the realized wave estimation, which is easy to calculate.Accurate estimates are gradually developing in the field of volatility research.This paper gives a detailed introduction to the realized volatility HAR and the development of the model, studies the applicability of the HAR model in the Chinese market, and makes a regression analysis of the CSI 300 index, which can represent the characteristics of the Chinese market, by using the HAR model.This paper discusses the prediction ability of CSI 300 index under the time scale structure of the model, and then optimizes the time scale structure of the model by using genetic algorithm.A new time scale structure which can better reflect the behavior of investors in Chinese market is obtained, which provides a reference and suggestion for the development of HAR model in Chinese market and the prediction of volatility of CSI 300 index.
【学位授予单位】：南京大学
【学位级别】：硕士
【学位授予年份】：2013
【分类号】：F832.51;F224;TP18

【参考文献】