基于RV-MSM模型的沪深300波动分析

发布时间：2018-03-15 10:45

本文选题：多重分形　切入点：沪深300　出处：《华中科技大学》2013年硕士论文　论文类型：学位论文

【摘要】：金融资产收益波动率一直是金融领域的一个研究热点。风险评估和定价作为金融市场的核心都需要对资产收益波动率进行有效度量，同时对波动性的辨识效果直接影响到风险管理、资产定价、资产配置等领域。。近期来，随着计算机和通讯技术的发展，基于日内高频数据的波动率模型开始兴起，记为已实现波动率模型。相对于低频数据，高频数据由于采集时间间隔小，具有更多市场微观结构信息，高频数据模型能够大幅度降低测量误差和噪声对其真实波动率的影响。分形市场理论突破了有效市场理论的正态、独立、线性等假定，有效的解释了许多有效市场理论下的异常现象‖。目前，，在国内的研究中，还缺乏利用将分形理论和高频数据相结合来探讨中国股市的波动特征的模型。因此，本文试图利用沪深300高频数据，基于MSM模型研究中国股指的波动性，无疑具有理论和现实意义。本文通过R/S分析等多重分形方法证实了沪深300数据存在多重分形特征；在此基础之上建立了马尔可夫转换多重分形模型（MSM），通过广义矩估计方法估计其参数；将其预测效果与GARCH模型进行比较分析，从中发现其预测效果要好于GARCH模型。在低频马尔可夫机制转换模型的基础上，考虑到高频数据的优越性，建立已实现波动率的RV-MSM模型。考虑到高频数据受到市场微观结构误差等的影响，本文对高频波动率进行相应改进，提出了ARV-MSM模型。从预测结果中可以发现，MSM模型预测效果优于GARCH模型；同时高频MSM模型优于低频MSM模型；改进过的已实现波动率模型，优于未改进波动率模型。
[Abstract]:As the core of financial market, risk assessment and pricing need to measure the volatility of asset return effectively. At the same time, the identification effect of volatility has a direct impact on risk management, asset pricing, asset allocation and other fields. In recent years, with the development of computer and communication technology, volatility models based on intraday high frequency data have been developed, which are recorded as realized volatility models. With more market microstructure information, high frequency data model can greatly reduce the influence of measurement error and noise on its true volatility. Fractal market theory breaks through the assumptions of efficient market theory, such as normal, independent, linear, etc. At present, in the domestic research, there is a lack of the use of fractal theory and high-frequency data to explore the volatility characteristics of the Chinese stock market model. This paper attempts to use the high frequency data of Shanghai and Shenzhen 300 to study the volatility of Chinese stock index based on MSM model, which is undoubtedly of theoretical and practical significance. In this paper, the existence of multifractal features in CSI 300 data is confirmed by using multifractal method such as R / S analysis, and then the Markov transform multifractal model is established, and its parameters are estimated by generalized moment estimation method. By comparing the prediction effect with the GARCH model, it is found that the prediction effect is better than that of the GARCH model. On the basis of the transformation model of low frequency Markov mechanism, the superiority of high frequency data is considered. The RV-MSM model of realized volatility is established. Considering that the high frequency data is influenced by the market microstructure error, this paper improves the high frequency volatility. The prediction results show that the ARV-MSM model is superior to the GARCH model, the high frequency MSM model is superior to the low frequency MSM model, and the improved realized volatility model is superior to the unimproved volatility model.
【学位授予单位】：华中科技大学
【学位级别】：硕士
【学位授予年份】：2013
【分类号】：F224;F832.51

【相似文献】