基于跳跃滤波和时变参数估计的中国股市微观结构研究

发布时间：2018-01-06 09:40

本文关键词：基于跳跃滤波和时变参数估计的中国股市微观结构研究　出处：《系统工程理论与实践》2017年06期 　论文类型：期刊论文

【摘要】：为了更为有效地探究微观市场结构对股票价格的影响,本文在状态空间模型框架下,同时将交易方向、带方向的交易量、交易间隔、微观噪声以及跳跃因素引入状态方程和观测方程中,建立全面的市场微观结构模型,以反映各变量对交易价格序列产生的暂时性和永久性影响.然后,在采用基于粒子滤波的非参数离群点检测方法检测跳跃并对跳跃进行滤波的基础上,运用贝叶斯最小二乘方法对微观结构模型系数、波动率以及状态变量进行实时更新和估计,在此基础上分析跳跃日度分布、日内分布、个股跳跃和共同跳跃特征,并同时判断不同交易方向、交易量的订单对价格产生的影响.最后选取2015年股灾期间、沪市股票的逐笔数据作为研究样本进行实证研究.实证结果显示:我国股票高频时间序列跳跃和共同跳跃均存在日内季节性,且跳跃频率和大小与市场信息密切相关;买卖订单对交易价格及股票价值的影响并不是完全对称的,交易价格对于微观结构的敏感度、微观结构噪声也随市场波动而发生变化,熊市中都相对更高;考虑跳跃滤波并对参数进行实时估计可以提高价格对于交易的敏感度,能更好地捕捉市场微观结构在价格发现中的作用.
[Abstract]:In order to explore the effect of market structure on stock price more effectively, this paper, under the framework of the state space model, will change the trading direction, transaction volume with direction, and trading interval at the same time. The micro noise and jump factor are introduced into the state equation and observation equation, and a comprehensive market microstructure model is established to reflect the temporary and permanent influence of each variable on the transaction price series. Based on the nonparametric outlier detection method based on particle filter to detect jump and filter jump, Bayesian least square method is used to analyze the coefficients of microstructure model. Volatility and state variables are updated and estimated in real time. On this basis, the diurnal distribution, intraday distribution, individual stock jump and common jump characteristics are analyzed, and different trading directions are judged at the same time. The effect of trading order on price. Finally, the period of stock crash on 2015 is selected. The empirical results show that the high-frequency time series jump and the common jump of Chinese stocks exist intraday seasonality. The jumping frequency and size are closely related to the market information. The influence of buying and selling order on the trading price and stock value is not completely symmetrical. The sensitivity of the transaction price to the microstructure and the microstructure noise change with the fluctuation of the market, and the bear market is relatively higher. Considering jump filtering and real time estimation of parameters, the sensitivity of price to transaction can be improved, and the role of market microstructure in price discovery can be better captured.
【作者单位】：中央财经大学管理科学与工程学院;新加坡国立大学环境与设计学院;
【基金】：国家自然科学基金(71271223,70971145) 教育部新世纪优秀人才支持计划(NCET-13-1054)~~
【分类号】：F832.51
【正文快照】： i引言随着高频数据的获取方式和处理手段日臻完善,以研究金融市场交易价格发现机制,订单形成规则,交易结算机制,信息披露机制等为主要内容的市场微观结构理论逐渐发展成为金融学领域重要分支,该理论为研究金融资产价格提供新的思路和方法·Glosten等W通过探究证券市场价格对均

【相似文献】