摘要
选用EEMD+小波软阈值降噪法对金融高频数据进行降噪处理,然后利用仿真数据和实证方法验证了该方法的有效性,最后把降噪后的数据进行跳跃和跳跃溢出检验.研究表明:1)EEMD+小波降噪方法可以较好的降低金融高频数据的噪音.2)大部分情况下CSI300股指期现货市场对信息的反应不充分,期指更容易出现信息反映过度,两市场间有共同跳跃,跳跃溢出显著.3)通常认为噪音的存在使得期指对新信息的冲击更敏感,跳跃频率、幅度更大,会领先于现货市场发生跳跃,降噪后两市场的跳跃特征无明显变化,所以噪音不是股指期现货市场具有这些跳跃特征的唯一因素.
This paper chooses the EEMD＋ wavelet soft threshold de-noise method in the finance high frequency data, and then simulation data and empirical method verify the effectiveness of the method. Finally, we use the L-M jump test and analyze the jump feature. Research shows： 1） EEMD＋ wavelet de-noising method can reduce the noise of financial high frequency data. 2） CSI300 stock index flltures-spot market is not sufficient to react information in most cases, index futures will be more likely to reflect the excessive information, between the two markets have a common jump, jumping overflow significantly. 3） It is generally believed that noise makes the stock index futures more sensitive to the impact of new information, jump frequency and amplitude are bigger, can cause leap ahead of the spot market, jump features of two markets have no obvious change after noise reduction, therefore noise is not the only factor to influence the jump characteristics of two markets.
出处
《系统科学与数学》
CSCD
北大核心
2017年第6期1509-1523,共15页
Journal of Systems Science and Mathematical Sciences
基金
国家自然科学基金面上项目(71471182)
国家自然科学基金项目(71661028
71401188)
中央财经大学青年科研创新团队
中国博士后第61批面上项目
新疆财经大学校级(2015XYB018)
新疆自治区普通高校人文社科重点研究基地社会经济统计研究中心招标(050315C05)资助课题
