The Long Memory Based On The Fund Market Of High-frequency Data

Stylized facts research of high frequency data have become the focus in economics and finance field in recent years. Because of the big capacity of data on a relatively short time span, it is more convenient to study the asymptotic properties of financial markets by using high frequency data. Researchers in this field begin to focus on intra-daily stylized facts rather than inter-daily, inter-weekly and even inter-monthly ones, which developed the research fields of market microstructure, provided a new view of financial derivatives pricing and market price mechanism and is also a new challenge.Long memory study is an important aspect in high frequency data research. Since Hurst found long memory in hydrology time series of tide data, and Mandelbrot laid the strict mathematical foundation of long memory, it becomes the focus of natural science and social science including economic and financial field. Long memory means that there is still impact on long lag of return or volatility series when a shock happened. It is very valuable to study long memory not only for analyzing market structure and estimating market tendency but also a impulsion to empirical study methods based on short memory. So it is also have important theory value.For above reasons there are many long memory studies of financial markets both abroad and at home. Those abroad studies include not only inter-daily low frequency data research but also many intra-daily stylized facts one based on high frequency data. However, contrast to abroad, there are many financial markets, for example stock market or bond market long memory study based on low frequency data, such as daily return, but few research by using high frequency data, though it is emerging in recent years. Moreover, long memory study of fund market, which is more and more an important tool used by institution investment is rarely seen. This article is a try for the aspect.The reminder of this paper is organized as follows: there is introduction of background and meanings of this paper and research situation both abroad and at home in chapter 1, mainly introduce the long memory research in the chapter.Because of the distinct intraday periodicity of volatility in fund market, which overshadows long memory properties research, we filter this intraday periodicity by using FFF regression in chapter 2.in this research we found that it is an excellent method to filter this intraday periodicity. After filtering that, intraday volatility series have no periodicities and can be found the long memory from the picture.In chapter 3,we introduce all kinds of long memory test methods and evaluate them. After that we use three of them to test long memory of intraday return series, volatility series and volatility series without intraday periodicity. We conclude that there is long memory in the intraday volatility series. Based on this conclusion, in chapter 4 we model the long memory in intraday volatility series by mainly using the GARCH long memory models, and contrast the results with that by using short memory model. Conclusion is in chapter 5.