Research On Decomposition Of Index Fluctuation Based On High-frequency Data

With the development and perfection of the financial market,the study on the volatility is playing a more and more important role in the academic realm.With the help of continuous improvement of computer storage technology,the study of volatility has begun to focus on the high-frequency field.Therefore,the study object of this paper is the high-frequency volatility of SSE 50 index.Specifically,this paper explores the volatility of the index in brand new way,which is quite different form previous research,by decomposing the volatility of the stock index into two components: sum of index constitute stocks' volatility and the degree of correlation among them.Also,the trend of the index within the sample interval can be divided into three parts: rising trend,fluctuating trend and falling trend.After a detailed introduction of calculation method and the corresponding explanation of the meaning of the dismantling of index,the statistical analysis will be carried under different market trends.The statistical data of two different volatility sub dimensions will be listed in order to analyze the difference of micro structure of index volatility under different trends.Additionally,different copula models will be applied to measure the correlation between index volatility and its two sub dimensions.By conducting quantitative analysis of the two sub dimensions of the index volatility under different stock market circumstances,the paper tries to reveal the economic principles hiding behind such differences.The last but not the least,due to the better understanding of the market information contained in the historical data this dismantling can bring,in this paper,we try to apply the additional information extracted from the index fluctuation to the fitting and forecasting model using high-frequency data.Based on the deformation of the classical HAR model cluster,a new volatility fitting and forecasting model is proposed,and the fitting and forecasting results of the new model will be detailedly compared with mainstream models.