Research Of The Ability Of Fitting And Forecasting The Volatility Of Stock Market Based On The Improved GARCH Mode

As an emerging capital market. China's stock market is at an early stage of development with a short history of SMSE Stock and Growth Enterprise Market. As a result, it always interacts with other market imperfection in the operating system or oversight mechanics. Because of these market imperfection, some semioccasional malice investment can lead to drastic fluctuations in China's stock market. Naturally, how to give an discipline of the stock price fluctuation and predict future yield volatility accuratly causes widespread concern in academic circles. Therefore, the research on the flitting result of fluctuation ratios and the predictive power of a model is of great importance. Above discussed points clearly prove this thesis to be of significant theoretical and practical importance.GARCH family models are a prevalent group of models used in empirical research on finance time series, for example, the stock yield. Howerver, it is one of the few areas where comparative study about the main board, small and medium-sized board and the growth enterprise market. Enlightened by the above-mentioned research review, this thesis adds two main variables--stock trading volume and price range--in three mature GARCH family models, respectively corresponding to three kinds of stock market (main board, small and medium-sized board and the growth enterprise market), to explores the result of model optimization as well as its explanation strengths for different market characteristics, so as to enable the regulator to make effective suggestions for the risk management and forecast for yield volatility.This thesis has six chapters. Chapter one is introduction, concluding research background and significance, the outline and framework of the research, research methodology and the deficiency in innovation. Chapter two is the summarization of status quo, containing the status quo and the development courses of the main board, small and medium-sized board and the growth enterprise market in China. Chapter three is literature review, in which contains review from two perspectives: the object of study and the research approach. Chapter four is model building and optimization based on modified GARCH family models, which introduces the model forms, parametric estimation and improved methods specificly. Chapter five is empirical analysis, first conducting a research on the market characteristics of the main board, small and medium-sized board and the growth enterprise market; then adopting modified GARCH family models to evaluate the fitting result and verify its validity in yield prediction; finanly find out suited models for three different parts of China's stock market according to the related index system. Chapter six is the research conclusion and suggestions. This part puts forward a series of research conclusion and prospectes the future development in related research field.The main conclusions of this paper are as follows:Firstly, the analysis of volatility characteristics shows several common points in different markets.(1)Every market has the characteristics of volatility clustering, long memory and leverage effect;(2)Descriptive statistics shows that the GEM market,SME board market and main board market rank from the first to the third according to the average rate of return and the standard deviation from high to low.That is because GEM market is younger than SME board market and main board market. Secondly, GARCH with volume factor and price range factor can improve the fitting and forcasting ability of GARCH models.Thirdly,according to the minimum AIC information criterion, the root mean square error (RMSE) and mean absolute error (MAE), the bwst model for mainboard is EGARCH-R with the Normal distribution.The best for SME board market is EGARCH-V-R with GED distribution,and the best for GEM market is EGARCH-R model with GED distribution.This paper mainly has two special points:one is research object and another is empirical method.In detail, firstly, a lot of literature study the single market with a single model,but in Chinese stock market, different markets have different applicable models.Therefore, this paper will analyze the main board market, the small and medium-sized board market and the gem market to screen out the best fitting and forecasting mode lfor each of them. Secondly, according to the mixture of distribution hypothesis, volume and price range can represent information flow as important alternative variables.Considering the price range and the volume reflect information flows from different angles, I try to put both of them into the GARCH models with the hope of that it is a nice try. The relevant literature show that split share structure reform in the long term to reduce the stock market wave dynamic, weaken the impact of early information on the current, but prolongs the impact time, which is neglected by lots of documents. So considering that, this paper will choose data from the point when the reform nearly ended, which is from 1/4/2007 to 6/30/2015.In adition, all ducuments apply the original or the logarithm value of volume and price range to GARCH models, but that often leads to negative coefficient of GARCH and TGARCH models which is not allowed by rules. So I think it is necessary to apply their growth rate to GARCH models.