GARCH Model Cluster In Our Application Of The Financial Sector

Since Stock market appeard, its distinguishing features is the price volatility. How to accurately describe the behavior of stock market prices in order to determine the future market rate of return is the interests to all investors and securities markets. Thus, one of the core content of the study becomes the quantitative modeling of volatility into the volatility of financial assets, A large number of empirical studies have shown that the presence of financial data volatility clustering and fat tail of the ARCH characteristics of the peak, so the traditional method of modeling using the general time-series model to fit the volatility of financial data becomes less. Appropriate, and the present, parameters, non-parametric, semi-parametric GARCH modeling more and more use in the financial sector, they are currently a powerful tool in financial market. One parameter model is the most commonly used GARCH model, but the parameters of GARCH model for model misspecification inherent defects,It had set up ahead of the model. To address the model misspecification problem, scholars made non-parametric GARCH model, but the parameters of GARCH model has the existence of "curse of dimensionality" and is difficult to estimate the model to explain the problem. Therefore, in order to compensate for parameter GARCH model and the non - parameters of GARCH model shortcomings. Scholars introduce the semi-parametric GARCH models. Semi-parametric GARCH model parameter part and the non-parametric part of the organically combined, even though it may be less than the predictive power of semi-parametric model, but its parameters can model a certain part of the explanation, and also reduce the estimated non-parametric part of the value and the error between the true value. It becomes an even more superior model.In this paper, application of GARCH model, EGARCH model, TGARCH model, the residuals obey a normal distribution and Student T distribution to model fitting model, and compared the model's predictive power. We found that the relative predictive power of GARCH model is better. then we proposed non-parametric and semi-parametric GARCH models, finding that two kinds of volatility parameters of GARCH model is more than the true value. The semi-parametric GARCH model is better than non-parametric GARCH model in explaining the model, and can solve the non-parametric GARCH model "dimension disaster" defects.In the End, the article mainly discusses the GARCH model in practice, the application of traditional comparative law historic volatility GARCH model can significantly improve forecasts of convertible bond prices, narrowing the gap between the real data. In practice, the application of the theoretical arguments put forward.