essay格式：GARCH模型的测试假设
在与资产回报相关的案例中，有两个重要特征。这两个特征是波动性聚类和高信度的存在。在这种情况下，波动可以被认为是变化或条件标准差。为了解决这些类型的事实，最好使用广义自回归条件异方差(GARCH)模型。在杠杆效应和误差分布的情况下，GARCH模型在某种程度上不能解释偏态或不对称，但能够完成所有其他需要的输出。GARCH、TGARCH、EGARCH和GJRGARCH可以与标准化t和标准化非对称t创新对50家公司的数据进行比较。
essay格式 ：GARCH模型的测试假设
首先假设GJR GARCH模型可以拟合到不对称的学生t分布。另一方面，考虑G=0时无杠杆模型和偏度参数= 1时模型创新的另一对称模型。在这两个模型的帮助下，这个假设检验可以用来验证上述事实的缺失或存在。GARCH、TGARCH和EGARCH模型也可以采用同样的过程来检验假设。在进行预测时，为了获得较高的准确性，将置信水平分别考虑为95%和99%。为了理解统计方法是预测目的的最佳方法之一，零假设被认为是检验似然比的最佳方法(Abor, 2007)。
essay格式 ：GARCH模型的测试假设
In cases related to the return of assets there are two important features. These two features are the presence of volatility clustering and high kurtious. The volatility in this case can be considered to be the variation or conditional standard deviation. To solve these styled facts, it is better to use generalized auto-regressive conditional heteroskedastic (GARCH) model. GARCH model is in some way not able to explain skewness or asymmetry in case of leverage effect and distribution of errors but able to do all other required outputs. GARCH, TGARCH, EGARCH and GJRGARCH can be compared with standardize t and standardize asymmetric t innovations for total of 50 company’s data.
essay格式 ：GARCH模型的测试假设
Assume first that GJR GARCH model can be fitted to the asymmetric student t-distribution. Then on the other hand, consider a model which has no leverage when G=0and another symmetric of the model innovation when skewness parameter = 1. With help of these two models, this hypothesis testing can be used to verify the absence or presence of these facts as discussed above. This same procedure can be followed for GARCH, TGARCH and EGARCH model to test the hypothesis. The confidence level is considered as 95% and 99% in order to get the high accuracy while making the predictions. The null hypothesis is considered to the best to test the likelihood ratio in order to understand a statistical method is one of the best method for prediction purpose (Abor, 2007).
