這是本研究的重要部分之一，以制定所需的和最好的模型，能夠處理這麽多的數據，並能夠做時間序列分析。在上述各種方法的幫助下，最終確定了GARCH模型。這是最終確定的，因為它是分析或回歸財務時間序列數據和特征的最佳方法或模型之一(Gurarda et al.， 2016)。該模型也適用,因為在中國上市公司收集到的數據高於50,有很多數據點的數量等一系列誤差項的方差或創新實際上函數相關的總體大小錯誤在之前的時期。
t2 =ω+αt-12 +βt-12
rt = t + t
t = tzt
∅= < 1
Methodology: Build the financial model based on the GARCH model
This is one of the important part of this study to formulate the required and best model that is capable of handling this much data and is able to do the time series analysis. With help of the various methods as discussed above, the GARCH model was finalized. This was finalized because it is one of the best method or model to analyse or regress the financial time series data and characterize (Gurarda et al., 2016). This model is also applied because in the collected data above for 50 Chinese listed companies, there are lot number data points which were there in a series such that the variance for the error term or the innovation it actually the function which is related to the overall size of the error taken during the previous time periods.
This was the reason for choosing GARCH model as the best suitable method with three different models for this study to provide the required results or outputs form the given data. In GARCH model the volatility is given as:
zt= independent and identically distributed process
t= conditional mean of rt
t2= conditional variance of rt
With given conditions as
These conditions are suitable for GARCH model to adopt in the process for analysing the leverage effect. The leverage effect is normally caused with the fact that negative returns have a greater influence on future volatility than do positive returns.