目前国内文献基本上算是极少。
这几天做DCC-GARCH,没有推进到MSR阶段,主要是数据量太大了。
生成了模型,光粘贴就得好些时间。
然后就继续向前推进,就到了GO-GARCH阶段, 多变量广义正交GARCH模型。
继续图形主义吧。懂行的人,一看便知其份量。
如何实现的,唯有code知道。哈哈
这是建模的两个FTS
然后是执行code。转身去看美剧,或喝茶。
生成的正交特征根和scree plot
Principal Components Analysis
Date: 11/16/14 Time: 20:19
Sample (adjusted): 1993M01 2014M06
Included observations: 258 after adjustments
Balanced sample (listwise missing value deletion)
Computed using: Ordinary correlations
Extracting 2 of 2 possible components
Eigenvalues: (Sum = 2, Average = 1)
Cumulative Cumulative
Number Value Difference Proportion Value Proportion
1 1.136757 0.273514 0.5684 1.136757 0.5684
2 0.863243 --- 0.4316 2.000000 1.0000
Eigenvectors (loadings):
Variable PC 1 PC 2
RESID_1_01 0.707107 -0.707107
RESID_2_01 0.707107 0.707107
Ordinary correlations:
RESID_1_01 RESID_2_01
RESID_1_01 1.000000
RESID_2_01 0.136757 1.000000
均值方程1
Dependent Variable: ASI
Method: Least Squares
Date: 11/16/14 Time: 20:19
Sample (adjusted): 1993M01 2014M06
Included observations: 258 after adjustments
Variable Coefficient Std. Error t-Statistic Prob.
C 0.631280 0.264037 2.390878 0.0175
R-squared 0.000000 Mean dependent var 0.631280
Adjusted R-squared 0.000000 S.D. dependent var 4.241059
S.E. of regression 4.241059 Akaike info criterion 5.731372
Sum squared resid 4622.552 Schwarz criterion 5.745143
Log likelihood -738.3469 Hannan-Quinn criter. 5.736909
Durbin-Watson stat 1.958134
均值方程2
Dependent Variable: CSI
Method: Least Squares
Date: 11/16/14 Time: 20:19
Sample (adjusted): 1993M01 2014M06
Included observations: 258 after adjustments
Variable Coefficient Std. Error t-Statistic Prob.
C 0.374025 0.703038 0.532013 0.5952
R-squared 0.000000 Mean dependent var 0.374025
Adjusted R-squared 0.000000 S.D. dependent var 11.29246
S.E. of regression 11.29246 Akaike info criterion 7.690017
Sum squared resid 32772.58 Schwarz criterion 7.703788
Log likelihood -991.0122 Hannan-Quinn criter. 7.695554
Durbin-Watson stat 2.254290
因子成分方程1
Dependent Variable: PCOMP01_1
Method: ML - ARCH (Marquardt) - Normal distribution
Date: 11/16/14 Time: 20:19
Sample (adjusted): 1993M01 2014M06
Included observations: 258 after adjustments
Convergence achieved after 9 iterations
Bollerslev-Wooldridge robust standard errors & covariance
Presample variance: unconditional
GARCH = 1*(1 - C(1) - C(2)) + C(1)*RESID(-1)^2 + C(2)*GARCH(-1)
Variable Coefficient Std. Error z-Statistic Prob.
Variance Equation
C 0.143606 -- -- --
RESID(-1)^2 0.170637 0.085560 1.994347 0.0461
GARCH(-1) 0.685756 0.134819 5.086499 0.0000
R-squared 0.000000 Mean dependent var 1.35E-17
Adjusted R-squared 0.003876 S.D. dependent var 1.001944
S.E. of regression 1.000000 Akaike info criterion 2.789961
Sum squared resid 258.0000 Schwarz criterion 2.817503
Log likelihood -357.9049 Hannan-Quinn criter. 2.801036
Durbin-Watson stat 1.993169
因子方程2
Dependent Variable: PCOMP01_2
Method: ML - ARCH (Marquardt) - Normal distribution
Date: 11/16/14 Time: 20:19
Sample (adjusted): 1993M01 2014M06
Included observations: 258 after adjustments
Convergence achieved after 8 iterations
Bollerslev-Wooldridge robust standard errors & covariance
Presample variance: unconditional
GARCH = 1*(1 - C(1) - C(2)) + C(1)*RESID(-1)^2 + C(2)*GARCH(-1)
Variable Coefficient Std. Error z-Statistic Prob.
Variance Equation
C 0.025978 -- -- --
RESID(-1)^2 0.117901 0.041296 2.855022 0.0043
GARCH(-1) 0.856120 0.054178 15.80208 0.0000
R-squared 0.000000 Mean dependent var -9.23E-17
Adjusted R-squared 0.003876 S.D. dependent var 1.001944
S.E. of regression 1.000000 Akaike info criterion 2.712852
Sum squared resid 258.0000 Schwarz criterion 2.740394
Log likelihood -347.9579 Hannan-Quinn criter. 2.723927
Durbin-Watson stat 2.255072
然后是DCC
成分一
成分二
GARCH01
GARCH02
然后就是因子FGARCH 01
然后就是因子FGARCH 02
正交的分解,基于多元模型的GARCH。
极其敏锐地捕捉到了国际金融大事。
哈哈。。。。。。。。。
然后就是喝茶,看美剧, walking daed S5E5
然后就是牛叉的模型就是这样诞生的。
然后就是***** 真牛叉。