我进行了滚动Garch估计,并将实际数据与其VaR进行了比较
gjr_garch_roll <- ugarchroll(garch_spec, xts_Obs_per_gjr_return, n.start = 1000, refit.every = 1,
refit.window = "moving", solver = "hybrid", calculate.VaR = TRUE,
VaR.alpha = 0.01, keep.coef = TRUE)
gjr_garch_roll
report(gjr_garch_roll, type = "VaR", VaR.alpha =0.01, conf.level = 0.99)
即使实际超出范围比预期的高,正确超出范围的Null假设也不会被拒绝。我在代码中做错了吗?
alpha:1%
预期超出:9.5
实际VaR超出:14
实际百分比:1.5%
无条件承保(Kupiec)
无效假设:正确的超越
LR.uc统计信息:1.908
LR.uc严重:6.635
LR.uc p值:0.167
拒绝为空:否