I want to analyze asset prices directly (not their returns) using ARIMA-GARCH model. Since the time series I have is stationary only if integrated once, I need an ARIMA (p,1,q) model.
Using rugarch though I cannot find how to specify an integration order 1 for my series. Using the following for example
spec <- ugarchspec(variance.model = list(model="sGARCH", garchOrder=c(1,1)),
mean.model = list(armaOrder=c(1,1,0))
distribution.model = "norm")
gives me an ARMA(1,1) (or specifically ARFIMA(1,0,1). If I use :
spec <- ugarchspec(variance.model = list(model="sGARCH", garchOrder=c(1,1)),
mean.model = list(armaOrder=c(1,1,1), arfima = T),
distribution.model = "norm")
will give me ARMA(1,1) and another variable ARFIMA with a coefficient, which to be sure I don't understand and in both cases I have a big mu (which is the mean of the price) which I understand - but that means that the series it's not integrated.
Is it possible to integrate series using ugarchspec?
question from:
https://stackoverflow.com/questions/66054968/rugarch-fitting-arima-non-stationary-series 与恶龙缠斗过久,自身亦成为恶龙;凝视深渊过久,深渊将回以凝视…