Comment by 0x008

My experience in general is that most time series model are inadequate in predicting time series except for very trivial cases of seasonalities or simple linear/nonlinear trends.

I think that you can throw any model you like at the problem but all you will do is overfit most of the time.