Is there any actual evidence that models are slow to take the effects of a SSW into account until it's under way. Similarly, it has been suggested here that the models are yet to factor in the impacts of the MJO.
It doesn't seem logical to say the GFS is not taking the amplitude of the MJO phase into account. And then make an independent forecast based of a GFS MJO plot... Similarly with strat impacts.
It's seems likely that the GFS/ECM/UKMO take many more drivers, signals and data into account than any human ever could when coming up with a forecast. Perhaps it's us enthusiast's are the ones failing to take drivers into account. And within there lies the mismatch between our expectations and NWP output, not the other way around.