The only way I could see it being falsified is if we got access to a new source of historical data that somehow was more accurate than all existing sources (which seems unlikely) or by waiting to see what happens and obtaining new data that way (which might give an unpleasant answer).

I'm not sure how the concept of a theory being falsifiable applies when you can't really run repeated controlled experiments.