Brendan Nyhan brings this graph and has more. Again, the point here is not that Romney didn’t gain ground after the first debate or that he will lose the election. He did gain ground, and he certainly could win. The point is that there was no change in the polls that would have justified either the Romney momentum narrative in the news media or the current “Slow Mo” narrative either. Since the race became essentially tied after the first debate, any movement in the polling margin in the graph above is indistinguishable from statistical noise — which polling averages reduce but do not eliminate.
My earlier thoughts are here.