This is kind of living in the weeds.
Even 538 got the chances of each winning wrong on the eve of the election. Not sure how “overall 538 was quite accurate” turns into a completely wrong prediction. In the end, it was off as much as everyone else.
So what if they hedged their bets by saying, we could be wrong because all statisticians say that when dealing with very low confidence levels. However, when the rubber hit the road, 538 missed it completely just like everyone else.