This - I believe - is exactly why those prediction are so misunderstood. Most pollster usually give the expected vote share of each candidate. In this regard, a 60/40 split between two candidates in the polls means that the first candidate is nearly guaranteed to win the election. And for most elections, the split usually is much closer (a 53/47 split is already considered a "wide" margin). So years after years you see split like 55/45, 52/48, 57/43, ... and the expected candidates always wins.

So when someone analysing an election comes up with a 70/30 split, the number seems incredibly high - there is now way we are getting this one wrong. It takes at least a second look to see that this guy is talking about something different and that the race actually is quite close this time.