The typical "top ten" P5 team may have four "real games" that are contested, four "pastry games" that are certain wins, and four games they could lose if things go wrong for them. Sometimes it's 3/5/4, but whatever.
A very very good team may have a 75% chance of winning the 4 contested games, but even so, the odds of winning them all are not good. Then they might have a 90% chance of winning the 4 possible loss games on top of that. Those games are like Clemson at Syracuse last year.
It's just not likely, even for an Alabama. Now, if you have 10 teams in that category, it starts to be likely for 1-2 of them, which is what we routinely see in practice.