You're right, it's more like 'outside of 2 standard deviations,' which would be around 2-4%.
But still.
I thought it incorporates the range of each well, as a strong FCS team is better than a bottom-feeder FBS squad.
Michigan is around 50-4 vs G5 teams the last 50 years. And I'll assume they're 11-1 or 20-1 or 35-1 vs FCS teams in that timeframe as well. We all know the one.....because it was a crazy outlier legendary event.
Idk how many wins vs FCS they have, but I doubt the win% is much higher than 50-4 (.925). Not enough games to get there with that 1 loss.
.
Anywho, I think my point stands. There's a cutoff between teams that have a reasonable chance at winning and those who do not. By definition, the latter group will win sometimes, but not often enough for it to be significant. For every major upset, there's 5 or 6 or 10 examples of 52-7 blowouts.