Here's a very simplified for why S&P likes UCF. Some of its principles are based on the Pythagorean theorem, which for sports originated in baseball (like a lot of advanced stats). Simply, it's this:
Win%=[(Points Scored)^2]/[(Points Scored)^2 + (Points Allowed)^2]
The exponent changes by sport, and is not always universally agreed upon in any particular sport.
It's a good second guess metric. This sort of metric shows flaws in a team's record. The positives for a metric like this is that it is based on scoring margin, which especially in an alternate possession sport like football, is a great predictor. The obvious negative is that it is prone to big results in small sample sizes having lopsided weights. I.E. you play 4 teams, 2 of those being lightweights you manhandle, and 2 being your relative equals. There are ways of working around that, though, with things like SOS filters and capping scores.
That's why it likes UCF, they have lopsided scores against all of their opponents. It's also why it likes Ohio State, who despite a loss, has been very impressive in its other showings.
Win%=[(Points Scored)^2]/[(Points Scored)^2 + (Points Allowed)^2]
The exponent changes by sport, and is not always universally agreed upon in any particular sport.
It's a good second guess metric. This sort of metric shows flaws in a team's record. The positives for a metric like this is that it is based on scoring margin, which especially in an alternate possession sport like football, is a great predictor. The obvious negative is that it is prone to big results in small sample sizes having lopsided weights. I.E. you play 4 teams, 2 of those being lightweights you manhandle, and 2 being your relative equals. There are ways of working around that, though, with things like SOS filters and capping scores.
That's why it likes UCF, they have lopsided scores against all of their opponents. It's also why it likes Ohio State, who despite a loss, has been very impressive in its other showings.