Which don't always tell the story. Our narrow wins against far lesser teams such as Ball State and Vandy play a major role, while Michigan's many blowouts helped them. So by the pure math, Sagarin's ratings are correct.
What they don't reflect is how the coaches coach the games. When Kelly gets a lead, especially against a poor team, he tends to take the air out of the ball at times instead of running up the score. He prefers to milk the clock and let the defense win the game, whereas other teams might try to keep their foot on the gas pedal.
If Sagarin had some sort of equal points bonus for winning a game, say 20 extra for every win regardless of opponent, it might better reflect how teams are ranked. So for example, Ohio State would have beaten Michigan 82-39 under this system, and Michigan would have been credited with a 43-point loss instead of 23.