Sagarin uses two methods and blends them. You'll note that Syracuse is #2 in elo chess, which only takes wins and loses into account. Predictor uses margin of victory as well. Historically Syracuse is better on elo chess than on predictor, because stall ball tends to tighten up games. Not so much that we lose, but enough to leave Syracuse underrated in Sagarin's predictor model.
So "Predictor" is a horrible predictor for Syracuse... And I'm still all in.