I've implemented Justin7's "strength of schedule" rating (from his book Conquering Risk) into a few of my models. The models are not based on NFL but I can't figure out for the life of me how using them has made the error on my model increase.

I computed the strength of schedule ratings, then ran a regression of SOS differential with error in margin of victory, found the adjustment factor and it has made my model worse. Does anyone have any other ways in which they factor in a team's strong/weak fixture?