Not Michigan-related, but there are enough good quant types here that hopefully someone can point me the right way.
What research is there to show that non-yardage ways of winning ball games, e.g. net turnovers, special teams, net penalties, etc., are or aren't sustainable?
Phil Steele has published some stuff in his annuals (e.g. 2009, I didn't get the 2010) that strongly suggests that you can't hope to get outgained and win consistently with a low yards-per-point. IIRC the only exception he noted was that you can get good at holding onto the ball, but not reproduceably good at forcing turnovers.
(As an aside, I wish Phil would take an offseason and learn to do simple regression analysis.)