Actually, it depends on far apart the teams are. If the teams are extremely unevenly matched (generally Colorado or Missouri in the 2000s for the most part), then it really does not matter. However, if a team is not significantly better, the rematch favors the loser of the first game. My best example is 2011 LSU. 2011 LSU was the best team in the country (and in my opinion still has the overall better resume than 2011 Alabama and should still have been the national champs after the two teams split the series). However, 2011 LSU was only slightly better than 2011 Alabama and Alabama easily won the rematch.
The Big 12 #1 and #2 rematching every year is going to be a lot more evenly matched and will produce far more rematch upsets than the Big 12 South v. Big 12 North where many times the Big 12 North team was far and away overmatched.
If Texas and OU had replayed each other for the Big 12 title in the 2000s, the loser of round one would have come away with a lot more victories than Colorado or Missouri did.
OU shortsightedly pushed for this rematch title game thinking it would be like playing Colorado and Missouri in the 2000s. The first game against TCU was like that. However, at no point, were Colorado and Missouri the second best team in the Big 12 in the 2000s. Rematching #1 v. #2 is not like that at all and will produce far more rematch upsets/hurt the Big 12's playoff changes in the long run.
I am not saying Texas wins this rematch this year. However, excluding outside factors like either team having injures, etc, I would say Texas has a far better shot in the rematch than it did in the cotton bowl in October just like OU had a far better shot last year in the rematch than it did in the cotton bowl last October.