This is one I have never forgotten, but I'm sure it's been over 10 years and so I don't know if there has ever been a correction - like the NBA zig-zag.
Once upon a time - the strongest trend out there was that in a divisional rematch - the second game went opposite of the first game, in terms of totals. It had an 80% success rate.
It strikes me that there would be two reasons for this: (1) The line would be skewed based upon the first game. So if the first total were 45, and the game went over at 54, you'd see an overreaction for that result in setting the second line or in betting it up, which is unjustified from a single game; (2) The coaching staffs would adjust their game plan defensively to counter what had worked offensively in the first game. Of course, the same would apply in the opposite direction for a first game that is low scoring, in terms of betting line overreaction and coaching staff adjustments.
You wouldn't see the same thing as to betting the side in these rematches, because of the highly touted revenge factor and the impact of home field.
I've been meaning to check on how well that trend has stood up, and so I am also interested in your query.