Mailbag: Do Underdogs Have An Edge In College Football Rivalry Games?

October 26, 2012 - by David Hess

We received a great question this morning from Andy Cox of Crashing The Dance.

Hey guys,

I’m curious (re: my Heels taking on N.C. State this weekend) whether you guys have done or are aware of any studies on the accuracy of point spreads in rivalry vs. “regular” games, particularly in college football. I’d take it on, but I don’t have the data.

There’s lots of Twitter chatter that the Heels as a 7.5 favorite is a bit high, especially considering the recent series history. Based on power rankings, the spread makes sense, but I’m curious how much the lines reflect the “intangibles” of rivalry games.

This made me curious, too, so I decided to dig into our historical data.

Defining Rivalries

We don’t have any “rivalry” labels in our database, so I had to improvise. I decided that grouping games by whether the two teams were A) from the same state, and B) from the same conference ought to do the trick. Obviously this will miss some rivalries (Michigan-Ohio State), but if there is a large effect, I would expect to see it in this data.

So, what did I find? Here is a table of ATS results, from the perspective of the underdog, for all games in our database involving two FBS or Division-I teams. The data goes back through 1995:

Underdog ATS Results In “Rivalry” Games, 1995-2012

State	Conf	# of Games	Cover%	ATS margin	Note
diff	diff	3674	51.5%	+0.4
diff	same	6524	50.0%	-0.1
same	diff	655	48.5%	-0.9
same	same	920	53.4%	+0.7	"rivalry" group

Hmm, pretty interesting! We see that in games in our makeshift “rivalry” bin, the underdogs have beaten the spread over 53% of the time, often enough that it would have been profitable to blindly bet them.

However, here is the same table, but looking at only the last 10 years. The effect disappears:

Underdog ATS Results In “Rivalry” Games, 2003-2012

State	Conf	# of Games	Cover%	ATS margin	Note
diff	diff	2169	51.1%	-0.3
diff	same	4261	50.5%	+0.0
same	diff	392	47.8%	-1.7
same	same	576	50.8%	+0.1	"rivalry" group

It looks to me like this piece of conventional wisdom actually started out as the truth, but is no longer accurate.

In other words, perhaps Vegas used to have a blind spot when it came to rivalry games, but it appears they’ve adjusted their rear view mirror.