Wednesday, January 9, 2013

Bowl Season Winners (hint: Alabama)



Bowl season is finally over for college football and now it’s time to dig into the data and see who won. We’ll start with the obvious. Alabama won. Alabama could scarcely have won by more. Moving to the conference-level we can take a quick look at the records before getting some more accurate answers.

The Conferences


W
L
W%
CUSA
4
1
0.800
ACC
4
2
0.667
SEC
6
3
0.667
WAC
2
1
0.667
BigEast
3
2
0.600
Pac12
4
4
0.500
SunBelt
2
2
0.500
Big12
4
5
0.444
Ind
1
2
0.333
B1G
2
5
0.286
MAC
2
5
0.286
MWC
1
3
0.250
Conference Records (2012-13 Bowls)

Since each team also had a Sagarin predictor ranking (the ranking predicts the score differential between any two teams), we can also look at what conferences were supposed to do, then compare that record with what actually happened.


W
L
W%
Fav
Und
eW%
Dif.
ACC
4
2
0.667
1
5
0.167
0.500
CUSA
4
1
0.800
2
3
0.400
0.400
MAC
2
5
0.286
0
7
0.000
0.286
BigEast
3
2
0.600
2
3
0.400
0.200
B1G
2
5
0.286
1
6
0.143
0.143
Ind
1
2
0.333
1
2
0.333
0.000
MWC
1
3
0.250
1
3
0.250
0.000
SEC
6
3
0.667
8
1
0.889
-0.222
Big12
4
5
0.444
6
3
0.667
-0.222
Pac12
4
4
0.500
6
2
0.750
-0.250
WAC
2
1
0.667
3
0
1.000
-0.333
SunBelt
2
2
0.500
4
0
1.000
-0.500
Conference Records plus Sagarin Predictions (2012-13 Bowls)

The story changes a bit here. The ACC and CUSA are still the class of the bowl season, but the SEC drops down to the bottom half on account of their teams’ lofty rankings from the regular season. The WAC and Sun Belt actually had pretty good bowl seasons in overall record – until you notice that their teams were expected to win every game.

Why use this method rather than their record? As I complained wrote about in my recent post on the Big Ten, teams are not necessarily selected to ensure competitive games. Even when bowls want a certain matchup to generate interest, conferences may step in as they did when the Capital One bowl wanted Texas A&M vs. Northwestern (they got conference championship game losers Georgia and Nebraska).

Why not use Vegas lines? Vegas lines, while primarily based on score expectation, may also incorporate information about how the public is betting to make sure that not all of the money comes in on one side. As an example, if BYU is playing UNLV in a bowl and the actual game estimate has BYU favored by 6, the line may shift toward UNLV (maybe BYU by 3 or 4) because BYU’s alumni base is unlikely to partake in the action (see here) while UNLV’s is close enough and numerous enough to overwhelm unbiased bettors.

Why not adjust the Sagarin lines to account for home field advantage? Because that doesn’t improve them. I took a look at the adjustment suggested by Sagarin – 3 points for the home team – with a further 1.5 points for schools that were “near” rather than home (think Texas Tech playing in Houston). The raw lines had an average error of 0.4 points. The adjusted lines had an average error of 1.0 points. Isolating just the Home and Near teams showed mixed results: Home results were significantly worse (1.1 raw and 4.1 adjusted) while Near were better (6.0 raw and 4.5 adjusted).

The Teams

I know, I know, I already said that Alabama won everything. For fun though, let’s take a look through the rest of the teams and see who showed up for their games and who came out flat. Since one team’s underperformance is another’s outperformance, this being a zero sum game, we’ll look at it by bowl game.

Performance of 2012-13 bowl winners relative to Sagarin predictions

This graph includes all of the winners in the 2012-13 bowl season. As elsewhere in this post, the Outperformance/Underperformance represents the actual margin of victory minus the projected margin of victory. Five of the winners underperformed with Texas Tech “leading” the way by winning despite underperforming by 12 points. Those schools that had head coaching changes underperformed by 6.6 points.

Alabama’s performance, the 9th highest outperformance, is remarkable because of the competition. As the one bowl that is supposed to match up relatively even teams – numbers 1 and 2 in the BCS – the National Championship Game should not see this kind of outperformance. Regardless of what should have happened, Alabama came out hot and Notre Dame had no answer throughout. I have a feeling there is a decent chance I’ll be writing this again next year with only one of those schools changed.

No comments:

Post a Comment