Uncategorized

Wherefore The Big 12 Championship Game

This year the Big 12 brings back its football championship game, which has lied dormant since Colorado and Nebraska left the conference in 2011, allowing that dreadful Alabama-LSU rematch.

It’s back. But is the Big 12 better for it?

The main football argument for bringing back the CG is that it would boost the profile of the Big 12 Champ. In order to evaluate that claim we ran the upcoming 2017 season through our simulator 10,000 times, both with and without a CG capping the Big 12 season. Note that raw “playoff appearance” stats are skewed by incomplete schedules in the MAC, MW, and SB; data comparisons between the two simulations are what is most relavant.

The Big Picture

Yes, a team that goes 13-0 or 12-1 looks better to the playoff committee than a team which goes 12-0 or 11-1. Here is the data.

Playoff Chance w/o CCG w/ CCG
Undefeated 88.64% 90.38%
1-loss Champ 32.45% 35.23%
1-loss At Large 11.57% 13.01%
2-loss Champ 1.48% 2.01%
2-loss At Large 0.11% 0.10%

It is not a big boost, but it ‘is’ something, roughly 2-3% given an “average” season. Arguably, TCU would have made the playoff in 2014 if they had redemptively defeated Baylor in a rematch to close out the season.

Oklahoma Case Study

Let us look at the numbers for a single team, the Oklahoma Sooners, our preseason favorites under either system.

Playoff Selections w/o CCG w/ CCG
Undefeated 1171/1218 (96%) 872/907 (96%)
1-loss Champ 721/1693 (42%) 732/1613 (45%)
1-loss At Large 119/647 (18%) 51/270 (18%)
2+ losses 26 times (<1%) 34 times (<1%)

We see many of the same numbers. We see the 3% jump among 1-loss Champs. There is no percentage jump in the third category, but this is in part because of an apples and oranges comparison; a 1-loss At Large under a CG system is almost always because of a undefeated season followed by a loss in the CG. The benefit of appearing in a CG balances the lateness of a loss in that game. The final category too shows the general strength of schedule increase a CG has, especially for 2-loss teams.

BUT. Look at the actual numbers, NOT just the percentages.

In 10,000 simulations, Oklahoma makes the playoff 348 times FEWER when the Big 12 uses a Championship Game. How can this be?!?

The Big 12 Championship Curse

No. it is not a ‘real’ curse; unless you count those cast by statistics.

Looking back at Oklahoma’s numbers it appears they have a 12% chance of running the table to 12-0. They then have a 77% chance of beating the #2 Big 12 team a second time. Notice that a win in this game has only a nominal affect on their playoff chances (96% with or without the game) while a loss dumps them down to an 18% chance of making the playoff.

Taking a look at 11-1, Oklahoma gets a 3% boost for winning that extra game, but that is no where near worth the risk of dropping to a nearly 0% chance of making the playoff if they lose.

This is not an observation privy only to computers. Teams in the AP Top 5 have been losing CGs a lot. Even if you ignore games between two such teams, Top 5 teams are 29-11 (72.5%) in CGs. And, despite its truncated existence, Top 5 teams lost more games in the Big 12 CG than in any other CG; that is quite a curse. It was only in the third-to-last Big 12 CG, Top 5 teams finally achieved a 6-5 record against non-Top 5 teams.

The ONLY team that could benefit from such a game is an 11-1 team that gets a second crack at winning the conference. An 11-1 At Large has an 18% chance of making the playoff from the Big 12 this year, but that number balloons to 45% if they win their CG. That anecdote about TCU in 2014 is thus rather fitting.

A Compromise?

As per usual this blog cites the healing power of a Wild Card Game, the format geared specifically toward helping those top teams who can no longer win their conference.

Rather than forcing a 12-0 Big 12 team into another game which they have a 25% chance of losing, such a team can elect to pass on playing a 13th game and will even be given a guaranteed playoff spot if they are ranked in the Top 3. The same is then true of any Big 12 team with the caveat that they must win their conference or win the Wild Card Game against another potential 11-1 division runner-up if they want to make the playoff.

Adding a CG drops the total number of Big 12 playoff appearances from 4490 to 3513. Approximately half of the CG participants would have 1 or fewer losses, and approximately half of those teams would instead play in the FBS Wild Card Game if that were the alternative.

Selecting only border-line playoff teams to play a 13th game would be a boon to the Big 12. Not to mention, having a “Champions Only” qualification metric means that even a #5 Big 12 Champ can sneak into the playoff if a previously undefeated team loses in their own CG yet fails to drop below the Big 12 Champ.

That 2014 TCU team? They’d wind up playing Baylor in either of the CG or WCG formats. Would they have staved off Ohio State or edged out Florida State? We’ll never know.

Data

Playoff (w/W, w/L), if Win Out (probability) TEAM with ccg
C=Champ, A=At Large, L=Lose CCG, #=More Losses
16.89% (17.9%, 3.07%), 96.1% (9.07%) Oklahoma
C0:872/907, C1:732/1613, L1:51/270, WC: (156/205)-54
8.72% (14.44%, 1.86%), 98.5% (4.72%) OK State
C0:465/472, C1:324/679, L1:44/202, WC: (85/111)-130
3.24% (6.39%, 0.23%), 81.9% (3.32%) W Virginia
C0:272/332, C1:50/453, WC: (25/33)-5
1.79% (1.85%, 0%), 74.6% (2.01%) Kᴀnsas St
C0:150/201, C1:29/304, WC: (17/28)-3
1.24% (2.39%, 0.06%), 95.5% (0.67%) Texas
WC: (13/18)-23
0.97% (1.04%, 0.13%), 83.8% (0.93%) TCU
C1:17/115, WC: (10/10)-2
0.82% (0.86%, 0.17%), 65% (1%) Baylor
C0:65/100, C1:15/145, WC: (6/13)-5
0.81% (0.88%, 0%), 89.2% (0.65%) Texas Tech
WC: (13/12)-8
0.4% (0.47%, 0%), 86.2% (0.29%) Iowa St
WC: (3/3)-1
0.25% (0.31%, 0%), 86.3% (0.22%) Kᴀnsas
WC: (1/7)-2

Playoff (w/W, w/L), if Win Out (probability) TEAM without ccg
C=Champ, A=At Large, L=Lose CCG, #=More Losses
20.37% (21.55%, 3.89%), 96.1% (12.18%) Oklahoma
C0:1171/1218, C1:721/1693, A1:119/647, WC: (293/412)-1904
11.38% (18.93%, 2.23%), 96.4% (6.57%) OK State
C0:634/657, C1:369/807, A1:125/609, WC: (109/300)-1213
4.96% (9.25%, 0.62%), 81.2% (5.06%) W Virginia
C0:411/506, C1:75/598, WC: (90/122)-582
2.47% (2.56%, 0%), 69% (3.33%) Kᴀnsas St
C0:230/333, WC: (67/60)-334
1.65% (3.26%, 0.07%), 96.5% (0.87%) Texas
C1:56/103, WC: (23/40)-163
1.13% (1.19%, 0.17%), 64% (1.64%) Baylor
C0:105/164, C1:7/138, WC: (25/24)-162
1.12% (1.22%, 0.11%), 91.2% (0.91%) Texas Tech
WC: (14/31)-124
0.84% (0.9%, 0%), 77.7% (0.99%) TCU
C1:6/108, WC: (23/19)-124
0.53% (0.62%, 0.06%), 95.1% (0.41%) Iowa St
WC: (6/7)-52
0.45% (0.56%, 0%), 86.9% (0.46%) Kᴀnsas
WC: (5/8)-46

c0: 90.38% Big 12
c1: 35.23% Big 12
c2: 2.01% Big 12 with ccg
a1: 13.01% Big 12
a2: 0.1% Big 12

c0: 88.64% Big 12
c1: 32.45% Big 12
c2: 1.48% Big 12 without ccg
a1: 11.57% Big 12
a2: 0.11% Big 12
8.284-3.716(5.9671-3.0329) W Virginia (94/324/1825)(1447/3403) 5 with ccg
3.416-8.584(1.8989-7.1011) Kᴀnsas (1/25/125)(95/220) 2
7.7256-4.2744(5.3296-3.6704) Kᴀnsas St (34/179/1124)(853/2100) 3
3.7452-8.2548(2.3965-6.6035) Iowa St (10/40/178)(134/290) 1
9.5264-2.4736(7.2918-1.7082) Oklahoma (360/1689/5324)(4149/6697) 54
8.6434-3.3566(6.5779-2.4221) OK State (154/872/3537)(2151/4531) 130
5.7304-6.2696(3.8344-5.1656) TCU (27/97/416)(290/663) 2
5.3135-6.6865(4.0337-4.9663) Texas (40/124/508)(309/753) 23
5.0787-6.9213(3.4845-5.5155) Texas Tech (23/81/355)(233/537) 8
6.443-5.557(4.1856-4.8144) Baylor (22/82/465)(339/806) 5

8.3644-3.6356(6.0246-2.9754) W Virginia (112/496/2078)(1533/0) 582 without ccg
3.4556-8.5444(1.9217-7.0783) Kᴀnsas (13/45/132)(100/0) 46
7.7527-4.2473(5.3661-3.6339) Kᴀnsas St (61/247/1282)(893/0) 334
3.7205-8.2795(2.3735-6.6265) Iowa St (10/53/195)(133/0) 52
9.5318-2.4682(7.2834-1.7166) Oklahoma (442/2037/5528)(3992/0) 1904
8.6107-3.3893(6.5417-2.4583) OK State (226/1138/3780)(2155/0) 1213
5.692-6.308(3.7953-5.2047) TCU (21/84/429)(262/0) 124
5.3243-6.6757(4.0397-4.9603) Texas (39/165/551)(329/0) 163
5.0557-6.9443(3.475-5.525) Texas Tech (29/112/387)(238/0) 124
6.4398-5.5602(4.179-4.821) Baylor (37/113/564)(365/0) 162
Top 5 vs. non-Top 5
SEC 14-4, 7-0 since 2005
B12 7-5, 2-0 since 2007
ACC 5-2, 4-0 since 2011
P12 2-0, 2-0 since 2011
BT 1-2
overall 29-11, 16-11

Advertisements

One thought on “Wherefore The Big 12 Championship Game

Leave a Reply

Please log in using one of these methods to post your comment:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s