2007 CGVRs

If you've followed my blogs for even a month ... ever ... then you know that Close Game Variance Ratio is something that I very much believe in. It's basically Phil Steele's idea that I gave a name to. CGVRs are calculated by finding the net ratio a team built in close games. If Team X had 4 wins of 8 points or less, and 2 losses of 8 points or fewer, they had a +2 CGVR. If a team has a +2 CGVR or higher, you can confidently predict that they will NOT improve their record. If a team has a -2 CGVR or worse, you can expect them to improve their record the following season. I've added in the 2006 outcomes to the ol' database, so from 2001-2006, here is how accurate the CGVR machine has been.

+4 and up - Has occurred 43 times, with 93% of the teams having the same or worse record the next season
+3 - Has occurred 30 times, with 73% of the teams having the same or worse record the next season
+2 - Has occurred 62 times, with 68% of the teams having the same or worse record the next season

Does your team fall into the parameters?

Hawaii +5 (12-1)
Mississippi State +4 (8-5)
Virginia +4 (9-4)
Oregon State +4 (9-4)
Tulsa +4 (10-4)
Tennessee +3 (10-4)
Northwestern +3 (6-6)
Kansas +3 (12-1)
Arizona State +3 (10-3)
East Carolina +3 (8-5)
Memphis +3 (7-6)
Toledo +3 (5-7)
Florida Atlantic +3 (8-5)
Kentucky +2 (8-5)
LSU +2 (12-2)
Wisconsin +2 (9-4)
Michigan +2 (9-4)
Texas +2 (10-3)
Missouri +2 (12-2)
Boston College +2 (11-3)
NC State +2 (5-7)
Louisiana Monroe +2 (6-6)
Connecticut +2 (9-4)
New Mexico +2 (9-4)
San Diego St +2 (4-8)
Bowling Green +2 (8-5)
Miami, Oh +2 (6-7)
Temple +2 (4-8)

If your team is listed above, be afraaaaid, be verrrry afraaaaaid. Now, the other side of the coin: negative CGVRs.

-4 and worse - Has occurred 29 times, with 86% of the teams having the same or better record the next season
-3 - Has occurred 46 times, with 72% of the teams having the same or better record the next season
-2 - Has occurred 62 times, with 71% of the teams having the same or better record the next season

Again, here are the teams that apply:

Minnesota -5 (1-11)
Washington -5 (4-9)
SMU -5 (1-11)
Michigan State -4 (7-6)
Vanderbilt -3 (5-7)
Maryland -3 (6-7)
North Carolina -3 (4-8)
UCLA -3 (6-7)
UNLV -3 (2-10)
Northern Illinois -3 (2-10)
Kent State -3 (3-9)
Louisiana Lafayette -3 (3-9)
Alabama -2 (7-6)
Kansas State -2 (5-7)
Florida State -2 (7-6)
Miami -2 (5-7)
Duke -2 (1-11)
Arizona -2 (5-7)
TCU -2 (8-5)
Colorado State -2 (3-9)
Marshall -2 (3-9)
Nevada -2 (6-7)
Idaho -2 (1-11)
Utah State -2 (2-10)
W. Michigan -2 (5-7)

If your team is listed above, you can hold out hope that the stats are in your favor. As always, these numbers are only tools, and are not the ONLY thing to look at. But you can still amaze your friends this autumn.

Revisiting 2006 CGVRs

In January of 2007, at my original Tennessee blog, I outlined the teams from 2006 that fell under the CGVR (Close Game Variance Ratio) parameters. CGVR is just the tool Phil Steele came up with years ago, but with a fancy name and specifically revisited every year. If you're new to CGVR, here's the deal: Mr. Steele found some strong statistical correlations between teams that either won or lost many close games and their record the next season. A close game is defined as any game that ends within one possession on the scoreboard (thus, 8 points). So if a team wins 3 close games and loses 1 close game, that team's CGVR is +2. Steele found that the correlations become strong at +2 and above and -2 and below. Teams with +2 or more cannot be expected to improve their record the next season, whereas teams with -2 or worse cannot be expected to regress the next season. Make sense? Good. Here are the 2006 CGVRs, the expectation, the 2007 record, and if the CGVR got it right.

Team	2006 CGVR	2006 Record	Expect	2007 Record	Correct?
Wake Forest	+5	11-3		9-4
Auburn	+5	11-2		9-4
Florida	+5	13-1		9-4
Maryland	+5	9-4		6-7
Oregon State	+4	10-4		9-4
Rice	+4	7-6		3-9
Boise State	+4	13-0		10-3
Iowa State	+4	4-8		3-9
Arkansas State	+4	6-6		5-7
Kansas State	+3	7-6		5-7
Texas Tech	+3	8-5		9-4
Wisconsin	+3	12-1		9-4
Notre Dame	+3	10-3		3-9
Ohio	+3	9-5		6-6
Troy	+3	8-5		8-4
Boston College	+2	10-3		11-3
Texas A&M	+2	9-4		7-6
LSU	+2	11-2		12-2
Tennessee	+2	9-4		10-4
Georgia	+2	9-4		11-2
Kentucky	+2	8-5		8-5
Ohio State	+2	12-1		11-2
Purdue	+2	8-6		8-5
Rutgers	+2	11-2		8-5
Cincinnati	+2	8-5		10-3
Southern Miss	+2	9-5		7-6
Idaho	+2	4-8		1-11
Lousiana-Lafayette	+2	6-6		3-9

As you can see, Phil Steele's theory continued to be accurate. Twenty of the 28 teams that qualified did not improve their records, which is 71.4%. Also, every single team with a CGVR of +4 or more had worse records the next year, with over 2 less wins than in 2006.

Onto the other side of the coin.

Team	2006 CGVR	2006 Record	Expect	2007 Record	Correct?
North Carolina State	-5	3-9		5-7
Memphis	-5	2-10		7-6
Eastern Michigan	-5	1-11		4-8
Florida International	-5	0-12		1-11
Louisiana-Monroe	-5	4-8		6-6
New Mexico State	-4	4-8		4-9
Colorado	-4	2-10		6-7
Duke	-3	0-12		1-11
Missouri	-3	8-5		12-2
Oklahoma State	-3	7-6		7-6
Vanderbilt	-3	4-8		5-7
Michigan State	-3	4-8		7-6
Illinois	-3	2-10		9-4
Air Force	-3	4-8		9-4
UAB	-3	3-9		2-10
Miami (OH)	-3	2-10		6-7
Clemson	-2	8-5		9-4
Florida State	-2	7-6		7-6
Alabama	-2	6-7		7-6
Mississippi State	-2	3-9		8-5
Iowa	-2	6-7		6-6
Minnesota	-2	6-7		1-11
San Diego State	-2	3-9		4-8
UNLV	-2	2-10		2-10
Houston	-2	10-4		8-5
Nevada	-2	8-5		6-7
North Texas	-2	2-10		2-10

Again, the theory holds water. Twenty-two of 27 qualified teams fit the mold. Overall, 42 of 55 teams that qualified somewhere within the parameters set my Phil Steele did what Steele predicted. That's 76.4% of the teams from last year - but people will still blow these facts off.

Coming later this week is the list of 2007 CGVR qualifiers.