JnJ Rankings

Better, But Don't Break the Bank

2012-01-19T18:15:00.000-08:00

Three Better, but Still Not Better Enough

The bowl-predicto for the revised/flattened model was indeed better, but went up only to 21 (from 18). More research required. Hopefully we can find an optimization statistic that yields a more solid/even result across =all= weeks of the season, at which point we will exercise it on some past historical data (entire other seasons plus post-season games) for further validation.

Experimental Ratings, Closer to W/L Only

2012-01-17T19:09:00.000-08:00

Post-Season Fudge Factor, or Next Season's New Model?

The ratings/rankings below use the same basic JnJ model, but flatten the results of games more than the standard method. The standard JnJ model attempts to flatten scores to some extent, but care must be taken to avoid the equivalent of integer round off, which can cause model instability.

This ranking improves both the overall season retrodiction (up about 2% to 89.1%, getting close to Coleman's theoretical MinV value of 91.47%) and correctly retrodicts all 35 bowl games. Tomorrow we will try to re-run the pre-bowl convergence to see how this model would have done on the actual bowl predictions. The ratings below are normalized to an average of 100.00.

Rank	Team	JnJ Rating
1	Alabama	175.32
2	LSU	173.01
3	Oklahoma St	160.41
4	Arkansas	153.37
5	West Virginia	151.48
6	Oregon	149.41
7	Stanford	148.76
8	Southern Cal	147.53
9	Southern Miss	147.3
10	Baylor	144.85
11	TCU	144.38
12	Boise St	143
13	Oklahoma	140.73
14	South Carolina	138.29
15	Kansas St	137.33
16	Michigan St	136.92
17	Wisconsin	135.31
18	Clemson	134.93
19	Michigan	133.3
20	Georgia	131.43
21	Auburn	130.78
22	Nebraska	129.58
23	Houston	128.78
24	Missouri	127.56
25	Washington	125.18
26	Virginia Tech	125.08
27	Florida St	122.78
28	Utah	120.8
29	Virginia	120.07
30	Texas	119.15
31	Mississippi St	118.32
32	Miami FL	116.75
33	Penn State	116.05
34	Florida	114.66
35	Notre Dame	113.65
36	Texas A&M	113.47
37	Brigham Young	113.15
38	Georgia Tech	112.9
39	North Carolina	111.49
40	Rutgers	110.44
41	Northern Illinois	109.52
42	Tennessee	109.01
43	Louisiana Tech	108.5
44	Purdue	105.76
45	Arizona	105.14
46	Tulsa	105.01
47	Iowa	105
48	Iowa St	104.74
49	Ohio State	103.39
50	Minnesota	102.24
51	Cincinnati	102.18
52	Ohio U.	100.22
53	Illinois	99.78
54	Northwestern	99.39
55	Utah St	97.52
56	Toledo	97.15
57	Vanderbilt	96.57
58	SMU	95.6
59	Wake Forest	95.24
60	Arkansas St	94.18
61	Texas Tech	93.96
62	Western Michigan	93.77
63	UCLA	93.23
64	Duke	92.35
65	Nevada	91.7
66	Marshall	91.08
67	Temple	90.33
68	Pittsburgh	89.58
69	Bowling Green	89.03
70	Wyoming	88.88
71	East Carolina	88.01
72	Louisiana-Lafayette	87.3
73	Boston College	87.27
74	California	86.98
75	North Carolina St	86.98
76	Central Florida	86.81
77	San José St	86.66
78	San Diego St	85.38
79	Navy	85.3
80	Florida Int'l	85.22
81	Western Kentucky	85.18
82	Kentucky	84.82
83	Miami OH	84.73
84	Rice	84.18
85	Louisville	83.47
86	Air Force	82.95
87	Kent St	82.22
88	Kansas	81.25
89	Connecticut	81.03
90	South Florida	79.57
91	UTEP	79.25
92	Mississippi	79.13
93	Arizona St	78.97
94	Washington St	78.57
95	Oregon St	78.11
96	North Texas	77.5
97	Ball St	74.47
98	Fresno St	73.07
99	Louisiana-Monroe	72.88
100	Maryland	72.66
101	New Mexico St	72.43
102	Hawai`i	72.39
103	Eastern Michigan	72.32
104	UNLV	72.24
105	Colorado	70.81
106	Colorado St	69.23
107	New Mexico	68.35
108	Army	66.5
109	Buffalo	66.49
110	Indiana	66.15
111	Troy	65.5
112	Syracuse	64.92
113	Idaho	64.49
114	Middle Tennessee St	61.26
115	Central Michigan	60.35
116	Akron	57.79
117	Florida Atlantic	52.85
118	Tulane	49.7
119	Alabama-Birmingham	49.34

Final 2011/2012 JnJ Rankings/Ratings

2012-01-10T18:51:00.000-08:00

Final Season JnJ Rankings/Ratings

These numbers represent the final season convergence for the JnJ Rankings. These values give an 87.0% retrodiction rate for the entire season, and a 94.2% rate (33 of 35) for the bowl games (missing only the ULL and Michigan State wins, MSU by only one ranking position). The average rating below is 245.

JnJ Rank	Wolfe Order	Wolfe/JnJ Name	JnJ Rating	Massey Order	Massey Name
1	3	Alabama	466.9	3	Alabama
2	51	LSU	458.4	47	LSU
3	78	Oklahoma St	400.5	74	Oklahoma St
4	7	Arkansas	385.9	6	Arkansas
5	116	West Virginia	377.0	118	West Virginia
6	92	Southern Miss	375.1	87	Southern Miss
7	79	Oregon	373.0	75	Oregon
8	93	Stanford	361.1	88	Stanford
9	91	Southern Cal	358.1	106	USC
10	77	Oklahoma	352.7	73	Oklahoma
11	13	Boise St	344.7	12	Boise St
12	12	Baylor	332.7	11	Baylor
13	95	TCU	332.4	90	TCU
14	44	Kansas St	332.3	42	Kansas St
15	89	South Carolina	332.0	85	South Carolina
16	37	Houston	325.8	35	Houston
17	22	Clemson	321.0	20	Clemson
18	119	Wisconsin	314.4	119	Wisconsin
19	63	Missouri	314.4	59	Missouri
20	34	Georgia	313.3	32	Georgia
21	58	Michigan St	309.4	54	Michigan St
22	99	Texas A&M	307.2	94	Texas A&M
23	10	Auburn	304.4	9	Auburn
24	32	Florida St	303.7	30	Florida St
25	57	Michigan	301.0	53	Michigan
26	65	Nebraska	295.1	63	Nebraska
27	98	Texas	292.0	93	Texas
28	74	Notre Dame	287.4	70	Notre Dame
29	81	Penn State	283.9	77	Penn St
30	112	Virginia Tech	280.6	112	Virginia Tech
31	29	Florida	277.7	28	Florida
32	62	Mississippi St	277.6	58	Mississippi St
33	114	Washington	270.5	116	Washington
34	72	Northern Illinois	270.1	60	N Illinois
35	16	Brigham Young	266.3	16	BYU
36	104	Tulsa	266.3	99	Tulsa
37	41	Iowa	265.4	39	Iowa
38	85	Rutgers	263.6	81	Rutgers
39	55	Miami FL	260.8	51	Miami FL
40	97	Tennessee	255.2	92	Tennessee
41	47	Louisiana Tech	253.6	45	Louisiana Tech
42	21	Cincinnati	252.1	19	Cincinnati
43	107	Utah	252.1	107	Utah
44	42	Iowa St	251.9	40	Iowa St
45	83	Purdue	250.6	79	Purdue
46	18	California	248.5	18	California
47	69	North Carolina	247.8	67	North Carolina
48	75	Ohio State	247.1	72	Ohio St
49	35	Georgia Tech	247.0	33	Georgia Tech
50	73	Northwestern	246.8	69	Northwestern
51	111	Virginia	246.1	111	Virginia
52	88	SMU	245.6	84	SMU
53	110	Vanderbilt	244.6	110	Vanderbilt
54	39	Illinois	242.3	37	Illinois
55	76	Ohio U.	240.4	71	Ohio
56	70	North Carolina St	236.0	62	NC State
57	82	Pittsburgh	235.0	78	Pittsburgh
58	105	UCLA	234.6	102	UCLA
59	46	Kentucky	234.2	44	Kentucky
60	8	Arkansas St	233.9	7	Arkansas St
61	100	Texas Tech	232.7	95	Texas Tech
62	52	Marshall	232.3	48	Marshall
63	66	Nevada	231.1	64	Nevada
64	64	Navy	230.9	61	Navy
65	101	Toledo	230.5	96	Toledo
66	108	Utah St	230.5	108	Utah St
67	113	Wake Forest	229.4	115	Wake Forest
68	5	Arizona	226.2	4	Arizona
69	6	Arizona St	225.7	5	Arizona St
70	118	Western Michigan	225.4	114	W Michigan
71	50	Louisville	225.4	46	Louisville
72	80	Oregon St	223.7	76	Oregon St
73	19	Central Florida	223.5	101	UCF
74	84	Rice	223.3	80	Rice
75	60	Minnesota	222.8	56	Minnesota
76	86	San Diego St	222.2	82	San Diego St
77	87	San José St	220.3	83	San Jose St
78	26	Duke	220.1	24	Duke
79	14	Boston College	218.7	13	Boston College
80	25	Connecticut	217.6	23	Connecticut
81	117	Western Kentucky	217.5	113	W Kentucky
82	96	Temple	217.1	91	Temple
83	27	East Carolina	217.0	26	East Carolina
84	61	Mississippi	216.9	57	Mississippi
85	90	South Florida	215.7	86	South Florida
86	48	Louisiana-Lafayette	214.8	103	ULL
87	120	Wyoming	214.1	120	Wyoming
88	115	Washington St	214.0	117	Washington St
89	31	Florida Int'l	213.6	29	Florida Intl
90	36	Hawai`i	213.6	34	Hawaii
91	109	UTEP	211.9	109	UTEP
92	15	Bowling Green	209.5	14	Bowling Green
93	33	Fresno St	207.8	31	Fresno St
94	23	Colorado	206.1	21	Colorado
95	1	Air Force	205.8	1	Air Force
96	94	Syracuse	204.8	89	Syracuse
97	68	New Mexico St	203.6	66	New Mexico St
98	43	Kansas	202.3	41	Kansas
99	45	Kent St	197.3	43	Kent
100	56	Miami OH	197.0	52	Miami OH
101	11	Ball St	193.0	10	Ball St
102	28	Eastern Michigan	188.8	25	E Michigan
103	38	Idaho	184.7	36	Idaho
104	71	North Texas	183.8	68	North Texas
105	24	Colorado St	179.1	22	Colorado St
106	53	Maryland	175.8	49	Maryland
107	67	New Mexico	173.4	65	New Mexico
108	106	UNLV	173.0	105	UNLV
109	49	Louisiana-Monroe	169.5	104	ULM
110	17	Buffalo	164.3	15	Buffalo
111	9	Army	155.0	8	Army
112	40	Indiana	153.8	38	Indiana
113	20	Central Michigan	145.0	17	C Michigan
114	102	Troy	142.9	97	Troy
115	2	Akron	138.2	2	Akron
116	59	Middle Tennessee St	133.4	55	Middle Tenn St
117	30	Florida Atlantic	108.5	27	FL Atlantic
118	103	Tulane	105.5	98	Tulane
119	54	Memphis	103.8	50	Memphis

Alabama 21, LSU 0

2012-01-10T16:57:00.000-08:00

The BCS Championship and Final Pick Standings

Only 4 of 131 systems picked Alabama over LSU.

Here are the final standings:

18/35 = 2 systems out of 131 (worst, at 51.4%, includes JNJ and Billingsley)

19/35 = 3 systems out of 131

20/35 = 11 systems out of 131

21/35 = 31 systems out of 131

22/35 = 36 systems out of 131 (this was closest to average, 62.9%)

23/35 = 25 systems out of 131

24/35 = 12 systems out of 131

25/35 = 7 systems out of 131

26/35 = 1 systems out of 131

27/35 = 1 systems out of 131

28/35 = 1 systems out of 131

29/35 = 1 systems out of 131 (best, at 82.8%)

Top systems were:

29/35 : Round Robin Win %

28/35 : Regression Based (seems to be a variation of Tempo-Free, both point to same website)

27/35 : Tempo-Free Gridiron

26/35 : Kellner MOV

The BCS ranking systems averaged 21 picks correct, and ranged between 18 and 23.

N Illinois 38, Arkansas State 20

2012-01-10T16:35:00.001-08:00

GoDaddy Bowl

North Illinois takes the GoDaddy Bowl, a result only 31 systems (one of which was JNJ) got correct.

SMU 28, Pittsburgh 6

2012-01-07T15:27:00.000-08:00

Compass Bowl

Only 55 of 131 systems got this one, and JNJ missed. RWP continues to lead, with 28 of 33 correct so far.

Arkansas 29, Kansas State 16 (Final)

2012-01-07T03:25:00.000-08:00

About 2/3rds of Systems Pick the Cotton Bowl Correctly

Alas, the JNJ was not one of the fortunate 83 who did get the game's W/L result correct, and we have now fallen back to the very back of the pack. Hard times, especially hard for us considering that the JNJ hit over 70% for the entire season. Several teams/games seem to have been particularly poorly picked and we will be looking at our minimization statistic in light of the success of some of the other systems to see what adjustments should be made. These days, it's also possible to find historical data for all past seasons, so we will go ahead and run the system against that data. Indications are that some form of score leveling helps, but various systems that use various variations on that method appear to have enjoyed differing benefits, so it will be a little bit of back to the drawing board. There are now only 3 bowls left. If Pittsburg, N. Illinois, and LSU each win their respective bowls, we won't finish last, but that's seems fairly unlikely at the moment.

Arkansas 19, Kansas State 16 (in progress)

2012-01-06T20:04:00.000-08:00

An Interesting Game So Far

This is the game we predicted would be very close, and at the moment we don't seem to be too far off. We continue to watch with great interest. Though it does seem unlikely there will be another blocked kick that gets run back for two points!

Stay tuned!

West Virginia 1000, Clemson 26

2012-01-04T21:09:00.000-08:00

A Forgone Conclusion

The actual score as I write this is West Virginia 70, Clemson 26, and any discussion of a Clemson comeback is moot.

Only 10 systems will have gotten the game's W/L result correct, and I am sure that nobody anywhere would have predicted this big a blowout.

So let's call this the upset of the 2011/2012 bowl season and take a well-deserved nap.

Michigan 23, Virginia Tech 20 (OT)

2012-01-03T21:06:00.000-08:00

Quite an Ending in the Sugar Bowl

In what will no doubt be one of the most talked-about-the-day-after finishes, Virginia Tech had a touchdown called back and then missed a field goal to open the door for Michigan, who wasted no time getting 3 points up in the air to close out the game.

107 of 131 systems got this pick correct. RWP continues to lead with their impressive 2 game lead, at 27 of 30 bowls. JNJ got this one correct, but remains tied at the bottom with two other systems.

Oklahoma State 41, Stanford 38 (OT)

2012-01-02T21:35:00.000-08:00

Another Offense-loaded Bowl Game in the Fiesta

Oklahoma State only lead once, but it was all they needed as they snatched victory away from Stanford. It will probably be difficult for them to argue for any claim to the number one position, given that it took two missed FG attempts in the last minutes of the game by Stanford to set up their victory, but it was still a pretty exciting game to watch.

Rose Bowl: Oregon 45, Wisconsin 38

2012-01-02T18:12:00.000-08:00

109 of 131 systems got this one

15/28 = 1 systems

16/28 = 4 systems

17/28 = 8 systems

18/28 = 27 systems

19/28 = 49 systems

20/28 = 28 systems

21/28 = 10 systems

22/28 = 2 systems

23/28 = 1 systems

24/28 = 0 systems

25/28 = 1 systems

Let's take a look at those results in a graph:

Four More Bowls

2012-01-02T15:35:00.000-08:00

Ticket City, Outback, Capital One, and Gator

122 of 131 systems correctly picked Houston (31) over Penn State (14) in the Ticket City Bowl.

Just 43 of 131 systems picked Michigan St (33) over Georgia (30) in the Outback Bowl.

Almost everyone (128 of 131) picked South Carolina (30) over Nebraska (13) in the Capital One Bowl.

84 of 131 systems correctly picked Florida (24) over Ohio State (17) in the Gator Bowl.

JNJ got only the South Carolina over Nebraska result correct of these four bowls.

14/27 = 1 systems (JNJ falls back to the bottom)

15/27 = 1 systems

16/27 = 8 systems

17/27 = 30 systems

18/27 = 41 systems

19/27 = 32 systems

20/27 = 11 systems

21/27 = 4 systems

22/27 = 2 systems

23/27 = 0 systems

24/27 = 1 systems (RWP continues to lead the field)

Auburn 43, Virginia 24

2011-12-31T22:07:00.000-08:00

Auburn Takes the Chik-Fil-A

Which 113/131 systems got correctly.

Round Robin Win Percentage maintains a 2 game lead on the best of the rest, and most of the BCS rankings are down in the 14/15 correct picks zone. JNJ continues to float near the bottom.

12/23 = 1 systems

13/23 = 6 systems (including JNJ)

14/23 = 21 systems

15/23 = 53 systems

16/23 = 32 systems

17/23 = 11 systems

18/23 = 5 systems

19/23 = 1 systems ("Regression Based")

20/23 = 0 systems

21/23 = 1 systems ("Round Robin Win Percentage" stands alone)

As always, thanks must go to Ken Massey for maintaining the CFB Ranking Comparison List.

BCS Systems:

15/23 Anderson

13/23 Billingsley

15/23 Colley

15/23 Massey BCS

15/23 Sagarin-Elo

14/23 Wolfe

Mandel: 14/23

Illinois 20, UCLA 14

2011-12-31T18:00:00.000-08:00

Kraft Fight Hunger Bowl

Only 50 of 131 systems got this result correct, and at last JNJ was on the good side of the split, but this far into the bowls, it's clear that although the JNJ system did fine at picking regular season predictions that it just isn't cutting the mustard for post-season play, at least this year.

Here are the current cumulative standings:

12/22 = 5 systems (including JNJ)

13/22 = 21 systems

14/22 = 52 systems

15/22 = 33 systems

16/22 = 7 systems

17/22 = 9 systems

18/22 = 3 systems

19/22 = 0 systems

20/22 = 1 systems

Round Robin Win % now not only leads the standings, but is a full 2 games up on the next level of systems, which include Congrove, D1A Football, and Regression Based with 18 picks each. Rothman has faded back into the pack with 15, a solid, but not remarkable standing at this point.

All lists can be found via links at Massey's Ratings Comparison page.

Cincinnati 31, Vanderbilt 24

2011-12-31T17:48:00.000-08:00

Liberty Bowl

101 out of 131 systems got this result correct, but JNJ was not one of them.

Utah 30, Georgia Tech 27

2011-12-31T17:45:00.000-08:00

Sun Bowl Goes to OT

Only 41 of 131 systems got this result correct. JNJ was not one of them.

Texas A&M 33, Northwestern 22

2011-12-31T17:42:00.000-08:00

Meineke Car Care Bowl

All 131 systems got this result correct.

Oklahoma 31, Iowa 14

2011-12-31T03:30:00.000-08:00

Halfway Home

We are now more than halfway through bowl season, with 18 of 35 games played. Everybody got the last two games right, so there has been no change in which system stands where. Here are the current standings:

9/18 = 1 system

10/18 = 9 systems

11/18 = 29 systems

12/18 = 62 systems

13/18 = 16 systems

14/18 = 9 systems

15/18 = 4 systems

16/18 = 1 system

14 systems are performing at 77.5% or better.

Mississippi State 23, Wake Forest 17

2011-12-30T20:22:00.000-08:00

Sweet Music for One and All at the Music City Bowl

Every system picked Mississippi State, so 131 contented sighs went up over that result. Well, surely there were some folks rooting for Wake Forest, but apparently the stats were solidly against them. At the moment Oklahoma leads Iowa, and if that holds up everyone will get another win.

Rutgers 27, Iowa State 13

2011-12-30T16:09:00.000-08:00

But Then We Pretty Much Give it Back

Only about 1/3 of the systems pegged the Pinstripe Bowl correctly, and JNJ wasn't one of them. So we mostly lost that game back to the field. There =is= one system that seems to be below us at the moment, but we like to underline the good performances, so if you want to know who it is, you'll have to crank some numbers yourself.

Meanwhile Round Robin Win Percentage got the Pinstripe correctly, and Pigskin didn't, so one streak ends and another goes on. RRWP is now up to 12 in a row and leads the entire pack in solo mode:

6/16 = 0 systems

7/16 = 1 systems

8/16 = 9 systems

9/16 = 29 systems

10/16 = 62 systems

11/16 = 16 systems

12/16 = 9 systems

13/16 = 4 systems

14/16 = 1 systems

15/16 = 0 systems

For those who care to revisit some of our pre-bowl-season humor, the JNJ is currently the owner of a 1 Flipping Monkey Scale rating. And we ain't proud of it! Congratulations to the 14 systems that are currently hitting at 75% or better!

Two more bowls tonight, but we won't make up any ground there: every single system picks both of them the same way.

Brigham Young 24, Tulsa 21

2011-12-30T15:53:00.000-08:00

JNJ Claws Back Off the Bottom

Only 25 systems picked BYU over Tulsa, and JNJ happened to be one of them, so at least for a little while we have clawed our way off the bottom. Like a lot of folks, we were working when this game happened, so we will have to read about it to know more details.

6/15 = 0 systems

7/15 = 3 systems

8/15 = 15 systems (including JNJ)

9/15 = 50 systems

10/15 = 43 systems

11/15 = 14 systems

12/15 = 4 systems

13/15 = 2 systems

14/15 = 0 systems

Pigskin and Round Robin Win Percentage are at the top, and Pigskin has the longest win active win streak with 12 in a row.

Baylor 67, Washington 56

2011-12-29T22:27:00.000-08:00

Someday We May Call It the No Defense Bowl

Make of it what you will, the Alamo will likely be remembered this year. All sorts of records were set in this crazy game, with over 100 points scored total.

7/14 = 5 systems

8/14 = 15 systems

9/14 = 57 systems

10/14 = 41 systems

11/14 = 10 systems

12/14 = 3 systems

Top spot amongst the ranking systems is currently shared by Pigskin, Rothman, and Round Robin Win Percentage with Pigskin perfect over the last 11 bowls. Kislanko has the last 10 correct.

Thirteen Bowls In

2011-12-29T20:07:00.000-08:00

With A Special Appearance By Mr. Mandel

Standings after the Champs Bowl (Florida State won that one):

5/13 = 0 systems

6/13 = 5 systems (including JNJ)

7/13 = 15 systems

8/13 = 56 systems (57 including Stewart Mandel, SI.com)

9/13 = 42 systems

10/13 = 10 systems

11/13 = 3 systems

12/13 = 0 systems

13/13 = 0 systems

Mandel was kind enough to project actual game scores for every bowl, and includes a brief paragraph for each bowl, but his list uses icons for the predictions, not text values, so we didn't fold his picks into the Massey comparison list. But we will check in with his standings now and then.

Meanwhile, Washington and Baylor are setting bowl scoring records over at the Alamo Bowl, with 59 points in the first half of the game.

The system with the longest winning streak is now Pigskin, with 10 in a row, tied at the top of the list with 11 total correct picks, but that may not last, since virtually every system has Baylor picked to win the Alamo, and Washington is looking to take advantage of the "Heisman curse", starting the second half with an 11 point lead. Whoops! Make that an 18 point lead just one minute into the second half!

Florida State 18, Notre Dame 14

2011-12-29T18:24:00.000-08:00

A Somewhat Unanticipated Result Slows the Rothman Rampage

There were actually several systems that had 9 correct picks in a row, and a few systems extended their streak (more on that after tonight's second bowl), but most systems got the FS/ND result incorrect. Let's take a look at the results so far, sorted by the number of systems that got it right (the basis is 131 systems -- Massey's official list minus the systems that don't rank all 120 FBS teams, plus a few of his combined/derived rankings; so, for example, the biggest upset so far appears to be ULL's win over SDSU, which only 5 systems pegged correctly):

5	San Diego St	30	ULL	32
22	NC State	31	Louisville	24
29	Florida St	18	Notre Dame	14
47	Florida Intl	10	Marshall	20
54	Purdue	37	W Michigan	32
72	Temple	37	Wyoming	15
102	Ohio	24	Utah St	23
122	Texas	21	California	10
123	Air Force	41	Toledo	42
126	North Carolina	24	Missouri	41
130	S'thern Miss	24	Nevada	17
131	TCU	31	Louisiana Tech	24
131	Boise St	56	Arizona St	24

Texas 21, Cal 10

2011-12-29T01:17:00.000-08:00

Texas Defense Roughs Up Cal QB

With five takeaways and six sacks, the Texas defense didn't seem to have much trouble causing Cal and Zach Maynard a tough time. Both of those numbers were apparently season highs.

That leaves us with about 1/3 of all the bowls now played, and the pack spreads out as follows:

5/12 = 0 systems

6/12 = 5 systems (including JNJ)

7/12 = 24 systems

8/12 = 58 systems

9/12 = 38 systems

10/12 = 5 systems

11/12 = 1 systems

12/12 = 0 systems

Toledo 42, Air Force 41

2011-12-28T18:43:00.000-08:00

About As Close As It Gets

Okay, it didn't come down to the very last play, but when Toledo punched Air Force's late game 2 point conversion attempt with less than a minute to go, you gotta admit that's pretty close. This was a game our official entry missed (dropping us back to the bottom of the pile), but at least it was a game where we predicted that it could be too close to call. Plus, since both J's are from northwest Ohio, it was tough to be rooting against Toledo.

In other news, check out this interesting article we found on a Buckeye fan page while doing more research on the Rothman system. It lists quite a few rating systems with long pedigrees. And check out this link for Billingsley's All Time All Team list. Does =your= alma mater make the grade?

Meanwhile, Rothman got the Toledo/Air Force game correct (as did most other sytems other than JNJ) and is still alone at the top.

Oh, by the way, how are the BCS systems doing?

2011-12-27T21:05:00.000-08:00

Well, They have Nothing to Complain About, it Would Seem

Anderson, Colley, Massey BCS, and Sagarin-ELO all currently stand at 7/10, which is quite respectable, and Billingsley and Wolfe appear to be at 6/10, which is still fine.

NC State 31, Louisville 24 // Purdue 37, W Michigan 32

2011-12-27T20:44:00.000-08:00

But the Classic Rothman Rating Leads the Pack -- Alone!

The Purdue victory finally gave us a game back against the pack, and for a few moments it looked like Louisville might keep a late drive alive to possibly tie the game and give us a chance to gain a little more ground, but instead they lost yardage and the game slipped forever from their grasp. With that result, we find that

3/10 = 0 systems

4/10 = 2 systems

5/10 = 17 systems (including JNJ)

6/10 = 64 systems

7/10 = 42 systems

8/10 = 5 systems

9/10 = 1 system
10/10 = zero systems

So it appears that just 1 system has 9 wins and leads the pack alone. Interestingly enough, that system is the classic Rothman (FACT) system, an open source method used as one of the computer ranking systems in the original BCS rankings, and dating back to about 1970. That link is definitely worth a read.

Missouri 41, North Carolina 24

2011-12-26T20:26:00.000-08:00

Holding Our Own

With the Missouri win, the JNJ gained no ground, but a small number of other systems lost a bit. The standings now appear to be:

3/8 = 0 systems

4/8 = 6 systems (including JNJ)

5/8 = 36 systems

6/8 = 83 systems

7/8 = 6 systems

8/8 = 0 systems

Two bowls tomorrow.

Seven Down, Twenty Eight to Go

2011-12-26T10:12:00.000-08:00

At Long Last, JNJ Gains(?) Ground

Let us not be proud. We are still at the bottom of the pack, with just 3 bowls correctly picked out of the 7 played so far. But after several days of being one of =three= on the bottom, we are now one of =four= at the bottom. So, truly, we gained no ground, but rather, one poor soul lost ground.

2/7 = 0 systems

3/7 = 4 systems

4/7 = 35 systems

5/7 = 86 systems

6/7 = 6 systems

7/7 = 0 systems

The next game where we differ from a fair number of other systems is Purdue/W Michigan (Purdue). And we differ from almost everyone with our Air Force/Toledo (Air Force) projection.

One Flippin' Monkey

2011-12-20T19:09:00.000-08:00

But He Does Pretty Good For Himself

[warning: click the links in this particular post at your own risk, some of the monkeys escaped from the research wing and injected some significant randomosity into the link-u-lator today]

I'm not saying "The Other J" owns a private island, but although his side of the organization is raising A Million Flipping Monkeys (who no doubt need certain special handling), "This J" can't currently afford quite that many primates. It is, after all, bowl season, um...er....I mean non-Grinch season, and with the economy needing a boost, I have (over)committed to "investments" in things like the Kindle Smolder and classic holiday gifts like these gems (or these) which I saw recently in my local house of merchandise.

But as part of a comprehensive review of the current JnJ ranking system, last night I rechecked the entire season (we eventually want to publish our Flipping Monkey Scale rating, of course, because we think we will be proud of it). Our full-season W/L results were pretty good, but in the process of doing that I realized that although a rather large gaggle (pride? bevy?) of monkeys could match our results, that there might be a way for just one (but properly trained [video]) monkey to do a pretty decent job.

My proposal is this: just train this monkey to always pick the home team. We all know that there is a built-in home team advantage in college football, but just how strong is it? Well, this year, if our properly trained monkey carried out his assignment from the beginning of the season to the end, he would hit an astounding 59.8% against the entire (pure FBS) slate! Yesterday we concluded that it would take something like 200 monkeys to hit 60%, but that one well-trained primate might be able to manage the same feat. I say, "Hoorah!" and "Let's give that monkey a great reward!".

But then I realize...that means that for a rating system to be truly great, it has to hit 5% or maybe even 10% higher than you'd think at first glance. I'm a little curious how the JnJ would do considering only something akin to the top 25 teams each week, or how a slightly more well-trained monkey might do picking the team with the better record, but that will take a bit more work. Which we will do. So let it be written, so let it be done!

A Million Flipping Monkeys...

2011-12-19T17:06:00.000-08:00

Some Deep Thoughts from "The Other J"

The Efficient Market Hypothesis states, in its crudest form, that a monkey throwing darts at a copy of the Wall Street Journal has the same chance at picking stock winners as you do. No, it really says that! The point is that the Efficient Market Theory gives us a reference for measuring the success or failure of any stock picking system. That is, is your system better or worse at picking stocks than a dart throwing monkey? An analogous metric for football rating systems would be is your system better at picking winning games than a coin flipping monkey. If your system does not beat this metric, you should probably just stop. Your system isn’t worth the TP it was written on.

One coin flipping monkey gives us a good start, in that it gives a hard pass/fail line. However, it does not help us differentiate between the myriad of football rating systems from the idiotic to the sublime. I propose a performance rating system based on the number of coin flipping monkeys it would take to duplicate the football rating system. Just as engine power is based on horsepower, the performance of football rating system would be based on monkey-power. That is, how many monkeys does it take before at least one of those monkeys will do as well or better than your football rating system?

During the college football season there are about 60 FBS games a week. To pick 30 of these games correctly is no great feat for even one flipping monkey. To pick 35 games correctly on average you would need only 10 flipping monkeys. To pick 40 games you would need 200 flipping monkeys. Forty out of sixty or two-thirds, frankly, isn’t all that impressive. Some systems consistently pick more than 70% of the football games correct and you would need about a 1000 flipping monkeys to do that. To duplicate a system that consistently picked 75% of the games correctly, you would need 17,500 monkeys! That’s a lot of flipping monkeys. As a matter of fact, the biggest problem with the Flipping Monkeys Scale is that over 75%, you practically need a supercomputer to determine the number of monkeys needed.

For those of you who thought this was going to be a serious discussion, the better way of thinking of this is that if you are correct only 50% of the time, then one monkey flipping a coin will beat your predictions about half the time. If you can live with that, God bless you. If your system is right 35 times out of 60, the coin flipping monkey would beat you only one week in 10. Before you start patting yourself on the back though, that means about once a year the monkey would have bragging rights for a week. For me, consistently predicting about two-thirds of the games correctly is where a system starts to show its merit. Then the monkeys would only best the system about once every 200 weeks. And, even a flipping monkey knows that being right once every 17 years is nothing to brag about.

In support of the Flipping Monkey Scale, the JNJ research department has a warehouse full of monkeys being trained to flip coins.* If your system is consistently picking less than half the games correctly, you might want to subscribe to our Flipping Monkeys Newsletter -- where each week we will highlight the picks of one of our monkeys.

Sincerely yours,

The Other J

* -- In the interest of full disclosure, you can’t train monkeys to flip coins. They are actually chimpanzees. Saying “flipping monkeys” is just funnier than saying “flipping chimpanzees”. On a side note, I’ve noticed that chimpanzees don't seem to be the least bit offended when you call them monkeys. They are good that way.

[Due to prior commitments all blog postings prior to this one have been made by "This J". But this post is by "The Other J"]

A Big Bowl in New Orleans

2011-12-17T22:15:00.000-08:00

But a Humbling Start for the JNJ

If we've done our number crunching correctly, the JNJ is just one of 11 systems that started the bowl season 0 for 3. 63 systems are 1 for 3, 55 are 2 for 3, and just 2 have a perfect 3-0 start. (Howell and Marsee)

Temple had no problems with Wyoming, Ohio slipped by Utah, and the University of Lousiana Lafayette won on a last-play field goal that might not have made it had San Diego State showed 5 yards of additional discipline in the second to the last play. But exciting games are always worthwhile, and with an 0 and 3 start on the line, this J was glued to the set on a game that otherwise might have gone by without much note.

72 of 131 systems got the Temple/Wyoming result correct.

102 of 131 systems got the Ohio/Utah State result, and

5 of 131 systems got the ULL result correct.

The Mandelbowl Set

2011-12-17T03:13:00.000-08:00

With Apologies to Benoit Mandelbrot and His Groovy Graphs

Half the fun of waiting for the bowls to actually begin is to find stuff that seems worth commenting upon and then to find a way to show the numbers. In one previous article we mentioned that we had a couple of outlier rankings that were quite a big different from the consensus. And in another we name-dropped a professional columnist, Stewart Mandel of CNN, who has updated his columns with a full list of bowl score prognostications. This J enjoys Mandel's writing, and we are always favorable to folks who are willing to take a stand with real numbers.

In this post, we take a look at a few bowls in more detail, and we include the outliers that we mentioned earlier and a couple where the official JNJ prediction is different from Mr. Mandel's.

First up, the BCS bowl. Most folks who are posting rankings pick LSU, and so does the JNJ. Not much to discuss there, but we include it to show what a close game with consistent rankings looks like.

Next up we have the Fiesta Bowl. Some systems pick Stanford, but most pick OSU, as do both Mandel and the JNJ. There is room in the numbers we see for either outcome, but the JNJ rates Stanford significantly lower than most of the entries in the Massey super list. As a matter of curiousity, we worked up their average performance ratings, and it turned out that Stanford was 10 or 15% below the line of greatness, while OSU was 10 or 15% above the line of greatness. Nothing is guaranteed, but our numbers suggest that OSU won't have any trouble handling Stanford. Mandel seems to agree. Standford is one of our two important outliers: JNJ rates it significantly lower than most. In this case, that doesn't seem to threaten the predicted outcome, but perhaps it will be a reflection on the quality of the game.

Next up we have the Cotton Bowl. Here, the majority of systems rank Arkansas above Kansas State, but it's a narrow margin. Our official Massey submission (everybody eventually has to pick their numbers and stick with 'em) reverses that, but by the same narrowest of margins. Various safety calculations seem to suggest that Arkansas will come out on top, but given the way the numbers fall across the board, this seems to be =the= game to watch -- nobody knows how it will turn out. We are hoping that means it's a close, hard-fought game. Mandel's pick is Arkansas by a sizeable margin (38 to 24 predicted score), so here he differs significantly from the consensus and JNJ.

Fourth, we list the Sugar Bowl, where most of the systems have Michigan as a solid favorite over Virginia Tech. JNJ parallels that stand, but rates both teams much lower than the consensus. However, the difference in rankings is about the same as the consensus.

Finally, we list the Ticket City Bowl. Here, the consensus has Houston as a clear favorite over Penn State. Both JNJ and Mandel see it differently, however. Mandel projects Penn State as the winner, 27-21, while JNJ suggests that the final point ratio should lean toward Penn by an even higher percentage.

In fact, Houston is the single biggest outlier from consensus in the JNJ this year. Curiosity about this and Stanford's ranking led to the Line of Greatness. As mentioned above, we found that Stanford fell somewhat below the Line of Greatness this year...but Houston, despite their 12-1 record and 16th place in the consensus, finds itself at 43rd in the JNJ. Their Line of Greatness rating was only about 45% due to both a weak schedule and a low performance against it. When January 2nd is over, we will have more evidence as to whether they were playing only as hard as they needed to win, or if their strength of schedule was truly as weak as we see it. Considering their average margin of victory was almost 30 points, and that the JNJ says that was =underperforming=, we are forced to conclude their strength of schedule was unusually low. It's also possible, however, that their games are "isolated" from the pack. We will do some safety calculations to see what we find.

The Top Sixteen

2011-12-15T20:30:00.000-08:00

Some Observations from the Massey CFB Comparison List

In preparing an entry on those two teams which JNJ rates low compared to almost everybody, we took a closer look at the current Massey CFB Comparison page. With over 130 ratings now being listed, it will probably come as no surprise that there is some odd variability on the highest and lowest rankings each team has. The JNJ has a few such outliers, but only one seems truly strange. But a good handful of other lists have their own, different, outliers.

Massey shows the standard deviation for the consensus ranking, but it seemed easier to try to show this in a graph.

With 130 to pick from, we decided it should be pretty safe to throw away the top 5 and bottom 5 for each team and do some plots. Here, for example, is a view of the top 16, showing the high, low, and median, as well as the bottom quartile and top quartile.

As you can see, Kansas State and Houston show some pretty wide variation across the various rankings. Click on the graph to see a somewhat larger version in a separate window or tab.

Who are the Great Teams in CFB this Year?

2011-12-15T13:49:00.000-08:00

The Line of Greatness

Imagine a nightmare scenario where the teams you play through the course of the year average out to be Top 25 teams. Then compound the nightmare by having high scoring games with every one of them, and in each game you lose by one measly point. In the JNJ system such a scenario would earn you a pretty high ranking. That's because you had an incredibly difficult schedule and you performed fairly well against it. Yeah, you didn't win a single game, but you =almost= beat 12 really tough teams. If you were matched against some normal non-super-human team, you could probably crush them.

This scenario could never happen in real life, but it gives us what we have decided to call, at least for now, the Line of Greatness. Perform =better= than that, and you are doing very well indeed, because nobody has that tough a schedule, but many teams can perform well against the teams they do play. Maybe a Line of Greatness based on the Top 25 average is too high. How about the Top 50 average? Nah, we want Greatness, not Pretty-Darn-Goodness.

So, how many teams performed above the Line of Greatness in 2011? Our numbers say only three: LSU, Oklahoma State, and Alabama.

There were a couple teams this year that most other systems rank a lot higher than JNJ does. We will talk about them next time, and we'll be referring to the Line of Greatness in that post.

Beating Up On Angry Birds

2011-12-14T02:28:00.000-08:00

Or ... Whaddya Mean I'm Not Using My Time Productively?

Check out Fareed Zakaria's CNN article talking about wasting time playing Angry Birds. According to his statistics, the world has wasted a total of 200,000 years so far (let's call those "entity years" so as to encompass humans of all genders, backgrounds, ages, and whatever other demographic you can think of, as well as any non-human creatures who might be trying to pass themselves off as human while waiting for their friends and relatives to come rescue them from the Don't Panic Planet here in the galactic backwaters. At the end of the article he muses "Just imagine what we could have accomplished if that time was spent productively."

But 200,000 years wasted is =nothing=. Why, the JnJ ratings alone have already wasted at least an entity-year of somebody's time, and there are hundreds of CFB rating systems that are published, which is just a tip of the iceberg. It's probable that CFB rating systems, in being created, edited, discussed, and reviled, have wasted a few thousand or even tens of thousands of years.

But that is a mere drop in the bucket compared to CFB itself. Hundreds of teams with hundreds of players, each with parents who tended their youth-entities through hundreds of games just in preparation and training for those games we rate and outcomes we await. And all of that going on for over a hundred years, at ever increasing levels of participation.

We love math, but the mind begins to boggle, truly boggle at the amount of time, the number of years, the intensity of spirit literally poured into CFB since it's inception. We conservatively estimate the CFB tally at something on the order of 150,000,000 years, and that's not even including time spent at lunch arguing the merits and limitations of the BCS selection system!

Take =that= Angry Birds!

The Best Worst Bowl of 2011

2011-12-13T19:47:00.000-08:00

It was Two Against the World -- and the Odds were Even!!!

First of all, let's remember that there are no bad bowls, so there can't possibly be a worst bowl.

But Stewart Mandel over at SI.com has a preview of all the upcoming bowl games, in which he has a little fun while ranking the Kraft Fight Hunger Bowl at number 33. He says "Illinois (6-6) vs. UCLA (6-7). Two teams with interim coaches meet in a baseball stadium on New Year's Eve. One squad just lost six in a row, the other just lost 50-0 to its arch-rival. Is it 2012 yet?".

But, including safety calculations, JnJ ranks that bowl as even closer than the Cotton! Both of these teams have something to prove and we are looking for a good old fashioned knock-down drag-out of a fight. Will that happen? Probably not -- all too many close game projections turn out to be blowouts ... but that won't stop us from hoping, and they don't call 'em the Fighting Illini for nothing!

So who do we like? Go Bruins! (It's not that we don't like Illinois, in fact our official pick IS Illinois, but "Go Fighting Illini" seems a lot less musical a phrase off the tip of the tongue, LOL!).

What Does it Take to be Number One?

2011-12-12T17:45:00.001-08:00

How the Search for the Higgs Boson relates to CFB Bowl Projections

So how well should a ranking system do when it comes to predicting the next round of games? There are mathematical methods to estimate that, but we don't want to bore you with mathematics. Suffice it to say that nobody thinks that picking the W/L result is a 50/50 shot. If all you did was pick the team with the better record, by the end of the season you'd do noticeably better than 50% on the W/L. So let's assume that it's maybe a 65/35 shot to pick the winner. If that's the case, then you'd have to pick something like 29 games to hit near the 3-sigma level. That's 83%!

Tomorrow, the physicists at CERN are expected to give a significant press event. They are probably NOT going to claim 3-sigma results in their search for the fabled Higgs Boson. But nobody would claim anything at less than the 2-sigma level. Let's call it 60/40 and figure the 2-sigma level. That would still be something like 26 games, which is 74%. It looks like the Massey comparison page could contain as many as 150 lists this year (it's up above 130 now). Using some very quick-and-dirty behind-the-scenes math, we predict that only about 8 or 10 lists will get as many as 25 games right this year. If anyone gets as many as 29, that would be, as the physicists say, a truly interesting result.

The Big Bowl

2011-12-11T22:22:00.001-08:00

JnJ Bowl Focus: The Cotton Bowl

They say everything is bigger in Texas, and according to this year's JnJ Rankings, the Cotton Bowl could shape up to be the closest bowl game in many years. The season ending JnJ rankings show Arkansas and Kansas State much closer than 1%, and a variety of safety calculations suggest that there's not a lot of variation from that slender margin.

Here are the big bowls of the year, with a variety of safety calculations shown. In safety calculations, the seed values are varied slightly to determine if any of the calculated results are strongly dependent on the initial conditions, a common problem in convergence problems with limited cross dependency. With only about a dozen games per team per season, there are always several sets of games that can legimately converge to higher =or= lower values without strongly affecting a lot of other games.

Let's take a closer look at the Cotton Bowl:

As you can see, the game is projected to be close under any circumstances, and in 6 of 7 convergences, Arkansas is favored. The official JnJ submission to the Massey comparison showed Kansas State ranked higher than Arkansas, but we also show that game as an "adjustment" game, meaning that the positions of those two teams also varied when we recalculated with bowl games weighted at double the normal values. Not many games fall into that category, but this one does. Consider it in the "too close to call" category, which we are hoping means that it will be an exciting game.

This is as close a game as we can remember showing up in the calculations in several years, and we will definitely be watching to see if it is as close as the calculations seem to indicate.

South Carolina vs Nebraska in the Capital One Bowl also looks like it could be a close game, but the Cotton still looks to be the King.

We'll know for sure come January. That seems like a very long time away!

10 10 5 10

2011-12-11T13:44:00.001-08:00

Looking Ahead

This year, there are 35 bowls. Ten of these are bowls where JnJ varies widely from the other entries in the Massey College Football Ranking Comparison. For ten more, we vary a fair bit, five are pretty close to the rest, and the final ten are almost identical to the rest of the list.

Massey goes to the trouble of showing the correlation to the consensus, and lists his comparison page ranking by that consensus. So we find ourselves in the interesting position of being way over on the right of his page, with the second lowest correlation to consensus on the list!

There =are= other lists which seem to be pretty high in their retrodiction percentage, and although all of them are fairly far from the consensus, none are anywhere near being as far out in right field as we are this year. I guess that's a mixed blessing/curse: if the consensus is accurate, we will do poorly in predicting bowl outcomes. But if we do well, it looks like we are not likely to have much company.

Bowl Projections 12/10/2011

2011-12-10T12:45:00.001-08:00

JnJ CFB Bowl Projections

Baseline projections use all regular season games as input.

The adjustment column (*) shows projections using all regular season games as well, but weighting bowl-bound teams at double the usual rate. A value of -1 indicates that the results using those special ratings showed a different result than the baseline. As expected, this may affect several games with rating ratios close to 1.000

The list is ordered by projected point ratio, so games expected to be close are in the middle of the list.

Bowl Num	Date	Location	On	Bowl Name	Team1	Score1	Team2	Score2	Projected	Ratio	*Adj*
10	Dec. 27	Charlotte, N.C.	ESPN	Belk	North Carolina St	1	Louisville	1	Louisville	0.373	1
17	Dec. 30	Nashville, Tenn.	ESPN	Music City	Wake Forest	1	Mississippi St	1	Mississippi St	0.492	1
1	Dec. 17	Albuquerque, N.M.	ESPN	New Mexico	Temple	1	Wyoming	1	Wyoming	0.554	1
8	Dec. 26	Shreveport, La.	ESPN2	Independence	North Carolina	1	Missouri	1	Missouri	0.621	1
2	Dec. 17	Boise, Idaho	ESPN	Potato	Ohio U.	1	Utah St	1	Utah St	0.638	1
18	Dec. 30	Tempe, Ariz.	ESPN	Insight	Iowa	1	Oklahoma	1	Oklahoma	0.652	1
19	Dec. 31	Houston, Texas	ESPN	Meineke	Northwestern	1	Texas A&M	1	Texas A&M	0.687	1
29	Jan. 2	Glendale, Ariz.	ESPN	Fiesta	Stanford	1	Oklahoma St	1	Oklahoma St	0.688	1
13	Dec. 29	Orlando, Fla.	ESPN	Champs	Florida St	1	Notre Dame	1	Notre Dame	0.834	1
23	Dec. 31	Atlanta, Ga.	ESPN	Chick-fil-A	Virginia	1	Auburn	1	Auburn	0.887	1
30	Jan. 3	New Orleans, La.	ESPN	Sugar	Virginia Tech	1	Michigan	1	Michigan	0.900	1
21	Dec. 31	Memphis, Tenn.	ESPN	Liberty	Cincinnati	1	Vanderbilt	1	Vanderbilt	0.925	1
26	Jan. 2	Orlando, Fla.	ESPN	Capital One	Nebraska	1	South Carolina	1	South Carolina	0.968	1
25	Jan. 2	Tampa, Fla.	ABC	Outback	Michigan St	1	Georgia	1	Georgia	0.990	1
32	Jan. 6	Arlington, Texas	FOX	Cotton	Kansas St	1	Arkansas	1	Kansas St	1.001	-1
27	Jan. 2	Jacksonville, Fla.	ESPN2	Gator	Ohio State	1	Florida	1	Ohio State	1.018	1
22	Dec. 31	San Francisco, Calif.	ESPN	Kraft	Illinois	1	UCLA	1	Illinois	1.026	1
11	Dec. 28	Washington, D.C.	ESPN	Military	Air Force	1	Toledo	1	Air Force	1.092	-1
34	Jan. 8	Mobile, Ala.	ESPN	GoDaddy	Northern Illinois	1	Arkansas St	1	Northern Illinois	1.132	-1
35	Jan. 9	New Orleans, La.	ESPN	BCS Champ	LSU	1	Alabama	1	LSU	1.163	1
28	Jan. 2	Pasadena, Calif.	ESPN	Rose	Oregon	1	Wisconsin	1	Oregon	1.216	1
20	Dec. 31	El Paso, Texas	CBS	Sun	Georgia Tech	1	Utah	1	Georgia Tech	1.218	1
15	Dec. 30	Dallas, Texas	ESPN	Armed Forces	Brigham Young	1	Tulsa	1	Brigham Young	1.281	1
24	Jan. 2	Dallas, Texas	ESPNU	TicketCity	Penn State	1	Houston	1	Penn State	1.328	1
4	Dec. 20	St. Petersburg, Fla.	ESPN	Beef 'O' Brady	Florida Int'l	1	Marshall	1	Florida Int'l	1.348	1
33	Jan. 7	Birmingham, Ala.	ESPN	Compass	Pittsburgh	1	SMU	1	Pittsburgh	1.493	1
9	Dec. 27	Detroit, Mich.	ESPN	Little Caesars	Purdue	1	Western Michigan	1	Purdue	1.528	1
31	Jan. 4	Miami, Fla.	ESPN	Orange	Clemson	1	West Virginia	1	Clemson	1.655	1
16	Dec. 30	Bronx, N.Y.	ESPN	Pinstripe	Iowa St	1	Rutgers	1	Iowa St	1.67	1
12	Dec. 28	San Diego, Calif.	ESPN	Holiday	Texas	1	California	1	Texas	1.675	1
3	Dec. 17	New Orleans, La.	ESPN	Carriers	San Diego St	1	Louisiana-Lafayette	1	San Diego St	1.685	1
7	Dec. 24	Honolulu, Hawaii	ESPN	Hawaii	Southern Miss	1	Nevada	1	Southern Miss	1.759	1
14	Dec. 29	San Antonio, Texas	ESPN	Alamo	Baylor	1	Washington	1	Baylor	1.792	1
5	Dec. 21	San Diego, Calif.	ESPN	Poinsettia	TCU	1	Louisiana Tech	1	TCU	1.919	1
6	Dec. 22	Las Vegas, Nev.	ESPN	Las Vegas	Boise St	1	Arizona St	1	Boise St	2.149	1

All rankings, ratings, and projections presented on the JnJ CFB pages are presented for educational, entertainment, and discussion purposes and are speculative in nature, but based on a consistent methodology. Copyright (c) 2011 JnJ Rankings. Permission is currently granted to publish the JnJ rankings on other CFB ratings comparison pages.

JNJ Rankings 12/05/2011

2011-12-05T18:48:00.001-08:00

JnJ CFB Rankings

The JnJ ("J and J") rankings fit the definition of an "advanced" rating system as found in David Wilson's definition(link can be found on the Massey pages). Results data are inserted into an Excel spreadsheet which is then repeatedlyrecalculated until the ratings stabilize. The minimization statistic is related to the W/L result and the game score, but not inlinear fashion. Stabilization is based on reaching a maximum for the retrodiction set, which may not be infinite depth/pureconvergence but is usually very close to it. Like the Coleman minV system, JnJ can have multiple rankings which achieve the maximumretrodiction rate. The starting values can affect the final ranking, as can the limits placed on the minimum or maximumallowable rating.

The JnJ rating effectively awards higher values for overperforming based on expected results and awards lower values forunderperforming based on expected results. Thus, a powerful team may lose rating points if they beat a weak team, but not byas much as the expected outcome. No manual inputs are made to reflect these adjustments, the model's minimization statistic includesall such adjustment as part of its definition.

Near the beginning of the season, past years results andratings may be used to initialize and stabilize the model. Near the end of the year, only this season's values are used, so thelist below includes only 2011 games. This list was calculated with all teams starting with the same rating/ranking and min/max limitsthat were near (but not at) zero/infinity.

JnJ Rankings as of 12/5/2011

JnJ Rank	Wolfe/JnJ Name	Wolfe Alpha Order	Massey Alpha Order	Massey Name
1	LSU	51	47	LSU
2	Oklahoma St	78	74	Oklahoma St
3	Alabama	3	3	Alabama
4	Oklahoma	77	73	Oklahoma
5	Baylor	12	11	Baylor
6	TCU	95	90	TCU
7	Kansas St	44	42	Kansas St
8	Arkansas	7	6	Arkansas
9	Boise St	13	12	Boise St
10	Oregon	79	75	Oregon
11	Missouri	63	59	Missouri
12	Stanford	93	88	Stanford
13	Southern Cal	91	106	USC
14	Texas A&M	99	94	Texas A&M
15	Texas	98	93	Texas
16	Southern Miss	92	87	Southern Miss
17	Wisconsin	119	119	Wisconsin
18	South Carolina	89	85	South Carolina
19	Notre Dame	74	70	Notre Dame
20	Georgia	34	32	Georgia
21	Michigan St	58	54	Michigan St
22	Nebraska	65	63	Nebraska
23	Clemson	22	20	Clemson
24	Michigan	57	53	Michigan
25	Penn State	81	77	Penn St
26	Iowa St	42	40	Iowa St
27	Auburn	10	9	Auburn
28	Iowa	41	39	Iowa
29	Virginia Tech	112	112	Virginia Tech
30	Florida St	32	30	Florida St
31	Virginia	111	111	Virginia
32	Miami FL	55	51	Miami FL
33	Texas Tech	100	95	Texas Tech
34	Purdue	83	79	Purdue
35	Mississippi St	62	58	Mississippi St
36	Northwestern	73	69	Northwestern
37	Georgia Tech	35	33	Georgia Tech
38	North Carolina	69	67	North Carolina
39	Washington	114	116	Washington
40	Ohio State	75	72	Ohio St
41	Florida	29	28	Florida
42	Minnesota	60	56	Minnesota
43	Houston	37	35	Houston
44	Louisiana Tech	47	45	Louisiana Tech
45	California	18	18	California
46	Illinois	39	37	Illinois
47	Brigham Young	16	16	BYU
48	Utah	107	107	Utah
49	UCLA	105	102	UCLA
50	San Diego St	86	82	San Diego St
51	Kansas	43	41	Kansas
52	Utah St	108	108	Utah St
53	Wyoming	120	120	Wyoming
54	Nevada	66	64	Nevada
55	Tennessee	97	92	Tennessee
56	Vanderbilt	110	110	Vanderbilt
57	Arizona St	6	5	Arizona St
58	West Virginia	116	118	West Virginia
59	Northern Illinois	72	60	N Illinois
60	Kentucky	46	44	Kentucky
61	Oregon St	80	76	Oregon St
62	Louisville	50	46	Louisville
63	Washington St	115	117	Washington St
64	San JosT St	87	83	San Jose St
65	Cincinnati	21	19	Cincinnati
66	Hawai`i	36	34	Hawaii
67	Rutgers	85	81	Rutgers
68	Pittsburgh	82	78	Pittsburgh
69	Arizona	5	4	Arizona
70	Mississippi	61	57	Mississippi
71	Colorado	23	21	Colorado
72	Fresno St	33	31	Fresno St
73	Air Force	1	1	Air Force
74	Navy	64	61	Navy
75	Western Michigan	118	114	W Michigan
76	Arkansas St	8	7	Arkansas St
77	Connecticut	25	23	Connecticut
78	Tulsa	104	99	Tulsa
79	South Florida	90	86	South Florida
80	Toledo	101	96	Toledo
81	Syracuse	94	89	Syracuse
82	Western Kentucky	117	113	W Kentucky
83	Ohio U.	76	71	Ohio
84	Wake Forest	113	115	Wake Forest
85	Bowling Green	15	14	Bowling Green
86	Duke	26	24	Duke
87	Louisiana-Lafayette	48	103	ULL
88	SMU	88	84	SMU
89	Florida Int'l	31	29	Florida Intl
90	Temple	96	91	Temple
91	Miami OH	56	52	Miami OH
92	Central Florida	19	101	UCF
93	Kent St	45	43	Kent
94	North Texas	71	68	North Texas
95	Marshall	52	48	Marshall
96	Boston College	14	13	Boston College
97	Ball St	11	10	Ball St
98	Eastern Michigan	28	25	E Michigan
99	Rice	84	80	Rice
100	North Carolina St	70	62	NC State
101	Louisiana-Monroe	49	104	ULM
102	UTEP	109	109	UTEP
103	Buffalo	17	15	Buffalo
104	Maryland	53	49	Maryland
105	East Carolina	27	26	East Carolina
106	Central Michigan	20	17	C Michigan
107	Army	9	8	Army
108	Troy	102	97	Troy
109	Akron	2	2	Akron
110	Middle Tennessee St	59	55	Middle Tenn St
111	Florida Atlantic	30	27	FL Atlantic
112	Memphis	54	50	Memphis
113	Tulane	103	98	Tulane
114	Alabama-Birmingham	4	100	UAB
115	New Mexico St	68	66	New Mexico St
116	Indiana	40	38	Indiana
117	Idaho	38	36	Idaho
118	Colorado St	24	22	Colorado St
119	New Mexico	67	65	New Mexico
120	UNLV	106	105	UNLV