Compu-Picks 2011 Analysis: Week 11_1

Compu-Picks 2011 Analysis: rating the top and bottom teams in college football after week eleven

As of the end of Tuesday night's games, these are the top 25 (+ Virginia Tech and USM for discussion purposes) and bottom 10 (I thought about doing 20/15, but it's pretty clear that the bulk of the interest in the results comes from the top-rated teams and not the bottom ones). I also Remember that this is a predictive model, designed to pick games and show how good a team actually is. Its results can be very different from what you'll see elsewhere. The workings of the model are confidential (it is, after all, designed to make winning picks), but I'm happy to answer questions about the models' results.

Rank	BCS Rank	Team	League	Score	Schedule Rank *	Result Rank
1	1	Louisiana State	SEC	0.91	6	2
2	3	Alabama	SEC	0.90	14	1
3	5	Oklahoma	Big 12	0.85	7	7
4	2	Oklahoma State	Big 12	0.83	8	5
5	4	Oregon	Pac-12	0.82	5	9
6	9	Stanford	Pac-12	0.73	23	6
7	10	Boise State	Mountain West	0.68	38	8
8	17	Wisconsin	Big Ten	0.58	64	4
9	18*	Southern California	Pac-12	0.49	20	18
10	18	Michigan	Big Ten	0.46	44	10
11	13	Kansas State	Big 12	0.46	2	38
12	14	Georgia	SEC	0.44	32	16
13	6	Arkansas	SEC	0.43	43	12
14	7	Clemson	ACC	0.41	48	13
15	11	Houston	C-USA	0.41	103	3
16	NR	Texas A&M	Big 12	0.41	10	30
17	16	Nebraska	Big Ten	0.41	29	20
18	23	Texas	Big 12	0.40	17	28
19	25	Florida State	ACC	0.40	51	14
20	NR	Notre Dame	Indep	0.39	24	23
21	NR	Missouri	Big 12	0.39	1	54
22	NR	Arizona State	Pac-12	0.37	11	35
23	12	South Carolina	SEC	0.36	39	22
24	15	Michigan State	Big Ten	0.35	36	25
25	19	Texas Christian	Mountain West	0.35	68	15
26	8	Virginia Tech	ACC	0.33	62	17
27	20	Southern Mississippi	C-USA	0.29	84	11
111		Colorado State	Mountain West	-0.52	115	98
112		Central Michigan	MAC	-0.54	108	102
113		Middle Tennessee State	Sun Belt	-0.56	118	95
114		Buffalo	MAC	-0.59	92	114
115		Alabama-Birmingham	C-USA	-0.66	112	113
116		Tulane	C-USA	-0.77	117	116
117		Florida Atlantic	Sun Belt	-0.79	97	119
118		Akron	MAC	-0.84	111	118
119		New Mexico	Mountain West	-0.84	105	120
120		Memphis	C-USA	-0.86	119	117

20 Best Rated Games

Game Rank	Team	Opponent	Location	Score
1	Oregon	Stanford	AWAY	53 - 30
2	Oklahoma	Kansas State	AWAY	58 - 17
3	Louisiana State	Oregon	NEUTRAL	40 - 27
4	Oklahoma	Texas	NEUTRAL	55 - 17
5	Oklahoma State	Missouri	AWAY	45 - 24
6	Oklahoma State	Texas Tech	AWAY	66 - 6
7	Clemson	Virginia Tech	AWAY	23 - 3
8	Louisiana State	Tennessee	AWAY	38 - 7
9	Louisiana State	Alabama	AWAY	9 - 6
10	Wisconsin	Nebraska	HOME	48 - 17
11	Alabama	Florida	AWAY	38 - 10
12	Stanford	Washington	HOME	65 - 21
13	Alabama	Vanderbilt	HOME	34 - 0
14	Oregon	Colorado	AWAY	45 - 2
15	Utah	Brigham Young	AWAY	54 - 10
16	Oklahoma State	Baylor	HOME	59 - 24
17	Alabama	Arkansas	HOME	38 - 14
18	Arizona State	Southern California	HOME	43 - 22
19	Oklahoma	Tulsa	HOME	47 - 14
20	Southern California	California	AWAY	30 - 9

Some thoughts on the list:

1) Please note that AA games are NOT counted for these ratings. This includes the schedule rankings. At some point later this year, I will post an adjusted schedule list that does account for the AA games, but they are not ready at this time. Please keep this in mind when looking at the schedule rankings, since a "true" schedule ranking would note these games.

2) This week, I'm adding in a data field called "Result Rank." This is simply stacking up each team's 1-A game scores, ignoring opponents, HFA, etc. If all we knew was (using Alabama as an example) that the game scores were 48-7, 27-11, 41-0, 38-14, 38-10, 34-0, 52-7, 37-6, 6-9 (OT), and 24-7, it would rate that set of results higher than any other. This is partially because there's a consistent pattern of extreme dominance, and partially because a 3-point overtime loss is treated as pretty close to a tie, so the model looks at is primarily as a close game (which does drop the average) as opposed to taking a much bigger bite out of it because it's a close loss compared to if it has instead been a close win. Again, 3 points in overtime is about as close to a tie as you can possibly get, so it probably shouldn't be stunning that this model takes such an approach.
Of course, LSU is still rated higher, but it's because LSU's schedule was sufficiently tougher than Bama's to make up for the (fairly small) results gap. If the schedule numbers had been flipped, then Bama would have been #1... though if Bama had played as tough a schedule as LSU, they probably wouldn't have been as consistently dominant as they have in fact been.

3) As usual, I'm only posting the compu-picks ratings for the very top and bottom teams (top 25 / bottom 10 this week), and will slowly expand the list as the season goes on. The reason I do this is that the teams at the very top and very bottom have largely separated themselves by now, while the teams on the next tier can largely be jumbled together.

Last week there were 7 elites, now there are 5, with Stanford and Boise both falling well off the pace of the top group. Interestingly, at this point the top five are not only rated much higher than the group just behind htem, they also all have schedule ratings much higher. That's a fairly odd coincidence, but it is interesting to see.

4) This week I'm going back to showing the top 20 best rated games, a list which rather strongly reinforces the idea that there are five elite teams with everyone else well behind them. All five top teams have multiple games on this list, and combine for eight of the top ten and fifteen of the top twenty. Moreover, of the five top teams, only Oregon only has two top twenty performances, and the Ducks have the #1 rated game to help balance it out. No other team has more than one top twenty performance, and the story wouldn't change much if I made it the top thirty instead.

5) This year I'm tracking the "Compu-Picks Curse" a bit more carefully. Below is the list of the teams that the system thought overrated each week (* means a bye/AA game or a game against someone else the model didn't like). So far teams have been exposed in one of twelve potential games, a huge reversal from last year's "at least one every week" pace).

After week 7 (0-2): Oklahoma St, Arkansas - none, though Arkansas came pretty close
After week 8 (1-3): Oklahoma St, Clemson, Nebraska*, Michigan St*, Arkansas, Virginia Tech - Clemson justifies my "wildly overrated" comment with a loss 31-17 at unranked Georgia Tech. Arkansas and Virginia Tech come very close against unranked opponents but pull out the wins. I won't count Michigan St since Nebraska was also on the list.
After week 9 (0-1): Arkansas*, South Carolina*, Virginia Tech*, Houston - none.
After week 10 (0-5): Oklahoma St, Arkansas, South Carolina, Virginia Tech, Houston - none, though Carolina got pushed pretty hard by an unranked, 4-loss (before the game) Florida team.

6) As usual, the BCS crushes teams for losses, even if they're close, even if they're against good teams, and regardless of how good they've been beforehand. Boise just had a 1 point loss to now-ranked TCU, and Stanford had a (not at all close) loss to an excellent Oregon team. Both of them dropped below Arkansas, Clemson AND Virginia Tech, which is just ridiculous. You can make a reasonable argument for Arkansas if you don't care about margin (i.e. the blowout loss to Bama and the handful of too close wins against mediocre opposition) and you emphasize recent results (both the Hogs' hot streak and the loss by Boise/Stanford)... but it's just silly to try and argue for Clemson or Virginia Tech over either Boise or Stanford. Boise and Stanford have played tougher schedules and done better against them. Losing by 1 at home to TCU is if anything a bit better than a 14-point loss at Georgia Tech, and you can very reasonably argue that Boise's schedule to date has been tougher than Clemson's. Ridiculous overreaction.

7) Ah, Virginia Tech. AWFUL schedule (ranked 62nd), home blowout loss, and squeakers against ECU, Miami and Duke. And it's not like they annihilated Ark St, Marshall, Wake and BC. This is simply NOT a top ten team. Them being ranked so high is simply insane, even after the 11-point Georgia Tech win. There's no other way to describe it. Completely embarrasing.

South Carolina is also pretty solidly overrated, with a meh schedule (rated 39th), a 16-point loss to Arkansas, a couple of way too close wins against teams that they shouldn't be struggling against (primarily Navy and Mississippi St), and nothing resembling a signature win. (the 3-pt win at Georgia was nice, but when that's your signature win, you don't have a signature win). Add it all up and a #12 ranking just doesn't make sense.

8) Team A beat Texas A&M 42-38 and hammered Auburn, South Carolina and Tennessee. Team B struggled against Troy (10 point win), Ole Miss (5 point), Vandy (3 point), and got thumped at Bama (38-14). Team A is a national elite that might not be at the LSU / Bama level, but isn't that far back either. Team B is a mess, and is iffy to even be in the top 25. Who are these two teams, you might ask? Actually, they're the same team. Two weeks ago, Arkansas's resume was mainly Team B with a really good 2-game stretch against A&M and Auburn as a seeming outlier. Now they added another two very good performances to their resume... and honestly who the hell knows? This team could beat Miss St by 30 and lose to LSU by 30... or they could blow the Miss St game and then somehow find a win at LSU the next week. Would either scenario REALLY surprise you? I'm not at all sure either should. So while the system still says "overrated" on the Hogs, it's a lot less of a strong opinion than it used to be (and much less strong than its opinions on South Carolina and especially Virginia Tech)... and when Team A shows up, they're not at all overrated. Of course, the question remains whether team B has been permanently gotten rid of or not... for that one, we'll just have to see what happens.

9) The following teams are ranked materially higher by the model than the BCS: Boise St, Stanford, Wisconsin, USC, Texas A&M.

Boise has played a tougher schedule than most give them credit for, they've been consistently dominant, and their one loss was an absolute nail-biter. They're simply a better football team than #10 in the BCS would indicate.

Stanford's rating seems fairly straight-forward: tougher schedule than most of the teams that Compu-Picks has under them, and they've done much better against that schedule than virtually everyone Compu-Picks has under them (Wisconsin and Houston did better, but against much worse schedules).

Wisconsin was simply punished far too harshly for their two close losses. Wisconsin remains very arguably the best in the Big Ten, and seems likely to win out.

Texas A&M is one of those teams the model likes because their losses have been very close (except for Oklahoma) and their wins (except for Tech) have been blowouts. While it's clearly too late to have a reasonable chance to play their way back into the Big 12 title race, they do have games remaining against KSU and Texas, and they've been playing well enough that they have a legit chance in each.

USC has an unusually strong schedule for a team in their rating range, which pushes them above most of their neighbors. The dominant win at Notre Dame win rates very well, especially compared to Michigan (who barely beat them at home) and Michigan St (who got hammered in South Bend).

10) The following teams are ranked materially lower than the model than the BCS: Arkansas, Clemson, Virginia Tech, South Carolina, USM, Baylor, Auburn.

I've discussed Arkansas, South Carolina and Virginia Tech above.

South Carolina has had a middling schedule for the Compu-Picks top 20, and has squeaked by way too often against that slate to be taken seriously. A 16-point loss, a 3 point win against Navy and a 2 point win at Mississippi St are simply bad performances for a team in anyone's top 15, much less top 10. And losing at home to Auburn (though it was close) really doesn't help either. Moreover, it's hard to look at their resume and find a single game that was all that impressive. A 3-point win at Georgia, an 18-point win against Vanderbilt and an 11-point win at Tennesse are fine showings, but when those are the shining moments of the season, that's a resume that just doesn't have much heft.

Forget how "they're not really THAT impressive" Houston's resume is... why the heck is USM suddenly a top 25 team? Their schedule has been a complete and total joke; they very arguably have yet to play a single top 40 team, and they have a loss (to a BAD Marshall team), and they have not one, not two, but THREE way too close wins against teams that a top 25 squad shouldn't struggle against (LA Tech, Virginia, UCF). A great example of an inflated record against a truly crappy schedule (as opposed to Boise's "we played a quality Georgia team and a few opponents with pulses" slate) suddenly getting BCS props.

Baylor has played a legit schedule, but they've been blown out twice and almost all of their wins have been nail-biters (except for an awful Rice team and a mediocre ISU one). And lest we forget, they just needed a massive comeback (not factored in to the model, but relevant for discussion purposes) just to register a one-point overtime win against Kansas.

Auburn just got blown out by Georgia (a good but not great team), got blown out by LSU, got blown out by Arkansas, lost by 14 to Clemson, and for good measure barely beat Utah St back in week one. Counter-acting these warts is... not all that much. A road win at South Carolina is nice, as is the 11-point home win over (5-loss) Florida... but they're a four-loss team with three blowout losses even before Bama comes into town. Even with the #10 rated schedule, a top 25 ranking just seems like a stretch here.

11) This isn't directly to do with the list, but here's fun lists of results:

@ Wisconsin 48, Nebraska 17
@ Michigan St 37, Wisconsin 31
@ Nebraska 24, Michigan St 3

Kansas St 41, @ Texas Tech 34
Texas Tech 41, @ Oklahoma 38
Oklahoma 58, @ Kansas St 17

@ Michigan 35, Notre Dame 31
@ Notre Dame 31, Michigan St 13
@ Michigan St 28, Michigan 14

If you try to apply "head to head is the only thing that matters" logic to this list, your head will explode. You can tease out certain information from these lists (Notre Dame's and Oklahoma's losses were close and their wins blowouts, therefore they get a bonus; Michigan's win came much earlier than their loss, therefore they get a demerit; etc.), but what it really does is highlight that each of these results was JUST ONE GAME. To properly evaluate a team, you need to evaluate the whole resume, not pretend that a single result means everything and the rest almost nothing just because of head to head "logic". That's why Compu-Picks doesn't give ANY special consideration to head to head results. You are what your resume says you are. Period.

Technical notes about the lists:

1) Conference ratings are straight averages of all of the teams in the league. There is no "central averaging" (like Sagarin does), or over-weighting the top teams, or anything like that. Such approaches would yield different numbers, and could potentially change the order of some of the leagues.

2) Games against AA teams are not counted. There are many good arguments both for and against counting such games (see this link for an interesting analysis of the issue). I have elected not to count these results in the Compu-Picks model. As is the case almost every year, this means that one or two especially surprising AA upsets don't make it into the numbers, skewing the results to a fair degree for a couple of teams. I believe that this is a more than acceptable tradeoff given the substantial issues that counting AA games would create, but you are certainly welcome to disagree with my decision on this matter.

3) As mentioned here, the purpose of this system is to make picks, not to create a list used for rankings. As such, I evaluate the system solely on the basis of how good a job it does making picks. I do not evaluate the system on the basis of whether or not it agreed with AP polls, BCS rankings, the BCS computers, or any other such list out there. In fact, the system has a long and established history of being substantially different than those sources. I am fine with these differences. To be honest, I publish these lists because I find them interesting and thought-provoking, and because I believe it is a good thing to introduce an approach that doesn't simply regurgitate the same avenues of thinking as you can find in most places.

4) The system is noisy, especially earlier in the year. This is why I start with only the top and bottom few, and slowly expand the list. While I believe that the numbers are reasonable, I certainly accept that they're not perfect. If you believe that a specific team is over- or under-ranked, you may well be right. I bring this up because if you're going to criticize the system for being wrong about a team, I'd appreciate it if you explain why you think the system is substantially wrong, rather than just marginally so (if it's just one or two slots off, especially well before the end of the year, I'd consider that well within a reasonable error range).

There are a few important notes and caveats I need to make about this model:

1) Compu-Picks does not endorse implicitly or explicitly any form of illegal gambling. Compu-Picks is intended to be used for entertainment purposes only.

2) No guarantee or warranty is offered or implied by Compu-Picks for any information provided and/or predictions made.

2011 Compu-Picks Blog

Questions, comments or suggestions? Email me at [e-mail]