2010 Compu-Picks Ratings - College

Mr Pac Ten
Posted Nov 3, 2010

The Compu-Picks model rates the top and bottom teams in college football after week nine

As of the end of MTSU - Ark St, these are the top 20 and bottom 10, along with conference ratings (straight average of the scores for each team in the confernece). 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 Team League Score Schedule Rank BCS Rank
1 Texas Christian Mountain West 0.88 41 3
2 Oregon Pac-10 0.87 20 1
3 Boise State WAC 0.84 39 4
4 Nebraska Big 12 0.70 25 7
5 Stanford Pac-10 0.67 17 13
6 Alabama SEC 0.66 34 6
7 Auburn SEC 0.64 12 2
8 Missouri Big 12 0.62 11 12
9 Arizona Pac-10 0.62 19 15
10 Virginia Tech ACC 0.60 58 22
11 Utah Mountain West 0.57 83 5
12 Ohio State Big Ten 0.57 74 11
13 Oklahoma Big 12 0.56 16 8
14 Louisiana State SEC 0.53 8 10
15 Iowa Big Ten 0.52 44 16
16 Oregon State Pac-10 0.51 1 NR
17 South Carolina SEC 0.49 14 19
18 Florida State ACC 0.46 29 24
19 Oklahoma State Big 12 0.46 55 17
20 Southern California Pac-10 0.44 18 NR
111 Middle Tennessee State Sun Belt -0.60 118
112 Western Kentucky Sun Belt -0.64 107
113 Ball State MAC -0.64 109
114 Louisiana-Lafayette Sun Belt -0.67 114
115 Buffalo MAC -0.69 87
116 Memphis C-USA -0.71 86
117 New Mexico Mountain West -0.78 90
118 New Mexico State WAC -0.79 108
119 Eastern Michigan MAC -0.82 105
120 Akron MAC -0.92 98

League Rating
Pac-10 0.36
SEC 0.30
Big 12 0.24
ACC 0.12
Big Ten 0.11
Indep 0.04
Big East 0.00
Mountain West -0.03
WAC -0.05
C-USA -0.22
MAC -0.42
Sun Belt -0.45

Some thoughts on the list:

1) Last week, I posted the compu-picks top fifteen and bottom ten, and this week I'm expanding to the top twenty and bottom ten. I will continue to 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. You can even see this in the scores, where there are a number of places where a few teams' ratings are packed very close together. This is even more the case further on towards the middle.

2) The picks have struggled so far, going 30 - 43 in the two weeks of ATS picks (though it did have a nice comeback this past week, with a 15-9 record). I am confident that it will return to the higher level that it has performed at in previous seasons, though if you want to feel skeptical about the whole thing until that happens, fair enough.

3) I'd like to give a tip of the cap to the model for being right about Michigan St. Before the Iowa game, they were the BCS #5, and not even in the compu-picks top 15. Vegas had them about a TD underdog... and turned out to have completely overrated them, as the Spartans just got humilated by the Hawkeyes. A team that needed a trick play in OT to beat a mediocre (and slipping) Notre Dame at home, and that only beat a really bad FAU team at home by 13 is probably not a top 5 team. Throw in that utterly humiliating result... and they're not a top 15 team. The BCS still has this team substantially overrated.

4) The following teams are ranked much higher by the model than the BCS: Nebraska, Stanford, Arizona, Virginia Tech. Putting aside Virginia Tech for now (as noted in previous weeks' notes, this difference is mainly due to not counting AA games... though 22nd is probably too low for how well they're playing), let's look at the other three. All three have had very tough schedules to date, and all three have been dominant to very dominant, well beyond what you'd expect just looking at the W/L record.

Nebraska (48-13 @ KSU, 31-17 vs Mizzou) and Stanford (41-0 @ Washington, 35-0 @ UCLA) have a pair of extremely impressive wins which have substantially helped their ratings. Arizona doesn't have any single really exceptional performance, but have a number of very solid ones, such as the 41-2 blowout at Toledo, the 34-27 win against Iowa, and the 29-21 win @ UCLA (who's rated better than you'd think; 3-5 isn't great, but it's been against a truly brutal slate, and that Texas win still looks pretty good, though not as good as it did at the time).

5) The following teams are ranked much lower than the model than the BCS: Auburn, Utah, Oklahoma, Wisconsin, Michigan St. Utah is the easy one: it's the schedule. They've been dominant, but the schedule just hasn't been good enough for it to really move the dial. It's also worth noting that in their toughest game to date, at Air Force, they won a pretty close game, by only five points. That's a worrisome sign since that was the first of a tough four-game stretch. If they do well in that stretch, their rating will rise, but right now, the system isn't high on their chances of a sweep; in fact, back to back games at Notre Dame and SD St could well prove tougher challenges than expect.

Michigan St was covered above; not especially dominant (especially after they just got waxed at Iowa), not much of a schedule to date (Notre Dame was actually the best by far of a really bad set of non-conference opponents, which continues to hold them down, though not as much now that they're well into the Big Ten schedule). Wisconsin is pretty much the same story, with an OK schedule, but similar issues with dominance, or lack thereof (how a "top ten" team escapes ASU by one point and, far worse, SJ St by only 13, is a mystery to me).

Oklahoma is getting dinged mainly because of all of those weirdly close wins. Nothing especially interesting to say here.

Auburn is a more interesting case. Clearly the schedule heft is there; right now their schedule to date is rated as the toughest of all the unbeaten teams. What's really been holding them back has been the unusual string of close games. 3 points at home in overtime to Clemson, 3 points each at Kentucky and Miss St... those really aren't the type of performances you expect from a truly elite team. On the other hand, they did beat both Arkansas and Ole Miss by more than 20, so that criticism is starting to fade. Should they run the table, including a win at Alabama, their rating will materially increase... though right now the model doesn't give them a great shot at winning at Tuscaloosa. For now, the model thinks that Alabama is actually the best team in the SEC (though it's close). Is the model right? We'll see in a few weeks (unless LSU beats the Tide, in which case we'll have an answer sooner than that). For what it's worth, I suspect that the model underrates them by a couple spots... though I think it's premature to say they're really a top 2 team. If they can pass their big road test at Bama I'll be convinced, but until then I'm on the fence about them to some degree.

6) Remember that gap between Oregon and the previous top 2 (Boise and TCU)? Yeah, so much for that. Oregon's extremely impressive 21-point win at USC made short work of it, passing Boise and basically moving into a tie with TCU. The Ducks now own two extremely impressive 21-point wins, against Stanford and USC, plus a number of other solid showings as well. Other teams will have a chance to move their own ratings up, but this Duck team is playing at a very good level. If they can continue their dominant pace in a difficult final three-game stretch (obviously the Washington game isn't as big of a hurdle), their rating would rise even further.

7) Boise has taken a hit even without playing (their game against LA Tech counted for the week 8 ratings). Their score itself moved about 0.01 down (some opponents' results helped, others hurt, net was a bit negative), but the main thing was that Oregon simply passed them. In the week 8 numbers, Boise had the tougher schedule, but after Oregon's big game at USC the Ducks' schedule strength grew substantially. The good news is that their schedule strength for their last five games is pretty comparable in aggregate to their first seven games, so their schedule rating shouldn't take much (if any) of a hit (it'll sag before Nevada, but then jump up after that game). The problem, however, is that, as expected, a number of the AQ teams may have a chance to pass Boise as their own schedules toughen up. Obvious examples of games where the winner should get a solid boost are Arizona-Stanford, Auburn-Alabama, and the Big 12 title game. The Broncos do have a solid lead on the rest, but as shown by Oregon's recent jump up in the numbers, that lead can be overcome. Boise needs to continue to dominate and hope that there's some chaos in the rest of the field, which pretty much reflects their reality in the rankings that actually count.

8) I'd say Akron reached a new low by getting shut out by Temple... but the ass-kicking Western Michigan gave them the week prior was actually a lot worse. This is just a horrible football team.

9) Last week I talked a bit about league ratings. This week I'll talk about one specific group of team ratings, why compu-picks rates them how it does, and why I feel that this is substantially superior to the BCS approach. The teams I'll look at are three one-loss teams: Wisconsin, LSU, and Stanford. The BCS has them rated 9th, 10th, and 17th in order, while compu-picks has their order exactly reversed: 24th (outside top 20), 14th, and 5th. Let's compare these three teams directly to see which ranking order makes more sense. Since compu-picks only looks at 1-A games, we'll do the same in this comparison (since each team won fairly comfortably against a AA opponent, I don't think ignoring those games loses much, though it's worth noting that Wisconsin really blew the doors off of their AA opponent, while LSU's was the closest [though still pretty convincing] ).

First of all, let's look at the losses. I don't think it's at all controversial to argue that Wisconsin has the worst of the three losses, by a substantial amount, since LSU and Stanford have lost to the top 2 teams in the BCS, while Wisconsin lost to a team that got demolished by Iowa, that barely (and I do mean barely) beat Notre Dame at home, and struggled mightily against a very bad FAU team. Comparing the LSU and Stanford losses is a bit tougher. Stanford's loss was less close than LSU's, but at least in compu-picks' mind, Stanford's loss was to the better team. Compu-picks thinks that overall they're comparable, but if you disagree with the idea that Oregon is a good deal better than Auburn, than LSU's is the least bad loss. For the sake of the argument, we'll go with that conclusion, which means that in the "how bad was their loss" category, it's:
LSU > Stanford >> Wisconsin

Next, let's look at the worst of each team's home wins. It's pretty obvious for each:
LSU 16-14 vs Tenn, Stan 38-28 vs WSU, Wisc 27-14 vs SJ St.
LSU had the toughest opponent, and Wisconsin the easiest. Meanwhile, LSU had the closest win, Wisconsin the most convincing. Honestly, all three wins were very unimpressive. The model doesn't think there's a huge difference here, balancing margin and difficulty. You may disagree, but I think it's fair to call it:
LSU ~ Stanford ~ Wisconsin (~ means about equal)

Next, let's look at the 2nd worst of each team's home wins. Again, this isn't going to be particularly controversial, though you can argue Wisconsin's way too close win against ASU belongs here instead, or that Wake is bad enough that it should be counted instead for Stanford:
LSU 20-14 vs WV, Stan 37-35 vs USC, Wisc 41-23 vs Minn
I don't think it's controversial to say that USC is sufficiently better than West Virginia for Stanford's win to register as a bit better than LSU's, though the margin difference certainly bridges the gap a fair amount. And I don't think it's at all controversial to say that both are substantially better than an 18-point win against a bad Minnesota team. So:
Stanford > LSU >> Wisconsin

Next, let's look at the worst of each team's road wins:
LSU 27-3 @ Vandy, Stan 37-14 @ ND, Wisc 41-21 @ UNLV
These aren't identical scores, but they're reasonably comparable, making it easy to just arrange them by difficulty level:
Stanford > LSU >> Wisconsin

Next, let's look at the 2nd worst of each team's road wins:
LSU 33-29 @ Florida, Stanford 41-0 @ Washington, Wisconsin 31-30 @ Iowa
Obviously in terms of difficulty level, Wisconsin's was toughest (though not by a massive margin), and Stanford's easiest (by a lot). However, it's the reverse in terms of dominance, with Stanford registering a hugely dominant win, while LSU's win was more dominant than Wisconsin's (though again, not by a huge margin). The model was VERY impressed with 41-0 (blowout AND shutout), partially because it rates Washington around 60th (remember, they beat up Syracuse [who's looking better by the week], and they beat USC and Oregon St [though both were VERY close], which for a bottom-half AQ team is a nice set of three wins). It thinks that Stanford had the best win by a solid margin, although you may well disagree with this one. Just remember, Iowa did lose twice, and Florida has three losses already. These aren't top 10 wins, so if you're willing to buy the idea that the extreme dominance more than overcomes the difficulty, then it makes sense to rate the wins:
Stanford >> LSU ~ Wisconsin
However, if you'd rather call them about a wash, for the sake of the argument I'm willing to go down that road. If you want to argue that Stanford's win is materially less impressive than the other three, though, I'm going to have to strongly disagree there.

Only two games left for each. Let's look at:
LSU 30-24 v UNC (neutral), Stan 68-24 vs Wake, Wisc 20-19 vs ASU
Wisconsin's way too close win against ASU was an obvious red flag in their rating; legit top 10 teams don't beat ASU at home by one point. The model rates LSU's and Stanford's wins as reasonably comparable, with a small edge to LSU, since they played the better team, and on a neutral site (Stanford and Wisconsin's games here were both home). Since I'm sure the margin emphasis is unpopular, for the sake of the argument let's say that gap is bigger than the model makes it, so we then have:
LSU > Stanford > Wisconsin

And then we have the final game for each:
LSU 29-7 vs Miss St, Stan 35-0 @ UCLA , Wisc 31-18 vs Ohio St
Certainly Wisconsin's game was the toughest test, but once again it was also the closest. What's interesting here is that LSU's and Stanford's games were rated as reasonably comparable in difficulty; Miss St was rated 26th and UCLA 45th (they still won handily against Houston and at Texas, and gave a very good Arizona team a fairly close game), and home-field advantage (since Stanford was away and LSU home) approximately covered that gap. The model was extremely impressed by the 35-0 win at UCLA, and thought it was one of the best performances anyone has had this year; the LSU and Wisconsin games were rated as pretty comparable overall, with a small edge to LSU because they had the more dominant win (and Ohio St has been a good but not great football team). So we have for this group:
Stanford >> LSU ~ Wisconsin

So what are we left with? Combining dominance with schedule difficulty, it likes Stanford the best of the three, LSU second, and Wisconsin third, each by a fair margin. However, if you think that dominance shouldn't be emphasized as much, it's absolutely reasonable to argue that LSU should be rated comparably to Stanford, and if you completely disregard margin, you can argue them ahead (as shown by their differing schedule strengths). However, it appears to make no sense whatsoever for Wisconsin to be rated above either LSU or Stanford, much less both. This is true whether or not you care about margin, as the Badgers have had the easiest schedule of the three (and factoring in margin only makes the comparison worse).

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

Texas 20, @ Nebraska 13
Nebraska 48, @ Kansas St 13
@ Kansas St 31, UCLA 22
UCLA 34, @ Texas 12

@ Hawaii 27, Nevada 21
@ Nevada 52, Cal 31
@ Cal 52, Colorado 7
@ Colorado 31, Hawaii 13

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 (UCLA had both games on the road, they get a bonus; Nevada's loss was the only close result from their list, therefore they get a bonus; 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 a top 10 / bottom 10 list, and slowly expand it. 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).

2010 Compu-Picks Blog

Questions, comments or suggestions? Email me at cfn_ms@hotmail.com

Related Stories
Martinez back at it
 -by BigRedReport.com  Nov 2, 2010
Analyzing Who the Top BCS Teams Have Played
 -by autigers.com  Nov 4, 2010
The Inside Slant
 -by TigerSportsDigest.com  Nov 2, 2010