Writers of Pro Football Prospectus 2008

Most Recent FO Features

HarvinPer09.jpg

» Impact of the NFL's Kickoff Rule Change

After three NFL seasons of kicking off from the 35-yard line, what has been the impact on touchbacks, returns, field position, scoring and injuries? Also, is this rule responsible for a record number of big comebacks?

14 Jan 2011

Varsity Numbers: Auf Wiedersehen, 2010

by Bill Connelly

When Auburn's Wes Byrum booted the title-winning field goal through the University of Phoenix uprights Monday night -- a little closer to the posts than some Auburn fans' hearts would have preferred -- one of the more interesting college football seasons I can remember came to an end. The 2010 season brought to the table everything one could like about college football -- just look through the recent Varsity Numbers archives to see how much fun I was having in October and November -- and everything one could dislike.

The Cam Newton investigation managed to hang a bit of a black cloud in the final month of the season, with the NCAA's eventual decision feeling both rushed and unsatisfying. The Ohio State suspension decision (key Ohio State players are suspended for a good chunk of next season ... but go ahead and play the bowl game, guys) was both creative and, yes, unsatisfying. But in the name of positivity, we won't focus on that. Instead, we look back on a bowl season that was enjoyable if you knew where to look, and we tie a big bow around the 2010 season.

Box Score of the Bowl Season

A game filled with mistakes, shoddy sod, and an odd, polarizing, all-but-title-clinching run, still managed to provide a healthy amount of excitement. A sloppy, nervous start gave way to big plays, goal-line stands, a dramatic game-tying drive, and a game-winning field goal.

We spent most of the first half wondering just how badly TCU might have beaten either one of these two teams, but by the end, the title ambiance took over. Congrats to Auburn for doing what it took to lock away a national title. Newton and the 2010 Tigers get to place their crystal football next to ... whatever hardware the 1957 Tigers and their dominant defense took home with their national title. Here's to hoping they don't have to relinquish the title in a few years. I want this one to end up legitimate, just for college football's sake, not to mention the sake of my two Auburn friends.

No. 1 Auburn 22, No. 2 Oregon 19

Auburn Oregon Auburn Oregon
Close % 100.0% STANDARD DOWNS
Field Position % 41.7% 53.4% Success Rate 57.8% 38.0%
Leverage % 76.2% 68.5% PPP 0.34 0.28
S&P 0.921 0.657
TOTAL
EqPts 25.3 22.6 PASSING DOWNS
Close Success Rate 52.4% 38.4% Success Rate 35.0% 39.1%
Close PPP 0.30 0.31 PPP 0.17 0.38
Close S&P 0.825 0.694 S&P 0.516 0.774
RUSHING TURNOVERS
EqPts 9.4 3.3 Number 2 2
Close Success Rate 54.2% 26.7% Turnover Pts 8.3 8.9
Close PPP 0.20 0.11 Turnover Pts Margin +0.6 -0.6
Close S&P 0.737 0.378
Line Yards/carry 3.49 1.91 Q1 S&P 0.231 0.634
Q2 S&P 1.131 0.921
PASSING Q3 S&P 0.798 0.681
EqPts 15.9 19.3 Q4 S&P 0.653 0.653
Close Success Rate 50.0% 46.5%
Close PPP 0.44 0.45 1st Down S&P 0.950 0.725
Close S&P 0.942 0.914 2nd Down S&P 0.699 0.631
SD/PD Sack Rate 0.0% / 15.4% 3.9% / 5.9% 3rd Down S&P 0.798 0.574
Projected Pt. Margin: Auburn +3.3 | Actual Pt. Margin: Auburn +3

Throughout the final three quarters, it felt like Auburn was in control, and it was but a matter of time before they broke the game wide open. This is reflected most clearly in both leverage rate and success rate. Auburn was consistently staying in standard downs and moving the ball effectively, while Oregon needed big plays to bail them out of what were far too many passing downs.

The Ducks were dominated on both sides of the line of scrimmage, but in times of need, they came up big through the air. Darron Thomas found Jeff Maehl for 81 yards to set up the Ducks' first touchdown, then found Lavasier Tuinei for 43 yards to set up what appeared to be the game-tying, third-quarter touchdown. Auburn freshman Demetruce McNeal made the tackle of his life, however, tripping Tuinei up at the three to set up Auburn's goal-line stand. Then, with a pick from the umpire, Thomas and D.J. Davis hooked up for 29 yards on fourth-and-5 to set up Oregon's game-tying touchdown late in the fourth quarter. Oregon's major problem: The 70 plays not mentioned above recorded only 296 yards, and they gained more on passing downs than standard downs. That is not a recipe for success.

What Auburn lacked in explosiveness, they made up for in efficiency. Their rushing success rate more than doubled Oregon's, and while the Ducks had the really big plays, the Tigers more effectively stayed on schedule. Field position wrecked their chances of sticking in the dagger -- their average starting field position in the second half was their own 21, and they would typically move the ball a decent amount then stall before stretching the lead to double digits.

This was not the festival of offensive fireworks that we imagined, of course. Both teams moved the ball, but they would usually trip before reaching the goal line. My colleague Rob already heaped a sufficient amount of praise on Auburn defensive tackle Nick Fairley and the havoc he wreaked, so I do not need to go too much further into it here. But in the end, Oregon's inability to move the ball on the ground resulted in them needing one too many magic tricks to win the game.

Ten Notable Games

Rose Bowl: No. 3 TCU 21, No. 5 Wisconsin 19

EqPts: Wisconsin 21.5, TCU 20.7
T/O Pts: none
Wisconsin > TCU +0.8

While the national title game may have been occasionally marred by turnovers and players slipping on the iffy turf, the Rose Bowl was a clean, enjoyable, nip-and-tuck game. It was also one of the quickest games I've seen in a long time. In the end, both teams did just enough to claim victory, but Tank Carder's deflection of Scott Tolzien's two-point pass attempt gave the edge the Horned Frogs. The S&P+ rankings have loved TCU for a couple years, and I wonder what they might have been able to do against Auburn. Alas, they appear perfectly thrilled with a second-place finish.

Allstate Sugar Bowl: No. 6 Ohio State 31, No. 8 Arkansas 26

EqPts: Ohio State 28.4, Arkansas 22.2
T/O Pts: Ohio State +5.0
Ohio State > Arkansas +11.2

Ohio State was on the right side of the turnovers on January 4, and it gave them the edge. We can debate whether Terrelle Pryor and the other soon-to-be suspended Buckeyes should have been playing in this game (it is much easier to make the case that they shouldn't have), but they did play, and it was an exciting game. Ohio State outgained Arkansas and maintained a comfortable lead until the fourth quarter, but two plays made the difference. Terrelle Pryor fumbled near the Arkansas goal line in the first quarter, and receiver Dane Sanzenbacher recovered in the end zone after two Arkansas players ran into each other, giving the Buckeyes a touchdown instead of a six-point turnover. Then, with the game in the balance after a blocked punt (Don't fall on the ball!!), Solomon Thomas picked off Ryan Mallett after Mallett failed to properly read a zone blitz.

Discover Orange Bowl: No. 4 Stanford 40, No. 13 Virginia Tech 12

EqPts: Stanford 36.0, Virginia Tech 13.2
T/O Pts: Virginia Tech +2.9
Stanford > Virginia Tech +20.1

Virginia Tech played a very Virginia Tech-esque first half, stealing points wherever it could, with a safety and an incredible touchdown pass from Tyrod Taylor to David Wilson, and this was a 13-12 game at halftime. But apparently the Hokies had punched themselves out. Stanford scored four unanswered touchdowns in the second half, and though the stats project a closer game, Jim Harbaugh's final game on the Stanford sideline was one of domination.

Tostitos Fiesta Bowl: No. 7 Oklahoma 48, Connecticut 20

EqPts: Oklahoma 31.5, UConn 14.2
T/O Pts: Oklahoma +2.6
Oklahoma > UConn +19.9

UConn won the Big East despite an offense ranked in the sixties in Offensive F/+ and a defense ranked in the forties in Defensive F/+. They did so by creating opportunities from special teams and turnovers, and doing exactly that allowed them to make it a game for a while in the Fiesta Bowl. They returned an interception for a touchdown in the second quarter, then returned a kickoff for a touchdown in the third. But they couldn't get closer than 14 points, and eventually Oklahoma pulled away. The Sooners scored off of two pick-sixes of their own. While their running game never really got going, they shredded UConn with the pass -- Ryan Broyles and Cameron Kenney combined for 20 catches and 324 yards. With Broyles and linebacker Travis Lewis returning for their senior seasons next year, Oklahoma is emerging as the favorite of everybody's Too Early 2011 Top 25 lists.

Capital One Bowl: No. 16 Alabama 49, No. 9 Michigan State 7

EqPts: Alabama 41.2, Michigan State 9.6
T/O Pts: Alabama +4.9
Alabama > Michigan State +36.5

Our FO metrics never stopped liking Alabama (in terms of F/+ rankings, they finished with a loss at home to the No. 1 team and road losses to No. 11 and No. 13) and, at the same time, didn't show a lot of respect to Michigan State (though 11-1, they ranked just 24th heading into the bowls). But we still didn't see this level of domination coming.

AT&T Cotton Bowl: No. 11 LSU 41, No. 17 Texas A&M 24

EqPts: LSU 31.1, Texas A&M 17.3
T/O Pts: LSU +14.9
LSU > Texas A&M +28.7

For the second time in three seasons, LSU quarterback Jordan Jefferson looked great in a bowl game. At his best, he is not a tremendously high-percentage passer, but his three career bowls have produced solid results -- 57-percent completion rate, 7.4 yards per pass, five touchdowns, two interceptions -- and have effectively teased Jefferson's potential. He will enter his senior season in 2011 with a new offensive coordinator and plenty of talent around him. Then again, the surrounding talent has never been an issue in Baton Rouge.

Maaco Bowl Las Vegas: No. 10 Boise State 26, No. 19 Utah 3

EqPts: Boise State 26.2, Utah 5.7
T/O Pts: Utah +4.4
Boise State > Utah +16.2

Boise State finishes with the highest-ranked defense in the country, according to Defensive S&P+. Their second-half fade against Nevada was the only chink in their armor in 2010, and they will return a ton of experience and talent next year. The offense loses its big-play receivers, but quarterback Kellen Moore and the Broncos will still win plenty of games behind the defense and outstanding running back Doug Martin. Martin is a power runner who rarely gets caught in the open field, and he's fun to watch.

Chick-Fil-A Bowl: No. 23 Florida State 26, No. 20 South Carolina 17

EqPts: Florida State 18.9, South Carolina 14.5
T/O Pts: Florida State +19.7
Florida State > South Carolina +24.1

Congratulations, Florida State! This bowl win (which the stats thought should have resulted in much larger than a nine-point margin) all but solidified your status, along with Texas A&M, as Next Year's Nebraska, the team every analyst picks as their "dark horse" title contender ... making you no longer a dark horse. Nebraska took the honors in 2010, Ole Miss and Oklahoma State in 2009, and Clemson and Texas Tech in 2008. Almost every one of those teams failed to live up to expectations, of course, but Florida State should be well-stocked with both experience and improved recruiting. Between the 'Noles and Texas A&M, Florida State will likely find itself on much more solid footing in 2011, and with a much less difficult conference schedule (though they do have to deal with both Oklahoma and Florida in non-conference play).

Progressive Gator Bowl: No. 21 Mississippi State 52, Michigan 14

EqPts: Mississippi State 38.8, Michigan 18.7
T/O Pts: Mississippi State +2.9
Mississippi State > Michigan +23.0

There's a chance that Florida State and Texas A&M will have to share the "dark horse" spotlight with the SEC West's latest up-and-comer. Mississippi State got Rich Rodriguez fired and established themselves with some serious momentum for 2011, especially when Dan Mullen remained the coach despite overtures from other schools.

Insight Bowl: Iowa 27, No. 12 Missouri 24

EqPts: Missouri 25.1, Iowa 22.8
T/O Pts: Iowa +7.7
Iowa > Missouri +5.4

Blaine Gabbert's final game in a Missouri uniform was going swimmingly until he suffered perhaps the biggest brain lapse of his college career. With Missouri ready to take a two-possession lead, Gabbert was flushed from the pocket and attempted to lob a pass to a receiver instead of throwing the ball out of bounds. The receiver, Wes Kemp, was assuming Gabbert would run and was blocking for him, and the pass itself did not have enough lob. Micah Hyde picked it off and returned it for the game-winning touchdown. It was the one glitch in a game that saw Gabbert make every throw an NFL scout would have wanted to see. In the end, the game was decided by the magnitude of turnovers. Missouri forced two semi-costly Iowa turnovers, while Iowa picked off one pass in their own end zone, and returned the other to Missouri's end zone.

Final F/+ Rankings

During the season, I unveiled the S&P+ rankings in Tuesday's Varsity Numbers, then the F/+ rankings on Friday. We'll skip a step here and just say that you can find the final S&P+ rankings here:

As expected, Boise State maintained its first-place S&P+ standings with its dominant-after-the-first-quarter performance against a solid, if faded, Utah team. Meanwhile, Oregon's performance against Auburn boosted its Defensive S&P+ ranking enough to get them up to No. 16.

And here's where the F/+ rankings find their value. If you look at the S&P+ rankings, you quickly begin to question No. 11 Notre Dame, No. 15 Miami and No. 16 Oregon, especially. On the FEI side, both N.C. State and West Virginia seem a bit high. F/+ balances out the outliers and produces what is, for me at least, a satisfying set of rankings.

RK TEAM RECORD F/+ LAST
WK
CHANGE S&P+ RK FEI RK OFF.
F/+
RK DEF.
F/+
RK
1 Auburn 14-0 +32.1% 1 +0 265.3 4 0.348 1 +21.6% 1 +10.5% 11
2 Boise State 12-1 +30.5% 2 +0 286.3 1 0.213 11 +16.3% 5 +14.2% 4
3 Alabama 10-3 +29.2% 4 +1 267.1 2 0.281 3 +17.0% 3 +12.2% 7
4 Stanford 12-1 +27.9% 5 +1 257.2 6 0.302 2 +16.3% 4 +11.6% 10
5 Ohio State 12-1 +26.4% 3 -2 265.3 3 0.234 8 +7.8% 23 +18.7% 1
6 TCU 13-0 +25.2% 6 +0 261.2 5 0.229 10 +11.0% 11 +14.2% 5
7 Arkansas 10-3 +23.2% 9 +2 249.9 9 0.245 7 +13.6% 6 +9.7% 16
8 Oklahoma 11-3 +22.7% 7 -1 250.6 8 0.231 9 +11.0% 10 +11.7% 8
9 Oregon 12-1 +21.5% 11 +2 238.2 16 0.267 4 +9.8% 15 +11.7% 9
10 Virginia Tech 11-3 +21.2% 8 -2 240.9 14 0.249 5 +17.6% 2 +3.7% 39
11 South Carolina 9-5 +21.0% 10 -1 253.1 7 0.185 16 +10.8% 13 +10.2% 12
12 Wisconsin 11-2 +20.4% 12 +0 245.1 10 0.211 12 +10.2% 14 +10.1% 13
13 LSU 11-2 +19.0% 13 +0 232.3 22 0.247 6 +1.5% 48 +17.5% 2
14 Oklahoma State 11-2 +18.4% 18 +4 242.7 12 0.184 17 +13.0% 8 +5.4% 32
15 Florida State 10-4 +17.9% 19 +4 241.2 13 0.182 18 +8.6% 19 +9.4% 19
RK TEAM RECORD F/+ LAST
WK
CHANGE S&P+ RK FEI RK OFF.
F/+
RK DEF.
F/+
RK
16 Missouri 10-3 +17.6% 15 -1 238.0 17 0.190 14 +8.6% 18 +9.0% 23
17 Notre Dame 8-5 +16.7% 22 +5 244.7 11 0.140 23 +6.9% 29 +9.8% 15
18 West Virginia 9-4 +16.4% 17 -1 233.1 20 0.190 14 +1.6% 47 +14.8% 3
19 Miami-FL 7-6 +15.5% 14 -5 239.0 15 0.143 22 +10.9% 12 +4.6% 36
20 Nebraska 11-3 +15.0% 16 -4 229.5 26 0.179 19 +8.4% 21 +6.6% 28
21 Iowa 8-5 +14.8% 21 +0 232.4 21 0.161 21 +5.2% 32 +9.6% 17
22 Pittsburgh 8-5 +14.6% 23 +1 230.3 24 0.168 20 +13.5% 7 +1.1% 48
23 NC State 9-4 +13.7% 26 +3 218.4 38 0.209 13 +7.1% 28 +6.7% 27
24 Texas A&M 9-4 +13.2% 20 -4 234.5 18 0.120 28 +0.4% 54 +12.9% 6
25 Clemson 6-7 +11.2% 25 +0 224.3 34 0.130 24 +1.9% 45 +9.4% 20

Was Oregon better than the ninth-best team in the country? Probably. But Monday night's performance, fair or not, took some of the sheen off of the Ducks' offense, which was the basis of their high rankings. If I had a vote, I'd have placed them in the Top 5, but ninth does not seem out of the realm of possibility. Meanwhile, Nevada is the only one-loss team to finish outside the Top 10, which was to be expected considering they began bowl season ranked 29th and beat No. 49 Boston College by just a touchdown.

Biggest F/+ Movers of the Week

There were not too many big movers because a) the sample size is pretty large now, and b) the F/+ rankings were pretty dialed in heading into bowl season. We may not have pounded our chests about this enough, but F/+ picks went 22-12-1 (64.3%) against the spread in bowl games. There were some misses, obviously, but the bowl season suggests that the biggest problem F/+ faces in terms of week-to-week picks is the value of momentum, injuries, suspensions, and other in-season developments. Force everybody to take three weeks off and cool off or warm back up, and the F/+ did reasonably well in terms of evaluation/prediction.

That, or the bowl performance was a total fluke. I'm open to either possibility.

Rises

Washington (14 spots, from 65th to 51st). Here's where we get into trouble making evaluations about bowl performances. The general consensus following the Holiday Bowl was that Nebraska had no interest in being there and laid a giant egg. To a certain extent, I'm sure that was true. But in evaluating each team's performance versus what was to be expected, Washington's own performance in this game was probably underrated. The Huskers fell four spots, which suggests that at least a small egg was laid, but the Huskies rose considerably after shutting down a still-decent Nebraska offense.

BYU (10 spots, from 63rd to 53rd). The Cougars were expected to handle UTEP, the bowl season's lowest-ranked team, easily. But they didn't necessarily have to do it that easily. BYU looked outstanding. Freshman quarterback Jake Heaps will enter 2011 with a ton of hype, and considering his September-to-December development (and his recruiting pedigree), it probably will not be unwarranted.

Mississippi State (five spots, from 31st to 26th). Granted, finishing 26th places you fifth in the SEC West, but this was still an outstanding showing, in a January 1 showcase, for Dan Mullen and the Bulldogs. They will go about their business hoping that Auburn and Arkansas both take tumbles next season after losing their quarterbacks.

Notre Dame (five spots, from 22nd to 17th). The Notre Dame defense became sneaky good in the last half of the season, while the offense improved as well. I'm curious where the Irish end up in the preseason polls. If the way they closed 2010 is any indication, they might actually have earned what are typically irrationally high preseason votes.

Other Rises: Troy (80th to 69th), San Diego State (53rd to 45th), Tulsa (67th to 60th), Central Florida (42nd to 37th).

Tumbles

Baylor (eight spots, from 54th to 62nd). Baylor fans will still look back at the 2010 season with fondness, but wow, did they take a tumble. The night of October 30, the Bears were 7-2 and coming off of their first win in Austin in ages. Then, in their last four games, they were outscored a Baylor-esque 188-96 and limped to a 7-6 finish. The defense must improve significantly in 2011 if the Bears want to move from simply salty to week-in-week-out dangerous.

Michigan (six spots, from 36th to 42nd). I really do think the Brady Hoke hire was decent. We tend to overreact to coaching hires -- Can't miss! Terrible! -- but the more I look into coaching changes, the more I realize how completely unpredictable they are. All we know is that Hoke led Ball State to an undefeated season -- yes, it took him a little while to do so, but the "He has a career losing record!!' cries are both unfair and disingenuous -- and needed just one year to bring moribund San Diego State to nine wins and bowl success. We don't know that he will succeed in Ann Arbor, but he has passed the tests he has taken as a college head coach.

Michigan State (six spots, from 24th to 30th). It's a shame that the half-pathetic performance the Spartans offered against Alabama will cast a pall on what was otherwise an exciting, successful season for the Spartans. They probably weren't "11 wins" good, but they were good, and they will likely be good again in 2011.

Miami (five spots, from 14th to 19th). Some teams find just the magic formula for statistical success, and Miami certainly qualifies. Anybody who watched the first half of the Sun Bowl would struggle to consider the Hurricanes a Top 50 team, much less Top 20, but every system has its outliers. This does suggest that Al Golden might have an opportunity for quick success at The U, however. There is at least enough talent in place to impress Brian's computer and mine. So they've got that going for them, which is nice.

Other Tumbles: East Carolina (58th to 67th), Ohio (85th to 91st), Fresno State (68th to 73rd).

The Playlist

"Bye," by J Dilla
"Bye Bye Baby," by Social Distortion
"Bye Bye Love," by The Everly Brothers
"Everytime We Say Goodbye," by John Coltrane
"Go and Say Goodbye," by Buffalo Springfield
"Goodbye Pork Pie Hat," by Charles Mingus
"Hello, Goodbye," by The Beatles
"Hey, That's No Way To Say Goodbye," by Leonard Cohen
"Long Slow Goodbye," by Queens of the Stone Age
"Never Say Goodbye," by Bob Dylan

Favorite Moment of the Bowl Season

I had no real emotional attachment to the Texas Bowl matchup between Illinois and Baylor. In fact, to the extent that I did have a rooting interest, it was in favor of Baylor quarterback Robert Griffin III. But watching Ron Zook's surprised, thrilled, and slightly psychotic reaction to the Gatorade bath he received after Illinois' blowout win was totally worth the price of admission. Each bowl season, coaches like Zook (whose tenure was left for dead a few months ago before "Hire a bunch of new assistants" Hail Mary actually worked for once), players like Washington's Jake Locker, and programs like Troy get to end the season as winners, which makes college football an entirely unique experience. For a couple of minutes, Zook probably felt like he had won the national title. All the NCAA investigations and Death to the BCS!! chants make little moments like that all the more rewarding and enjoyable to watch.

Posted by: Bill Connelly on 14 Jan 2011

25 comments, Last at 15 Jan 2011, 10:41pm by zlionsfan

Comments

1
by cfn_ms :: Fri, 01/14/2011 - 1:49pm

The obvious headline was "Nebraska doesn't care". A less obvious (but also as defensible) one was "Nebraska's offense sputters again" (see: Texas, Texas A&M, SD St, and to a lesser extent the mediocre showings against Kansas's lousy defense and the 4-turnover mess against Oklahoma in the title game).

2
by cfn_ms :: Fri, 01/14/2011 - 2:08pm

Higher than consensus:
Boise (big outlier)
Alabama
South Carolina (big outlier)
Arkansas
Notre Dame (big outlier)
Miami (big outlier)

Lower than consensus:
Auburn
TCU
Stanford
Oregon (big outlier)
LSU (big outlier)
Nevada (big outlier)

Any thoughts on the differences b/w S&P results and general consensus? Are there ones you especially agree with and ones you especially disagree with? And are you planning on adjusting the system due to what apparently big outliers like Boise, South Carolina, Notre Dame, Miami, and especially Oregon?

My guess (as noted previously) is that the schedule adjustments are the biggest thing to fix for the 2011 version. I'd suggest going at least 4 layers deep (instead of just "avg opponent results" and "avg opp-opp results"), or do a recursive rating structure, or use some form of regression.

I'd also pretty strongly suggest factoring in home-field advantage, since it's a pretty meaningful predictive variable. Plus it should help explain a couple of the outliers, since Notre Dame only had 3 true road games (it's probably reasonable to count Army as a road game to bump it to 4, but Navy was clearly a neutral-site game), and A&M and Arkansas only had 4 true road games (plus neutral site vs each other). And on the flip side, Oregon, Stanford and Nevada all had 6 road games, and would have gotten a bump from factoring in HFA.

3
by Brian Fremeau :: Fri, 01/14/2011 - 2:24pm

Actually, for Notre Dame, the Army game was more of a home game than Navy, I'd say. It was an off-site home game in terms of ticket sales, attendance, pep rallies, etc. ND/Navy in attendance is split more evenly, but ticket sales belong to Navy.

Agree on the three true road games.

6
by cfn_ms :: Fri, 01/14/2011 - 4:02pm

Without having paid much attention to either of the two games, you could at least make a case that Army was a road game since it was in the same state as Army, even though it was a much different part of the state. I think I counted it as a road game in my numbers, though maybe I misremember.

wrt Navy, that's a neutral site game on the very face of it (even if Navy wants to call it a "home game"). Substantial distance for both to travel, and you know that ND brings more casual fans than Navy.

But the big thing is that HFA is such a consistently huge predictive factor you simply HAVE to account for it when putting together schedule numbers.

4
by Kal :: Fri, 01/14/2011 - 2:39pm

"Was Oregon better than the ninth-best team in the country? Probably. But Monday night's performance, fair or not, took some of the sheen off of the Ducks' offense, which was the basis of their high rankings."

Okay - their offense didn't look as solid.

But then you give zero credit for their defense? While Auburn's D definitely did a job, the difference in points, yards and overall result was exceedingly close. The Ducks finished with (in S&P) the 19th-ranked defense in the country - behind teams like Alabama (who gave up 28 to Auburn), Notre Dame (seriously?), Clemson (who gave up 27 to Auburn), and narrowly ahead of Arizona. And that's the final rankings; I remember Oregon's D in the low 30s before this.

I guess it's insane to me to think that a game that was as close as it was, won by the team who had the ball last in regulation by 3 points on what was one of the most fluky plays anyone's ever seen, was considered by you to be a dominant game against something like the 9th ranked team. Furthermore, I don't see how you can justify the S&P ranking by itself of Oregon (which is 16th) - behind South Carolina (who Auburn beat 56-17), Boston College, Illinois...etc.

Heck, Boise State is #1 despite losing to Nevada. Ohio State played a squeaker against the 4th best SEC West team - but is #2. TCU barely beats Wisconsin and is #3. Notre Dame, who got obliterated by Stanford, is #4.

Compare this to FEI's results - the top teams are Auburn, Stanford, Alabama, Oregon, and VTech. Of those which do you quibble with most? Maybe VTech. Then it's LSU, Arkansas, Ohio State, Oklahoma and TCU. Of those the only oddity to me is Arkansas ranked slightly ahead of Ohio State. TCU and Wisconsin are inches apart. The weirdest result of FEI's value is maybe WV - but they're still 10 spots below S&P.

I guess I can see how FEI and S&P make a more accurate, reasonable ranking system - but that's largely in my mind because FEI has very reasonable rankings, and S&P doesn't. S&P by itself looks to be very, very confused here.

Are you going to get a breakdown of where S&P predicted success and failure in the bowl games this season?

To add to this - based on the box score of the NCG Auburn should have won by 3 - and they did. But if you went and used S&P before the game to predict the result, what would it have been? My guess is that Auburn would have been expected to win by double digits, at least.

5
by Tampa Bay Mike (not verified) :: Fri, 01/14/2011 - 3:40pm

Having watched the way the game went - Auburn dominating both lines and steadily moving the ball and Oregon having to pull out all the stops in boom and bust fashion, 10 points in Auburn's favor on average if they played ten times seems a little high but certainly not unreasonable.

There were fluky plays on both sides but Oregon's was necessary just to have the chance to go to overtime and have a chance at winning. Auburn's just gave them a better chance of winning in regulation. In short Oregon's fluke was absolutely critical and Auburn's was not.

As for the box score, strategic effects came into play at the end. It's unlikely that Oregon would've stopped Auburn from scoring a TD from the goal line with time for three or four tries if they had wanted. So the scoring margin on average accounting for that would be more than 3.

7
by cfn_ms :: Fri, 01/14/2011 - 4:02pm

Wait... so big-yardage plays are "fluky" but goal-line stands aren't? How about Auburn's long 3rd down conversions by Newton running?

Let's be honest: there were plenty of "fluky" plays on both sides. Auburn largely dominated LOS, Oregon largely dominated the edges of the field. It felt to me like a fairly even game where Auburn was a bit better, which is the same thing the scoreboard showed.

10
by Tampa Bay Mike (not verified) :: Fri, 01/14/2011 - 5:38pm

Doesn't Auburn's goal line stand get negated by Oregon's goal line stand? That is just assuming that they were equal.

Boom and bust offense, getting dominated at the line of scrimmage = hard work running in the red zone (as we saw twice for Oregon). Once they scored and once they didn't.

12
by cfn_ms :: Fri, 01/14/2011 - 6:01pm

I would say that the odds of getting in from 1st down on the 3 are materially better than 1st down on the 10, but have admittedly not researched it. Probably also fair to argue that the safety Oregon incurred on the next play (but Auburn didn't) was also fairly fluky.

But my larger point was that there were plenty of fluky plays on both sides, as opposed to somehow Oregon catching all the important breaks but Auburn almost none. Saying there was close to a 10 point difference (your original statement) in that game is VERY excessive IMO.

13
by Tampa Bay Mike (not verified) :: Fri, 01/14/2011 - 7:16pm

"as opposed to somehow Oregon catching all the important breaks but Auburn almost none"

Get him! Beat that strawman! Hit him again! Yeah!

"Saying there was close to a 10 point difference (your original statement) in that game is VERY excessive IMO."

Let me repost what I said:

"10 points in Auburn's favor on average if they played ten times seems a little high but certainly not unreasonable"

This is in reference to your speculation about a model predicting a double digit win. I chose 10 as the smallest of the double d's.

To rephrase this, I think that while predicting that Auburn would win by 10 is not unreasonable I also think that it's a little high. If we remove strategic end game concerns about the field goal vs the TD from things Auburn probably wins that game by about 5 points typically. From watching the game I'd say that's probably right, I might bump it up to 6 but that's still really close. They'd usually win by a single score, occasionally by two, and they'd lose occasionally by a single score.

9 - slightly less or essentially the same
8 - a little less
7 - less

...and I realize that this is a conversation without point. Feel free to attribute to me whatever meaning you desire and argue against it.

15
by cfn_ms :: Fri, 01/14/2011 - 7:30pm

Wait, now you're hitting my strawman, Kal made the original 10-pt statement, I was merely responding (perhaps overly so) to your statement that 10 points was "reasonable." Now, if you meant that as "reasonable prediction", then fine, but if you meant that as "reasonable given the flow of the game", that seems excessive to me.

Now in reference to your last statement, I can't believe you're arguing that Vandy would have beaten Oregon by 80. You really have no shame :P

8
by Kal :: Fri, 01/14/2011 - 4:31pm

And yet FEI has Oregon doing significantly better than Auburn in the actual game.

Both teams had fluky plays and weird luck. The Duck first interception was just a tip ball. Both teams had a 4th and goal that they did not make by a very short margin. If Auburn doesn't intercept the ball (once on batted play, once on a guy falling) the game is very different.

The fact is that Oregon did have a number of successful drives - but more importantly Oregon was able to consistently hold Auburn's offense down. While Auburn probably could have scored a TD at the end, the reason that they were there at all was because of an odd running play that had been stopped for a fairly small gain; Auburn had not been able to successfully run up and down the field all game, much less all half (that half they had 3 points, total).

Every fan I talked to -both Oregon and Auburn - saw the game as a game between two very close teams. While Auburn was able to run more effectively Oregon was able to pass effectively even on non-bomb plays. Defensively it was a struggle and a half, and Oregon was able to consistently stop Auburn eventually. Saying that it was a huge blowout but Oregon got lucky belies what FEI, S&P, and what folks' eyes say. It was the closest title game so far.

Really, I think watching it in person was a very different thing than watching it on the telecast where they appeared to be anointing Cam Newton before the game was over. I don't see the dominance that is ascribed by you, and honestly neither does S&P; S&P rates this as a likely 3-point Auburn win after looking at the stats.

11
by Tampa Bay Mike (not verified) :: Fri, 01/14/2011 - 5:42pm

Saying that it was a huge blowout but Oregon got lucky belies what FEI, S&P, and what folks' eyes say. It was the closest title game so far.

You can beat that strawman to death if you like.

9
by Bill Connelly :: Fri, 01/14/2011 - 4:34pm

I don't have the data available to me at this exact moment, but the last time I ran correlations between FO ratings and win%, it was something to the effect of...

S&P+: 0.80-0.81
FEI: 0.82-0.84
F/+: 0.85-0.86

Something to that effect. Like I've mentioned before, my initial impressions are that S&P+ are more effective week-to-week, and FEI is more effective over the course of the season. But the two improve each other, resulting in an F/+ rating that ties very well to overall wins. You don't want your ratings too close to 1.00 correlation by any means -- otherwise what the hell's the point of using anything other than win% to evaluate teams? -- but my happy place would be to get F/+ into the 0.86-0.88 range. We're pretty close. And honestly, when I run the numbers again post-bowl season (when, again, the F/+ picks were pretty dialed in), it might have indeed crept up into that range.

Both systems may have their oddities, and I'll straight-up admit that Oregon and Miami are, to me, the oddest of this year's outliers with either system. But it happens. S&P+ and F/+ were both more 'accurate,' so to speak, than they had been in previous years (you'd have to ask Fremeau about FEI), and I hope to be able to say the same thing a year from now.

Okay - their offense didn't look as solid. But then you give zero credit for their defense? While Auburn's D definitely did a job, the difference in points, yards and overall result was exceedingly close.

The rankings themselves give plenty of credit to the defense. I was trying to stay abbreviated in my thoughts on the game, since it was already well-covered, but yes, holding Auburn to just a good overall offensive day instead of a great one, was quite beneficial to their overall ratings, which did go up despite an average offensive performance (in terms of S&P). My intention with what I did write about the game was to suggest that the way the game played out, with Auburn holding a much higher leverage rate and S&P, shows that Auburn had a much nicer margin for error. They almost blew it all, and Oregon was basically one more big play away from winning the title, but Auburn gave themselves more chances for success, and through a combination of their own mistakes and Oregon's good plays, they almost lost regardless.

(This goes to show that I should never abbreviate my throughts. If you look at the average length of Varsity Numbers, it should be clear that I'm not very good at it.)

To add to this - based on the box score of the NCG Auburn should have won by 3 - and they did. But if you went and used S&P before the game to predict the result, what would it have been? My guess is that Auburn would have been expected to win by double digits, at least.

The official F/+ pick for this game was Auburn by 11, so I think we can discern that S&P+ would have been higher than that, yes. If you'd used F/+ to pick, straight up, every Oregon game this year, you'd have gone 12-1 (missing only the Stanford game), so in a broad sense, it's not out of the realm of possibility that our (okay, my) data was too horribly off on them. But it was clearly not dialed in too much. F/+ picks went only 6-7 ATS with Oregon this year, which was the worst of any team in the F/+ Top 15. F/+ had a really good read on the major conferences and top teams overall -- our ATS win % for the teams ranked in the F/+ Top 40 was 57%, which I think is outstanding. But as I admitted many, many times this season, the numbers (mine in particular) weren't reading Oregon very well for one reason or another. I couldn't possibly admit that more than I did over the course of the season. It was driving me nuts.

14
by Kal :: Fri, 01/14/2011 - 7:21pm

I guess that looking at the numbers it's not just about Oregon; while that's one of the extreme outliers by any means, it's not the sole one. The two versions of the top 25 look very little like each other. Boise State is #1 in S&P, #11 in FEI. Ohio State's #8 in FEI, #3 in S&P.

And then we get into weirdness.

Stanford is ranked below TCU. Whuh?
South Carolina is ranked below Arkansas.
Notre Dame is right there with Wisconsin and very close to Stanford.
Oregon is below Miami and barely ahead of Utah.
Alabama is ahead of Auburn.
LSU is below Texas A&M.

I understand if teams are close to each other and the ranking shift just a bit, and I understand there aren't that many differences between some of the top-ranked teams. I understand that just because Boise lost doesn't mean they should be catapulted into oblivion. But at the same time, a system that sees LSU obliterate TA&M and ranks them lower has something empirically wrong. A system that has Oregon so low despite crushing teams like Stanford has something empirically wrong. And while F+ might be a fairly decent indicator, that isn't to say that it can't be better in some ways (or that S&P can't be better in some ways). A system that sees the performance that Stanford put on VTech and then compares it to what TCU did to Wisconsin - and Stanford loses out - is somehow flawed.

I don't know what the flaw is. I think part of it is issues on the boundaries, where you rate games as being close vs. blowouts (and what that does to the numbers). I think you'll have to end up refining that, but I suspect you're already on that path after reading the Georgia game description.

I think part may be opponent adjustments as alluded to in the above; while there's an indicator (albeit a small one) that in the NFL teams that stomp weak opposition are better than teams that squeak by good teams, I'm not sure that correlation applies in the NCAA. I can think of teams like Ohio State that prove that completely wrong, for instance, and Auburn this year is another example.

I suspect that home vs. away is also going to be more enlightening. I don't believe FEI deals with this but I think seeing some of the outliers in Oregon's schedule a lot of it is simply explained by playing okay teams on the road. That isn't to say that it's not a fair indicator, but it's there.

Finally, I wish to apologize; I didn't mean to come off as hugely combative as I did, and I'm sorry. It would be neat to see a postmortem of both FEI and S&P, where they did better than each other and where they did worse and see if there are correlative values. It would be interesting to normalize some of the values too, so that S&P doesn't blow away FEI rankings. But first and foremost you and Brian are doing a great job measuring something that is very difficult, statistically, to quantify, and I applaud you for doing it and for being honest about both your wins and your losses in doing so, and keeping the dialog open.

16
by cfn_ms :: Fri, 01/14/2011 - 7:50pm

I'm presuming you're talking about S&P with your list:
(my comments below)

Stanford is ranked below TCU. Whuh?
(I don't necessarily have a huge problem w/ this one, given TCU being undefeated. For the the "Whuh" is Stanford below Boise and Ohio St, especially Ohio St)

South Carolina is ranked below Arkansas.
(I'm guessing you mean above which S&P has. Below is where they should be)

Notre Dame is right there with Wisconsin and very close to Stanford.
(yeah ND is one of those outliers that kind of fell under the radar a bit, but deserves attention as well. Intuitively, team #11 shouldn't get waxed by Navy, lose at home to Michigan or lose at home to Tulsa, much less all three. Dominant wins vs Utah and Miami were nice, but IMO not enough to overcome the three crap games)

Oregon is below Miami and barely ahead of Utah.
(the Oregon one has been beaten into the ground, but I'll take a few more whacks for kicks. There isn't a single team within 5 spots in S&P's ratings whose resume is remotely comparable to Oregon. Now that the season is over and Oregon's rating is still wacky, it's fair to pile on some more on that one)

Alabama is ahead of Auburn.
(I have a bit less of a problem w/ this than some others. Auburn's resume has some meaningful warts, and Bama HAS been a really good team. I disagree w/ the ordering, but IMO more defensible than some others here)

LSU is below Texas A&M.
(that one strikes me as strange as well, though at least they're rated close. S&P seems to really dislike LSU for some reason, don't recall the details there)

I think the flaws are, in some order:

Schedule adjustments - I'm pretty sure this is the biggest source of S&P error. When your schedule adjustments are x% avg opp performance and y% avg opp-opp performance, even in a highly connected environment (like the NFL) it's a fairly questionable approach. In a very poorly connected environment like CFB, it's simply wrong (moreso since it appears S&P ignores HFA). I'm guessing Bill fixes this in the offseason.

Overly weighting certain situations at the expense of others - Bill has actually talked about this a bit, and I agree that his approach is generally defensible but hurts certain specific types of teams like Oregon. I'm guessing it makes sense to investigate lightening the adjustments a bit (arbitrary example: 10-25% less adjustment as a starting point) and see how it affects predictions. As long as it isn't hurting the system, I'd think a tweak like that makes sense. There's also been some discussion about tweaking the "blowout" rule so that if the game ever becomes close again (i.e. big lead builds up then goes away) all plays up to that point get full weight, which intuitively makes a fair amount of sense.

FWIW, I also really appreciate the time and effort both Bill and Brian put into this. Modeling sports events is HARD, and a huge part of the process is (unfortunately) absorbing body-blows from people who disagree with what you're doing and/or your results. Despite my harping on points of disagreement, I think that both models are valuable and useful, and I enjoy seeing them both. Thanks to both of you for doing such a good and (far too) thankless job.

18
by Bill Connelly :: Fri, 01/14/2011 - 7:54pm

Out of curiosity a few weeks ago, I looked into what would happen if I cut the schedule adjustments in half. That's obviously drastic, but again, I was curious. Oregon's ranking made a lot more sense ... and the overall accuracy, both in terms of week-to-week picks and overall ties to win%, sank dramatically. I'm always tinkering, but that was a tinker that didn't take.

20
by cfn_ms :: Fri, 01/14/2011 - 8:19pm

I think the big problem isn't with the size of your schedule adjustments (which I would guess are fine) but the structure of them. S&P looks at Oregon as having had a mediocre to poor schedule; post-bowl I would look at them as having at the very least a top 20 schedule, and (IMO very) arguably a top 10 schedule.

So in my mind, the key point isn't "schedule shouldn't matter so much" (I think it should, a point reinforced by your experiment [in fact, were you to investigate giving more weight to schedule you might find that it should actually matter MORE] ) but instead "Oregon should be recognized as having had a very tough schedule" (which I think they have - 6 of 12 regular season games on the road, 10 of 12 regular season games against AQ's, playing in a league that was good to [IMO] very good, plus the NCG against Auburn).

17
by Bill Connelly :: Fri, 01/14/2011 - 7:52pm

A system that sees the performance that Stanford put on VTech and then compares it to what TCU did to Wisconsin - and Stanford loses out - is somehow flawed.

We're all entitled to our opinions, so I'll just say that my opinion of this differs significantly. I think TCU is a very, very, very, very good team that could compete with anybody. Boise State too. I'll defend those rankings strongly.

In general, Brian and I get different rankings because we measure different things. We work in completely different methods to get different results ... both of which tie relatively well (and getting better) to win%. The fact that we DO get differing results tells you what kind of crazy world lives below the surface when you look past simple points scored and win-loss records. There's a lot of data there, and we're really only beginning to figure out what we can do with it. I've already made a couple of S&P+ tweaks, and I'll make more whenever it makes sense to do so (and it takes more than one odd year to force a change -- it has to make sense for ALL the data I have, not just 2010).

Home-road is one of the millions of things I want to look into at some point.

(I looked completely past the combative tone. If we replaced "Oregon" with "Missouri" in all of these examples, I'd be sensitive in all the same ways. It comes with being a fan. As somebody who has always rooted for the non-traditional powers as much as possible, I'd have really preferred that Alabama or Ohio State or a traditional power be the subject of this weird outlier.)

19
by Kal :: Fri, 01/14/2011 - 8:19pm

Actually that's the funny thing, Bill - I don't see myself as a homer here. I have way more problem with Boise and OSU having such a high rank compared to where Oregon ranks. And I have a really hard time with ND being so good. I can actually understand completely why Oregon ranks the way it does in the system; while I think it's wrong, I get it. They played a bad ASU team close and probably should have lost were it not for lucky turnovers. They played a truly horrendous WSU game much closer than it should have been. They played a meh Cal team far closer than it should have been. All of these things in the methodology of 'stomps uber alles' point to DANGER DANGER OVERRATED TEAM.

I get that, I really do. It makes sense.

Similarly, if you favor teams being consistent and early in winning against bad teams(as honestly you likely should) Oregon looks bad; turning on the jets against Stanford, Washington, USC, and even Oregon State all don't look as impressive as what they did against UCLA. I understand that methodology and don't decry it.

What I don't get as far as Oregon's concerned is why games like Stanford weren't higher 'wins' for Oregon in the system. For every game where Oregon looked meh against non-quality opponents there was a game where Oregon simply looked like the best team ever. That seems to have a huge ripple effect, where all of a sudden Stanford looks meh because they lost to Oregon and thus all of the teams Stanford played take a hit.

Mostly, I really don't get things like Notre Dame or Miami. LSU I kinda get; you can't play Tennessee to a 'loss' (as far as stats were concerned) and be considered a good team, and I'd even agree there. But Notre Dame just didn't look good.

21
by cfn_ms :: Sat, 01/15/2011 - 1:00am

http://cfn.scout.com/2/1039643.html

As always, the dumbest "computer" system around. Fiu's "System" makes Billingsley look reasonable, which pretty much says it all. I love how they say "Forgetting the opinion side of the equation" as if they didn't throw a bunch of random numbers together and use their own arbitrary opinions to decide what does and doesn't matter. Mock away, I'm sure not going to defend it.

22
by Alexander :: Sat, 01/15/2011 - 2:51am

The top of the list doesn't bother me at all. I think it is likely that the Stanford-Oregon game would have gone the other way if it was not in Eugene.

Sadly, I feel that perhaps my Pitt Panthers are the true outlier in this system. They had some truly inexcusable results this year.

23
by Bill Connelly :: Sat, 01/15/2011 - 12:39pm

Pitt is proof that I don't have the opponent adjustments set up quite correctly. They were rewarded a bit too much for a tough non-con schedule. Miami, too. For some reason, I don't really have a problem with where Notre Dame ended up...probably a bit too high, but I think they turned into a helluva team late in the year...and weren't terrible early on.

24
by cfn_ms :: Sat, 01/15/2011 - 7:37pm

Is S&P set up to over-weight more recent games? If that's so, then I could consider it defensible (though I thought S&P was time-neutral).

If not, though, they had three inexcusible games for a team ranked #11: losing at home to Tulsa, losing at home to Michigan, and getting waxed against Navy. Losing at Michigan St and getting waxed at home by Stanford couldn't have helped the rating either. The Utah and Miami wins were quite nice, but unless you're really strongly over-weighting recent results, they're not close to enough to overcome those losses.

If I had to guess, I'd suspect you're overrating ND's schedule. My numbers have them 29th. Do you have them top 10 or the like? That might be playing a part as well.

25
by zlionsfan :: Sat, 01/15/2011 - 10:41pm

He coached them to a loss in the MAC championship (to Turner Gill's 7-5 Buffalo team), then left for San Diego State prior to the Cardinals' bowl loss.

I'm not sure I agree that it's unfair to look at his career record. Five of the six coaches who preceded Hoke at Ball State had winning records; the exception, Bill Lynch, demonstrated that he could struggle farther south as well. (If I-69 ever extends through Bloomington, they can name that stretch the Bill Lynch Memorial Highway ... the stretch of road on which his career was buried.) Lynch actually had Ball State at .500 the two season before he was let go (and 5-6 the two before that), but it took Hoke five seasons to break .500.

He did have an excellent season there, but there's nothing to suggest it was anything more than decent talent plus a favorable schedule, and even at that, they couldn't even win the conference title and lost their bowl game as well. Stan Parrish seems to have removed our ability to assess whether or not Hoke actually built a solid program at Ball State ... if he'd at least broken .500 once, it'd be easier to argue that Hoke left something behind. (Compare this to Brian Kelly and Butch Jones at Central Michigan.)

The Aztecs' turnaround is definitely a good thing, but again, after a single season, it's hard to be sure whether it was significant progress or just good timing.

Hoke may be a good fit at Michigan - one thing's for sure, he won't inspire cries of "YOU AIN'T FROM AROUND HERE" that his predecessor did - but his resume raises just as many questions as it answers. Impressive resumes do not always lead to impressive results - again, a reference to Rodriguez - but given the way in which Brandon communicated his intentions, I think a good portion of the fan base was expecting something different.