This week's DVOA commentary is all about worsts. Come find out where Washington stands among the worst special teams in DVOA history, whether San Diego has the biggest gap between offense and defense, and whether Baltimore or Jacksonville has the worst running game we've ever tracked.
27 Nov 2009
by Bill Connelly
How in the world can OU still rank this high? Is replacing Charlie Weis really a safe bet? Are stupid, immature rivalries as good as I make them out to be? Could the numbers possibly hate Rutgers any more than they already do? Answers to all these questions and more below!
(That was my best soap opera lede. How was it?)
We focus this week's VN Box Score on a source of annoyance for yours truly.
Sub-header: "And Yet OU is Still Ranked in the S&P+ Top Ten."
Before we try to investigate exactly how Oklahoma has managed to stay in the S&P+ Top 10 despite a 6-5 record (as I mentioned yesterday, I'm growing considerably bitter about how OU is always getting in my way, as both a fan and a numbers nerd), let's take a look back at last week's game in Lubbock, which probably exorcised at least a few demons for Tech fans still licking their wounds from last year's trip to Norman.
|Field Position %||29.0%||47.7%|
|Close Success Rate||23.8%||46.8%|
|Close Success Rate||23.5%||50.0%|
|Close Success Rate||24.0%||45.5%|
|SD/PD Sack Rate
||0.0% / 17.7%||2.6% / 6.3%|
|Turnover Pts Margin
|1st Down S&P||0.632||0.917|
|2nd Down S&P||0.301||0.886|
|3rd Down S&P||0.589||0.714|
|Projected Pt. Margin
|Actual Pt. Margin
When you give up 549 yards of total offense, your defense clearly did something wrong. Oklahoma entered last week's game ranked third in the country in Defensive S&P+; and, in the past, they have defended spread offenses as well as anybody in the country. In the end, this loss falls once again onto the shoulders of a beleaguered Sooners offense.
While this game was within the confines of being considered "close" (it was no longer "close" when Taylor Potts found Alex Torres for a touchdown to put Tech up 24-6 with 10:38 left in the third quarter), OU's offense fell apart while the defense fought to keep things close. An interception and two field goals gave the Sooners a 6-3 lead early in the second quarter, but here were OU's next seven drives (as evidenced by positively putrid Q2 and Q3 S&P figures):
By the time Landry Jones found Ryan Broyles for a 51-yard touchdown midway through the fourth quarter (which led to one of the most egregiously stupid taunting penalties you'll ever see), the game had gotten out of hand, and OU had learned the same lesson a lot of teams learn playing Tech. No matter how good your defense is, if your offense is producing nothing and losing the field position battle, you are eventually going to wilt and give up points.
Being that a good portion of Tech's yardage came after the game left "close" status, OU's defense wasn't punished and stayed at third in this week's S&P+ defensive rankings. Combined with the fact that OU's offense wasn't ranked very high to begin with, their lack of production didn't drop them very far. In other words, despite the fact that the Sooners lost by 28 to a team ranked 27th entering last week, they only fell from fifth to seventh in the S&P+ rankings. And I can't tell you how annoying I find that.
As we worked our way through some small sample size issues earlier in the season, we saw some interesting rankings pop up from week to week. Teams with .500 records found themselves ranked quite high simply because of who they played. But as they have slowly weeded themselves out (relatively speaking -- Arkansas, Tennessee, and South Carolina still rank quite high, but with their losses to mostly top teams, this is understandable), OU has remained higher than they appear to deserve. So let's look into why that is.
In all, a vast majority of S&P+ categories I collect are pretty well-correlated to win percentages, which makes sense. However, nine categories have a direct correlation of 0.6 or more. Here are those ten categories, with Oklahoma's rank in each category:
(Oklahoma aside, it is interesting to see which categories seem to make the most difference in terms of wins and losses this year -- rush defense, standard downs offense, pass offense. Plus, PPP+, which is only one component of overall S&P+, seems to be more important than S&P+. The ability to make and prevent big plays is more important than efficiency this year; that has been the case for all three years of this data, but the difference is a little stronger this year. All that said, all of the correlations are pretty strong, so we probably shouldn't read too much into it.)
Aside from big-play offensive ability on standard downs, OU ranks well where it counts the most. In other words, they are good at almost all of the things the best teams have done this year, except winning games. (They haven't been quite as good as that.) And the areas in which they have struggled this year -- rushing offense, for one -- are not as closely tied to wins and losses, so maybe they are not punished as much for struggling there.
This brings up a good point, however. When analyzing play-by-play data, how important is it that there really is not a direct way to take into account teams who simply "know how to win," whatever that exactly means? Is S&P+ meant to reward teams who are best at pure play-by-play performance, or reward teams who have been best at winning games, when the two lists are far from identical? Let's see what would happen if "ability to win" was taken into account.
Regardless of why it is happening, OU is still ranked high despite five losses. But what would happen if we applied a loss adjustment like the BCS formulas used to do? How would the rankings look if we subtracted 10 S&P+ points for each loss?
|Loss-Adjusted S&P+ Top 25|
|5||Boise State (11-0)||239.9||11||0||239.9|
|7||Penn State (10-2)||251.2||6||2||231.2|
|9||Virginia Tech (8-3)||253.4||5||3||223.4|
|10||Ohio State (10-2)||241.5||10||2||221.5|
|Loss-Adjusted S&P+ Top 25|
|11||Georgia Tech (10-1)||223.5||25||1||213.5|
|19||Oklahoma State (9-2)||222.5||26||2||202.5|
|Loss-Adjusted S&P+ Top 25|
|24||Ole Miss (8-3)||225.8||23||3||195.8|
|25||Texas Tech (7-4)||231.5||16||4||191.5|
You've still got two- and three-loss teams ahead of one-loss teams like Georgia Tech and Pittsburgh, but, in general, this seems to be a solid mix of both evaluation of play and reward for wins. If this ranking system were to be used in the actual BCS formulas (and hey, why shouldn't FEI and S&P+ have a seat at the table?), it wouldn't be too bad an idea to apply an adjustment like this. In the end, winning does matter, no matter how teams figure out ways to do it (or not do it).
Switching gears for a moment, let's talk about Charlie Weis. It is all but a foregone conclusion that Weis' days in South Bend are coming to a close. He has basically the same record as his two predecessors, Bob Davie and Tyrone Willingham, did, and since he went out of the way to call that level of achievement unacceptable when he took the job, he's between a rock and a hard place. Even though we know the odds are good that Weis will be removed from his duties, my question today is, Should he? If the last two coaches have achieved at precisely the same level as Weis, then what are the odds that a new coach wouldn't do exactly the same job? Should Notre Dame actually keep the same guy around for more than a handful of years to see if a long-term building project is the best idea?
Two things are relatively certain when discussing Weis' tenure in South Bend:
1. The program fell off the face of the planet in 2007 when Brady Quinn, Jeff Samardzija and company left town
2. They have shown steady improvement each of the last two years.
The improvement is hard to see when just looking at wins and losses -- the records show relative stagnation between 2008 and 2009 -- but the statistical improvement is there. And since the relative strength of Notre Dame's schedule seems to shift from very hard to very easy in any given year, a schedule-adjusted approach is probably the best way to measure per-season progression.
Lucky for us, we have a nice schedule-adjusted tool at our fingertips! Using S&P+ ratings for the last three years, and Estimated S&P+ ratings (discussed a couple of weeks ago -- scroll down to "The Andrews Effect") for the time before that, let's take a look at how Willingham's teams, Davie's teams, and even Lou Holtz's final few teams performed. Below is the following data:
Season (Record): Overall S&P+ ranking (Offensive Ranking, Defensive Ranking)
Lou Holtz's Last Five Years (44-13-2)
1992 (10-1-1): 6th (4th offense, 8th defense)
1993 (11-1): 5th (4th offense, 10th defense)
1994 (6-5-1): 18th (23rd offense, 18th defense)
1995 (9-3): 5th (5th offense, 10th defense)
1996 (8-3): 7th (7th offense, 16th defense)
Average Ranks: 8.2 overall, 8.6 offense, 12.4 defense
Bob Davie (35-25)
1997 (7-6): 24th (36th offense, 19th defense)
1998 (9-3): 31st (47th offense, 15th defense)
1999 (5-7): 29th (25th offense, 36th defense)
2000 (9-3): 27th (32nd offense, 21st defense)
2001 (5-6): 33rd (60th offense, 19th defense)
Average Ranks: 24.0 overall, 40.0 offense, 22.0 defense
Tyrone Willingham (21-15)
2002 (10-3): 13th (58th offense, 6th defense)
2003 (5-7): 42nd (54th offense, 29th defense)
2004 (6-6): 29th (30th offense, 32nd defense)
Average Ranks: 28.0 overall, 47.3 offense, 22.3 defense
Charlie Weis (35-26)
2005 (9-3): 11th (4th offense, 29th defense)
2006 (10-3): 23rd (6th offense, 49th defense)
2007 (3-9): 90th (118th offense, 46th defense)
2008 (7-6): 54th (63rd offense, 60th defense)
2009 (6-5): 41st (29th offense, 52nd defense)
Average Ranks: 43.8 overall, 44.0 offense, 47.2 defense
Average Ranks w/o 2007: 32.3 overall, 25.5 offense, 47.5 defense
What is most impressive about this data is that, aside from an offensive upswing from 2004-06 (followed by a dramatic downturn in 2007) and a defensive upswing from 2000-02, this team's performance really has been relatively consistent under each of the last three coaches. Offense ranked in the 40s, defense ranked in the 20s. In Weis' last couple of years, the strengths have flipped from defense to offense (If Weis does manage to keep his job another year, the Irish need some significant defensive improvement), but as a whole Notre Dame has ranked between 23rd and 42nd in nine of the last 13 seasons. What has made as much of a difference as anything has been the varying strength of schedule. Easy schedules have allowed for higher win totals and increased expectations and ridiculously tough schedules have taken them back down. Willingham's final team showed solid improvement, but the schedule dictated only a one-game increase in the win total. Meanwhile, Davie's 5-7 team in 1999 (which faced just two teams with losing records) was actually a bit stronger than their 9-3 team in 1998 (which faced seven).
So what does this mean? Depends on your point of view. Maybe they should have never fired Davie. Maybe the fact that the Irish rank just 41st in their fifth season is clear proof that Weis should be gone. Personally, I always err on the side of giving a coach more time. There's no question that 2007 was a disaster, and however Weis handled the massive turnover in personnel from 2006 to 2007, he erred significantly. But they improved in 2008 and have improved again in 2009. The personnel could dictate more improvement in 2010 no matter who the coach is, and it might as well be Weis leading (and potentially winning with) the players he recruited.
Then again, if Jimmy Clausen and Golden Tate bolt for the pros, he might be in for a worse season in 2010 and now might be a pretty good time to break in a new coach. This is why nobody pays me to make decisions. I have no idea what would work best for the future of the Irish, but I know that is far from cut-and-dried that Weis deserves to go. Good luck to you, Jack Swarbrick.
Few teams made dramatic rises or falls this week, at least outside of New Jersey and Oregon.
Texas Tech (11 spots, from 27th to 16th). As I have mentioned before, one of the entertaining things about tracking the week-to-week moves in the S&P+ rankings is how when there is a surprising result -- like Texas Tech's blowout of Oklahoma (the blowout was the surprise, not a Tech victory) -- it will sometimes impact one team's rankings a lot more than it impacts the other's. Whereas Oklahoma barely fell for getting blown out, Tech vaulted to 16th. They have been impressive in rising that high despite the strength-of-schedule punishment for playing two FCS teams and the still-baffling blowout loss to Texas A&M.
Northwestern (11 spots, from 89th to 78th). The ratings have not been very impressed with Northwestern this year; for the most part, they have played the part of a poor man's Iowa, winning games while not looking very good. The difference is, Northwestern has done it against a much easier schedule, and their ratings have not been very impressive. Upsetting Wisconsin, however, gave them a nice little end-of-season boost. And regardless of their rankings, Pat Fitzgerald deserves some major credit for piecing together a decent team after having to break in new personnel in the offseason and looking less than stellar in September.
Arizona (9 spots, from 52nd to 43rd). The ratings have been slow to warm on Arizona, but even though the Wildcats blew a late lead and lost in overtime to Oregon, building that lead in the first place gave them a boost.
Syracuse (9 spots, from 63rd to 54th). Syracuse was clearly a few notches below quite a few of their opponents in terms of athleticism (they were unable to stay too competitive against the likes of Penn State, West Virginia, and Cincinnati), but credit new coach Doug Marrone for piecing together a team that has taken advantage of opportunities to win. Against less athletic teams, they have stayed competitive, beating Northwestern, almost beating Minnesota and Louisville, and whipping freshly ranked Rutgers last week in an Upset Watch Special.
Other notable rises: Utah State (96th to 82nd), Buffalo (93rd to 80th), East Carolina (76th to 65th), Troy (44th to 34th), Vanderbilt (98th to 89th).
Rutgers (20 spots, from 80th to 100th). At kickoff last week, Syracuse ranked 63rd and Rutgers ranked 80th, but when the Orange beat the Scarlet Knights as predicted, it made a tremendous impact on both teams' rankings. I mentioned last week that Rutgers' horrid strength of schedule was holding them back; you know what else holds a team back? Getting whooped by Syracuse.
Oregon State (19 spots, from 36th to 55th). Here's where the strength-of-schedule adjustment is entertaining. When you play a really weak team, your ranking won't automatically drop, but you better put them away as quickly as possible. Oregon State led lowly Washington State just 21-10 at halftime, and even though they eventually put the Cougars away (final: 42-10), their ratings suffered quite a bit. The bar for beating Wazzu is set at "You should be up 21 after the first quarter," and the Beavers certainly missed that mark.
Connecticut (9 spots, from 38th to 47th). Further insult to injury for hurting Notre Dame fans: Connecticut beats the Irish in South Bend and sees their ranking fall nine spots. UConn's offense did relatively well when all was said and done, but Notre Dame's defense is poor enough that UConn needed to do more to avoid slippage.
Baylor (8 spots, from 48th to 56th). You can't lose to a schizophrenic Texas A&M team by 35 and keep your spot in the Top 50.
Other notable falls: UTEP (85th to 97th), Air Force (28th to 39th), Toledo (88th to 98th), Hawaii (83rd to 91st), Fresno State (50th to 58th).
To fate, for giving us not only the typical conference championship games next weekend -- Big 12, SEC, ACC, Conference USA, MAC -- but also two more big-time, de-facto championship games as well. The Pac-10 title will be decided next Saturday in Eugene, when Oregon and Oregon State face off in this year's Civil War. Meanwhile, no matter who wins tonight's Pittsburgh-WVU brawl, next week's Pittsburgh-Cincinnati game will determine the Big East champion. You've got five of six BCS conferences determining their champion on the same day (meanwhile, the Big Ten's regular season ended, what, six weeks ago? Six years ago?). Needless to say, that's one fantastic day of television. In just one day, a relatively lackluster season low on on-the-field drama (at least at the top of the polls) could become remarkable. Or not.
Call this a rant revisited: Tell me again why we still have award watch lists? Heading into this weekend, here were the 2009 statistics for four receivers:
Here are the November stats for these four receivers:
Receivers A, C, and D were on the preseason Biletnikoff watch list, while Receiver B was not. Receivers A, C, and D, therefore, are the three Biletnikoff finalists. Receiver B, the most statistically successful BCS conference receiver in the country, was not, and all because of the concept of watch lists, which became outdated the moment the Internet reached prominence.
Receiver A: Freddie Barnes, Bowling Green
Receiver B: Danario Alexander, Missouri
Receiver C: Jordan Shipley, Texas
Receiver D: Golden Tate, Notre Dame
Sure, Yankees and Red Sox fans probably don't like each other very much. Nor do Cowboys and Redskins fans (or Eagles fans and anybody else). But only in college football can you truly embrace the hatred of your neighbor simply because of the colors and logos he chooses to wear. Texas and Texas A&M fans, Georgia and Georgia Tech fans, Auburn and Alabama fans, Missouri and Kansas fans ... they work together, their kids go to the same day care, and they probably enjoy each other's company from time to time, but like Crips and Bloods with 90 percent less violence, when the school pride comes out, so does the good, clean, old-fashioned hate. In theory, allowing yourself to embrace immaturity should clear up the rest of your life for acting like an adult (sadly, however, that's not the way the world always works, but you get the idea). So in that way, thank you, college football, for giving us all this outlet. And M-I-Z ... F***-K-U.
We'll just say this was the most predictable Random Playlist theme of all-time.
"Cold Turkey" by John Lennon
"Jive Turkey" by Ohio Players
"Stuffy Turkey" by Thelonious Monk
"Thank God for Kids" by Oak Ridge Boys
"Thank You" by Led Zeppelin
"Thank You (Falletin Me Be Mice Elf Agin)" by Sly & the Family Stone
"Thank You Too" by My Morning Jacket
"Thanks" Chatham County Line
"The Thanks I Get" by Wilco
"Thanks That Was Fun" by Barenaked Ladies
It was a 2-2 week for Upset Watch last week, as Texas Tech most certainly beat Oklahoma and all my non-committal talk about the "Syracuse > Rutgers" pick ended in a lovely reverse jinx. We almost went 3-1, but Les Miles' well-documented brain farts at the end of the Ole Miss-LSU game prevented us from potentially nailing that one with a game-winning field goal.
Louisville over Rutgers. Spread: Louisville +3 | S&P+ Projection: Louisville by 11.4.
This pick was predictable, as all the factors that applied to last week's "Syracuse over Rutgers" pick apply here too. The numbers simply do not like Rutgers, and it's Senior Day at Louisville.
South Carolina over Clemson. Spread: S. Caro. +3 | S&P+ Projection: S. Caro. by 4.5.
The S&P+ ratings have liked Clemson for a while, first picking them to make the ACC championship game despite early struggles, then (currently) picking them to beat Georgia Tech and take the ACC title. That said, South Carolina is a rock solid football team victimized by a brutal schedule (four of their last five opponents rank in the S&P+ Top 12), and they could very easily knock off the Tigers in Columbia.
Arkansas over LSU. Spread: Arkansas +3.5 | S&P+ Projection: Arkansas by 1.3.
Mississippi State over Ole Miss. Spread: Ole Miss -8 | S&P+ Projection: Ole Miss by 0.3.
While the SEC's two best teams aren't expected to struggle with their respective rivals (Florida should handle Florida State easily, and Alabama shouldn't struggle too much with Auburn), other SEC rivalry games could be barn-burners. Arkansas is an S&P+ darling, enough that they are projected to win in Baton Rouge against an LSU team still trying to figure out what happened in the last minute last week in Oxford; meanwhile, Mississippi State hosts an Ole Miss team that is good, but not necessarily great.
This Thanksgiving, I am beyond thankful for the opportunity to write for Football Outsiders and interact with a lot of other stats geeks who are as (or more) interested in the impact of numbers on football as I am. I want to wish you all a happy holiday season.
12 comments, Last at 29 Nov 2009, 3:04pm by captain anonymous