Comparing the final Football Outsiders F/+ projections to the AP poll and conventional wisdom.
02 Sep 2011
by Bill Connelly
I spent most of my summer biting off almost more than I could chew: a 120-team, 300,000-word project in which I wrote statistical profiles and 2011 previews for each FBS program for SB Nation. It was a rewarding experience -- if nothing else, it proved once and for all that I am incapable of burning out on writing about college football -- but it would be a shame if I instantly put the project behind me and looked toward the season at hand.
Below are ten lessons I learned from this offseason of writing.
I have always thought there were two primary uses for college football stats: evaluation and prediction. We have made progress (and still have progress to make) in both, but there is a third avenue that I began to find extremely worthwhile during these profiles: personality. We are working from the most basic of play-by-play data, but we can certainly do quite a bit with it.
Here are a couple of examples:
From this, we quickly learn the following: A) Oklahoma threw a disproportionally high amount of the time on passing downs under then-offensive coordinator Kevin Wilson, making Landry Jones' passing line (7.7 yards per pass, 66-percent completion rate, 38 TD, 12 INT) even more impressive. B) Their pace was, as was typical for Wilson's OU offenses, astronomical. C) With the variability measure, we see that they were reasonably similar in their tendencies, no matter the field position or scoring margin.
Combine that with what we knew from Oklahoma's other stats -- 52nd in Rushing S&P+, 90th in Adj. Line Yards, 20th in Passing Downs S&P+ -- and we get a pretty clear picture. Oklahoma wanted to establish the run on standard downs, but they weren't very good at it (thanks mostly to their offensive line) and relied on Jones, Ryan Broyles and a strong passing game to bail them out. It worked most of the time, but one can see why this formula failed to produce enough on the road at Missouri and Texas A&M.
With this chart, we get the impression that opponents knew they couldn't run on South Carolina, so they didn't really try. Any success opponents thought they could have was going to come through the air. We also see that whatever pass rush the Gamecocks generated, it didn't necessarily come on passing downs, and that S.C. went after the ball (forced fumbles, passes defensed) quite well.
The Gamecocks' stats back this up. They ranked third in Rushing S&P+ and just 20th in Passing S&P+ -- they were good against the aerial game, but obviously not as good. End Devin Taylor broke up eight passes, and four different returning defensive backs defensed at least five passes.
The footprint is a nice complement to the rankings we have developed. It gives us a certain amount of the "how" behind the numbers, and I hope it continues to develop as an interesting tool.
Toward the start of each profile, I charted out each team's Adj. Scores to look for interesting trends. I went into detail about Adj. Score here. (NOTE: In that column, I was looking just at non-garbage time plays. After some consideration, I changed that to look at all plays, just to give a more accurate representation of all 60 minutes. For this measure, that seemed like the right move.)
For instance, we see that Oregon's defense was incredible early on, then began to show cracks just as their offense began to surge. Meanwhile, after they got drubbed by Oregon, Stanford was perhaps the best team in the country, and it wasn't really that close.
Really, the biggest issue right now with Adj. Score is that I am not sure I have explained it well enough. Here's another attempt: Adj. Score attempts to tell you how a team would have done against a perfectly average opponent, getting a perfectly average number of breaks, each week. If two very good teams play each other, the odds are good that they will both end up with an Adjusted Win because they both played good enough to beat an average team. Meanwhile, if two bad teams play poorly against each other, one will get the win, but neither are likely to get an Adjusted Win because they couldn't beat an average opponent. It is a great measure for observing trends, and that's how I used it in these profiles.
USC is the best team in the division, and they are ineligible for the title. Arizona is replacing all five starters on the offensive line. Colorado and UCLA are nowhere near ready to challenge for a division title. That leaves Utah and Arizona State. Look at the schedules for these two teams and Arizona.
Arizona: at Arizona State, Utah, at USC, Oregon, Stanford
Arizona State: Arizona, at Utah, USC, at Oregon, no Stanford
Utah: at Arizona, Arizona State, at USC, no Oregon, no Stanford
The Utes get ASU at home and avoid both Oregon and Stanford. This is why they are projected to tie for the Pac-12 South title in Football Outsiders Almanac 2011 despite a lower projected ranking. When you throw in a simply incredible number offseason injuries for the Sun Devils, it feels right to give a huge edge to the conference newbies, the Utes.
(For now, we'll ignore how unimpressive they looked last night.)
Here's one of my favorite charts from the series. It is a look at two different teams and their F/+ performances over the last four years.
Team A is the Kentucky Wildcats. Team B? The Michigan Wolverines. Not to go cliche on you, but the mighty, they have fallen. But they made one hell of an offseason hire. Brady Hoke has turned around two mid-major programs -- it took him just two years to turn around a dour San Diego State squad -- and his new defensive coordinator, Greg Mattison, inherits a defense that almost literally cannot be worse than they were last year. If their experiment with Denard Robinson, Pro-Style Quarterback, does not end in disaster, then a simply mediocre defense will make the Wolverines strong in a hurry. They avoid Wisconsin and Penn State from the other division, and they get Nebraska at home. That's enough for me to go all-in on the Michigan bandwagon. The Legends Division could potentially become a four-team logjam between Michigan, Nebraska, Iowa and Michigan State, and the Wolverines could very well be a major sleeper in the race.
Oklahoma State is getting a lot of pretty deserved hype this year, and they get Oklahoma at home to finish the season. That could signal that this fall's Bedlam battle will be the biggest ever. But ... the 'Pokes go on the road to face almost every other good conference team: Texas A&M, Missouri, Texas and Texas Tech. That is brutal, and it will be a major hindrance. If OSU wins the Big 12 this year, they'll have really, really earned it.
Football Outsiders' projections are perhaps best at locating the paper tigers: the teams who are being given too much credit, too quickly. This year's major "too much, too soon" candidate is obvious: Texas A&M. The Aggies played like a top 40 team for the first half of the 2010 season and a top 20 team over the last half. Now they start 2011 as a top 10 team. They will be good, but they might not be that good. The Florida State Seminoles, on the other hand, have the profile to live up to the hype. Their base of talent (i.e. five-year recruiting ranking) is as strong as anybody's, and last year's surge to the ACC title game wasn't actually much of a surge.
FSU's four-year F/+ average ranks them in the top 25; they've finished in the top 30 in each of the last three years, and last year's "surge" only took them from 29th to 15th. The major difference was, before 2010, this was certainly a "whole is less than the sum of its parts" situation. The foundation has always been rather strong.
Whether the 'Noles can get past what might be an undefeated Virginia Tech team in the ACC title game is a different story. But this is going to be a damn good team.
TCU has performed at a Top-Five level in recent years, and with a move to the Big East coming next year, they will be getting a lot more attention this season. In theory, that's good; Gary Patterson has done an unbelievable job in Fort Worth, and he deserves all the commendation he gets. One problem: the Horned Frogs return just eight total starters, four on each side of the ball, and they must replace both quarterback Andy Dalton and four offensive line starters, including Rimington Award-winning center Jake Kirkpatrick. The defense will be good because, well, Gary Patterson doesn't put a bad defense on the field, but the offense could regress enough to lead to an early slip-up on the road. TCU plays Baylor, Air Force and San Diego State over the first half of the season, then heads to Boise State in November. That's a lot to ask of such a young team.
The new Texas defensive coordinator did some amazing things at Middle Tennessee and Mississippi State in recent years, and evidently he's a fan of PPP+, or so he told Bruce Feldman. We see you, Coach.
Remember all those crazy stories of guns and drugs from places like Oklahoma and Miami in the 1980s? It was easy to think we had somewhat moved past that. At least, it was until Yahoo! investigative journalist Charles Robinson started digging. His story on the drugs-and-yachts-and-abortions saga at The U was incredible. Discouraging, somewhat depressing, and incredible.
The NFL got a lot of polite claps for the way they have decided to enforce Terrelle Pryor's NCAA suspension. Personally, that is rather terrifying. If we want to have any shot of returning some sort of perceived amateurism to college football, then we have to a) disconnect the NCAA from the NFL as much as possible and b) give players a more direct path to professional opportunities so that they aren't forced to go to college just to eventually make the pro ranks. Granted, the CFL exists and should be utilized. That would help. But while we're figuring out how to move in one direction, the NFL's treatment of Pryor took us very far in the other.
Just so people can see the entire picture of how we came about developing our F/+ projections, here are the S&P+ projections that matched Brian Fremeau's FEI data. These have not been updated to account for recent suspensions, injuries, etc. I just ran out of time on that one.
|Rk||Team||Proj. S&P+||Rk||Team||Proj. S&P+|
|1||Alabama||247.4||61||San Diego State||204.5|
|2||Boise State||243.2||62||Kansas State||204.3|
|Rk||Team||Proj. S&P+||Rk||Team||Proj. S&P+|
|23||Texas A&M||223.3||83||Louisiana Tech||194.8|
|29||Michigan State||218.4||89||Miami (Ohio)||190.6|
|32||Penn State||216.9||92||Central Michigan||189.6|
|35||Mississippi State||214.5||95||Washington State||186.6|
|Rk||Team||Proj. S&P+||Rk||Team||Proj. S&P+|
|42||BYU||211.9||102||San Jose State||184.2|
|45||North Carolina||211.1||105||Colorado State||183.2|
|52||South Florida||206.9||112||Western Kentucky||179.6|
|55||Boston College||206.5||115||Ball State||178.0|
|58||Southern Miss||205.2||118||New Mexico||172.7|
|60||Tulsa||205.0||120||New Mexico State||166.8|
I cannot tell you how happy I was to see actual football on my hospital television last night. Here's to a damn good season, one that doesn't end with "We'll fondly look back at this brilliant game when Vacated beat Vacated, 22-19" jokes like last year.
14 comments, Last at 04 Sep 2011, 6:35pm by Dales