Brian Fremeau is back for his 10th season of college football advanced stats at Football Outsiders, with a lesson on average teams (like Arkansas State) and updated FEI projections for 2016.
14 Nov 2008
by Bill Connelly
In 2004, Missouri coach Gary Pinkel decided to make an adjustment. The Tigers had a breakthrough season in 2003, Brad Smith's sophomore year. But Pinkel saw that defenses were starting to adjust to Smith's tendencies (i.e., keeping the ball a lot); they were starting to try to contain him instead of pursuing him, and they were attempting to make him throw the ball to succeed. While he had had huge games against teams like Texas Tech, Nebraska, he had struggled against others. Pinkel saw that if he could make Smith a better passer, it would open up rushing lanes. Square peg, meet round hole.
After a 5-6 season that almost got Pinkel fired, he had come to a conclusion: Brad Smith is not a pocket passer. He switched to more of a spread offense, he got Chase Daniel to commit, he got Mizzou back to a bowl game in 2005, and the program took off. But it almost never happened.
I had this in mind when I decided to write a "What's wrong with _____?" column and couldn't decide between Auburn, Tennessee, and West Virginia. When Bill Stewart started speaking of "making Pat White a better passer" this past offseason, my radar perked up. I feared what might happen to the Mountaineers. The "make the runner a passer" concept is a familiar one for me, so we're going with West Virginia.
As teams like South Florida had been slowly adapting to West Virginia's style and tendencies, new WVU coach Bill Stewart made Pinkel-like proclamations in the offseason. Teams won't be able to gang up on the Mountaineers running game because Pat White's going to pass really well in 2008, he said. It makes sense, unless by attempting to strengthen a player's weakness, you are also changing the mindset that made the player special in the first place.
On their face, WVU's 2008 offensive numbers really are not that bad. They are No. 14 in the country in rushing offense, and Pat White is No. 25 in pass efficiency. That seems to be a pretty good combination. But the rushing yards are not coming at key moments.
The Mountaineers are only No. 65 in the country in scoring offense, and though their passing has been efficient, they are still No. 108 in overall passing offense. Meanwhile, more defenses have followed the "Stopping Pat White" blueprint successfully. East Carolina held WVU to 3 points, Colorado held them to 14 points, Syracuse held them to 17, and they were trailing Cincinnati, 20-7, in the fourth quarter this past Saturday until a late rally. In the end, they are averaging 80 fewer rushing yards and 14 fewer points per game; this was not expected with a backfield of Pat White and Noel Devine that torched Oklahoma in last year's Fiesta Bowl.
The entertaining blog West BY GOD Virginia has placed the blame for the offensive struggles on one man: Offensive Coordinator Jeff Mullen. But is it so simple? What is West Virginia actually doing differently in 2008?
To attempt to answer this, we should first look at the situations in which WVU is indeed passing more. They have thrown the ball 39 percent of the time in 2008, as opposed to 31 percent in 2007. They haven't exactly become Texas Tech in their tendencies, but there is a difference.
|WVU: Situational Run-Pass Splits|
|Situation||Pass %||Pass S&P||Run %||Run S&P||Pass %||Pass S&P||Run %||Run S&P|
|All First Downs||22.1%||0.825||77.9%||0.881||40.5%||0.946||59.5%||0.678|
|All Second Downs||30.2%||0.895||69.8%||0.812||35.8%||0.726||64.2%||0.796|
|All Third Downs||51.7%||0.937||48.3%||1.016||38.5%||0.487||61.5%||0.983|
* Short yardage is defined as less than five yards to go for the first down. Medium yardage is 5 to 10 yards, and long yardage is anything over 10. These categories (and more specific ones like "very short," "medium-long," etc.) were derived and categorized by looking at success rates for each yard increment and grouping the ones closest to each other.
There are a lot of numbers here, and while they back up the fact that WVU is throwing more, they do not show whether they're throwing more out of strategy or necessity. Opponents are ganging up on the run on first down, and it appears that WVU has to throw more there (40.5 percent passes on first downs in 2008, 22.1 percent in 2007) because the run is ineffective. Pat White is doing a decent job of getting yards out of the passing game in those situations; WVU is averaging 0.35 PPP (Points Per Play) on first downs, as opposed to 0.38 in 2007. It's a difference, but it's not a tremendous dropoff.
However, that number falls slightly to 0.34 on second downs (0.38 in 2007), and their 0.37 PPP on third downs is much lower than the 0.49 of last season. In all, the big plays simply are not there, especially on third downs and key situations. While Noel Devine's 0.747 is less than Steve Slaton's 0.785 (and Devine's own 0.947) from 2007, the biggest individual drop comes from Pat White. White had a 0.964 Rushing S&P in 2007; that has fallen to 0.765.
In all, WVU's productivity has dropped in a lot of categories, though. Where have the drops been most precipitous?
|WVU: Biggest statistical drops in 2008|
|Third Down Passing Success Rates||-53.8%|
|Rushing Points Per Play, Non-Passing Downs||-44.5%|
|Rushing EqPts per Game||-42.8%|
|Third Quarter Rushing PPP||-41.8%|
|Fourth Quarter Rushing PPP||-40.8%|
|Red Zone Rushing PPP||-40.2%|
|First Down Rushing PPP||-36.8%|
|Fourth Quarter Rushing S&P||-34.9%|
|Third Down Rushing PPP||-33.6%|
|Third Down Passing S&P||-33.0%|
Almost all of these categories involve rushing PPP and third down passing. Of course, these two go together pretty well. The fewer yards/EqPts the Mountaineers are getting on the ground, the longer their third downs will be. Led by Devine, the running backs are less explosive this season, and Pat White is both less explosive and less efficient.
Whatever the problem with the rushing game, it doesn't appear that the offensive line is to blame. West Virginia is averaging 3.22 line yards per carry in 2008; they averaged 3.25 in 2007. In close-game situations (scoring margin within 16 points or less), WVU is actually doing better in line yards -- 3.25 in 2008 after a 3.12 average in 2007.
So we seem to have reached a chicken-or-egg situation. Is the running game failing because of a higher focus on the pass? Is the pass a necessity because teams are shutting down the run?
In 2004, attempting to make Brad Smith a pocket passer took away his running instincts, and even when he did run, he wasn't as aggressive. His yards per carry fell from 6.8 in 2003 to 3.4 in 2004. Is the same thing happening to Pat White? Has the "pass more" mindset affected his running instincts? Has the play-calling just been terrible? Have more defenses simply been able to follow the South Florida-esque "Stopping Pat White" blueprint by making the harnessing of the White-Devine zone read the No. 1, No. 2, and No. 3 priority and ignoring the passing game? Stats alone cannot tell that story, but they certainly do suggest that Jeff Mullen needs to start getting a lot more creative in figuring out ways to get White and Devine -- two obscenely talented runners -- in open space. He only has a couple more games to figure out how to unshackle the running game.
|Iowa 24, Penn State 23|
|Field Position %||62.1%||37.9%|
|Points Per Play (PPP)||0.26||0.33|
|S&P (Success + PPP)||0.746||0.806|
|CLOSE GAME ONLY|
|S&P by Quarter||Q1 S&P||0.648||1.306|
|S&P by Down||First Down S&P||0.714||0.742|
|Second Down S&P||0.912||0.603|
|Third Down S&P||0.519||1.384|
|Total T/O Pts||4.29||9.70|
|Turnover Pts Margin||+5.41||-5.41|
Here's what I said about this game last week:
Iowa's defense is the second best among BCS teams in terms of shutting down teams on passing downs. Meanwhile, Penn State has a rather high ratio of passing downs success to non-passing downs success (the national ratio is 0.802). And they're hosting Penn State this weekend. Consider that an upset alert.
This was the first game for which I entered the play-by-play this week, as I wanted to see just how much of an impact Penn State's Passing Downs numbers had on this upset. They still managed to score a respectable 23 points, but it's clear that failures on passing downs (their 0.450 S&P on PDs was less than half of their 0.904 season average) held up the HD offense at key moments. So consider this a victory for PD-to-NPD disproportionality!
Next up on the disproportionality upset watch: During their three-game streak ridiculous offensive output, Florida's S&P on passing downs and non-passing downs have been virtually even. In their remaining three games, they face two defenses with top-ranked passing downs defense numbers: South Carolina and Florida State. I don't expect South Carolina to be able to keep up with the Gators in The Swamp, but watch out for the Florida State game in Tallahassee in a couple of weeks. FSU is running the ball well, and they're by far the best in the country at clamping down on passing downs. That could be a good combination for a surprising result.
Also, we're in "irresistible force vs immovable object" territory with next weekend's Oklahoma-Texas Tech game. Both teams have offenses and defenses that are great on passing downs. Something's got to give there, and I think the advantage of Owen Field (which is the loudest place I've ever attended when the road opponent has a big play) means the Sooners have the edge.
Yeah, Iowa's one of those teams that's just quietly putting together a solid, but unspectacular seasons due to a bunch of close losses (much like Ole Miss), and is therefore just flying under the radar completely. They haven't lost a game by more than 5 all year.
But I'm not sure that just looking at passing down S&P is a great way of finding an upset, because in order to get to a passing down, you have to stop a team on a non-passing down. So really, without knowing what the non-passing down S&P comparison is, it doesn't really help much.
Very true, but A) even great offenses are going to see passing downs 15 to 20 percent of the time (as compared to 35 to 40 percent for bad offenses), which still means that there will be opportunities for drive-killing stops, and B) the ideas for finding potential upsets are based around red flags. The PD-to-NPD disproportionality is certainly one, and I'm always on the lookout for more.
[I]s it better to get another first down on first down, or a second-and-1?
Back in my very first VN column, I mentioned that I looked at potential EqPts values based on 1) yard line, 2) down and yard line, and 3) down, distance, and yard line. In the coming weeks, we'll take a look at (2) and (3), at which point this will be a very interesting question to consider.
2 comments, Last at 15 Nov 2008, 6:19pm by Anon