Writers of Pro Football Prospectus 2008

Most Recent FO Features


» Week 3 Quick Reads

We break down the very ugly game Joe Flacco had against Jacksonville, how completely ineffectual the Ravens offense was -- and why it wasn't even the worst game among quarterbacks this week.

30 Oct 2015

VN: Passing Downs Magic Acts

by Bill Connelly

In my ongoing exploration of old Varsity Numbers pieces, I discovered exactly the thing I hoped to discover in this exercise: an old concept that I had allowed to completely fall by the wayside.

Back in 2008-09, I talked quite a bit about disproportionality -- an offense's ability to consistently bail itself out on passing downs while struggling on standard downs. As the idea went, if your success on passing downs is disproportional to your success on standard downs, it will be hard to maintain that success in the future.

In the 2008-09 offseason, I explored this idea for both sides of the ball. First, some examples of disproportionality on offense:

Teams that returned star quarterbacks and major receiving threats from 2007 to 2008 (teams like Georgia, Oklahoma, and Texas Tech) improved in 2008. Meanwhile, teams with offensive deficiencies (Washington State and Washington, for example) or teams having to replace key offensive cogs (Kentucky, Wisconsin) regressed, some significantly.

Looking again at the 2008 list, Texas might be OK. They return Colt McCoy and his top weapon, Jordan Shipley, plus major big-play weapons in Malcolm Williams, Dan Buckner, and others. They still don't have a go-to running back, and while they could regress a bit, the damage might not be too bad. Meanwhile, Ole Miss returns Jevan Snead but loses big-play threat Mike Wallace, and Penn State has to replace almost their entire receiving corps. Look for a potential drop from the Rebels and a potential large drop from the Nittany Lions.

The results according to the S&P+ data I now use: Texas did fall a bit, from second in Off. S&P+ in 2008 to 15th in 2009. The other two examples fell more -- Ole Miss from 24th to 63rd, Penn State from seventh to 28th. And of course, when Colt McCoy graduated, Texas plummeted to 86th in Off. S&P+ in 2010. I called him a passing downs magician for years because of this.

On the defensive side of the ball, the conclusions were similar:

In all, ten of 12 teams with a disproportionality ratio of at least 1.18 saw a fall in Defensive S&P+ from 2007 to 2008, and eight fell at least 8 percent. Clearly experience and overall quality play a role here -- there is not much concern for Miami (OH)'s defense regressing considering they were 113th in 2008 -- but disproportionality alone seems to be a decent warning sign for a fall. With a depleted pass rush and by far the highest disproportionality, Texas Tech should see a decent-sized fall from their spot as the 45th overall defense. They didn't have an amazing defense in 2008, but they made big plays, especially on Passing Downs, and that allowed them to utilize their experienced offense and go 11-1. With less experience on offense and a regression on defense, they should fall back to the 7-5 or 8-4 range in 2009.

The next team on the list, Iowa, has had a reputation for always having solid defenses under Kirk Ferentz, but until 2008 they had not finished in the Top 15 in Total Defense since 2004, so this surge back to near the top of the rankings will probably come with at least a slight fall in 2009. Meanwhile, both Connecticut and Pittsburgh have had decent defenses in the past, but they will probably regress a bit as well; Connecticut especially should fear at least a slight fall, especially with the loss of second-round draft picks Darius Butler (cornerback) and Cody Brown (linebacker).

Iowa: eighth in Def. S&P+ in 2008, sixth in 2009. Connecticut: 29th, then 50th. Pitt: 16th, then 21st. Nailed UConn, at least.

Let's look at some more current examples.

Using 2014 standard downs and passing downs data*, and comparing teams' change in Off. S&P+ from 2014 to 2015, we get the following.

* One note: I found that correlations are stronger using simply the opponent-adjusted success rate figure (SR+) for standard and passing downs instead of the combined S&P+ figure. So that's what you see below.

Ratio of PD SR+ to SD SR+ Count Avg. change in Off. S&P+
<0.85 16 +2.19
0.86 to 1.00 41 +1.75
1.01 to 1.15 51 -0.60
1.16+ 19 -2.26

What does this tell us? That if your Passing Downs SR+ was quite a bit better than your Standard Downs SR+, you are probably going to regress the next year. And if it's quite a bit lower, you're probably going to improve.

What happens if we look at the Colt McCoy Effect, i.e. the impact a returning quarterback has on these numbers? Looking at only the top range and bottom range, we find this:

Ratio of PD SR+ to SD SR+ Avg. change in Off. S&P+
(teams with returning QB)
Avg. change in Off. S&P+
(teams with new QB)
<0.85 +3.38 -1.38
1.16+ -1.04 -5.66

So if your ratio is too high, but you're returning your starting quarterback, you have a chance at holding steady. (And it should be noted that, if you remove UCF from the "returning starter at QB" list because Justin Holman has been injured, the average change rises from minus-1.38 to minus-0.38.) But if your ratio was high and you have a new starter, you're regressing by almost a touchdown. That's pretty significant.

Meanwhile, if your ratio was too low, and you have a new starting quarterback, you might get even worse. But if you return your starting quarterback, therefore maintaining more experience at the toughest position on the field, you will improve by about half a touchdown.

This is something I perhaps shouldn't have allowed to drop!

What about the defensive side of the ball? As you might expect, the connections there are a little more tenuous. (There is no quarterback after all, even if we call the middle linebacker or someone else the "quarterback of the defense.")

(Remember: since S&P+ is presented as an adjusted point total now, when it comes to defense, negative numbers mean improvement.)

Ratio of PD SR+ to SD SR+ Count Avg. change in Def. S&P+
<0.85 15 -3.09
0.86 to 1.00 45 +0.93
1.01 to 1.15 54 -0.18
1.16+ 13 +0.63

The numbers are more all over the map here, but of the 15 teams in the lowest range, the ones with a PD S&P+-to-SD S&P+ ratio under 0.85, 10 have seen their Def. S&P+ improve, all by at least 3.5 points. (The averages are skewed by Rice, with its 0.80 ratio in 2014, managing to regress by 11.0 points.)

Out of curiosity, I looked into one more piece of defensive data. Instead of looking at whether your quarterback is back or not, what happens if we look at whether your pass rush is better or worse? My thought was that your pass rush has a pretty clear, obvious impact on your passing downs defense in particular. If you lose the key components of your pass rush, then perhaps that PD-to-SD ratio means something pretty clear?

Ratio of PD SR+ to SD SR+ Avg. change in Def. S&P+
(Teams with worse pass rush)
Avg. change in Def. S&P+
(Teams with improved pass rush)
<0.85 -0.86 -4.58
0.86 to 1.00 +1.19 +0.74
1.01 to 1.15 +3.29 -2.76
1.16+ +0.38 +1.21

The data is still a bit on the loosey-goosey side there, but teams in that particular "<0.85" range are separated more clearly. If your defense was a little too unsuccessful on passing downs, and you figure out a way to get more pressure on the passer, voila. You are all but guaranteed a better defense.

College football projections are a funny thing. I look for more and more factors that can lead us to understanding more clearly about clear expectations for a given team ... knowing that a handful of injuries, suspensions, coaching changes, and random "he's 20 years old, and 20-year-old males are flaky" instances will render about half the projections moot the moment the season begins. But still! Here's one more thing to add to the arsenal.

By the way, we're still barely midway through the season, and data will change, but which teams might expect to improve or regress -- either in this season's final weeks or next year -- based on the same standard and passing downs data?


  • Offenses with 0.84 ratio or lower (25): Wyoming (0.54), Appalachian State (0.63), Utah State (0.65), Charlotte (0.68), Georgia (0.70), New Mexico (0.72), Akron (0.72), Colorado (0.76), Arkansas State (0.76), Baylor (0.76!), Kent State (0.76), Air Force (0.78), UCF (0.78), Iowa State (0.78), Oregon State (0.80), North Carolina (0.81), North Texas (0.81), Penn State (0.82), Navy (0.82), UTSA (0.83), Purdue (0.83), Pitt (0.83), Oregon (0.83), Georgia Tech (0.84), UL-Monroe (0.84)
  • Offenses with 1.16 ratio or higher (28): MTSU (1.16), Arizona State (1.16), Idaho (1.17), LSU (1.17), WMU (1.18), UCLA (1.19), South Alabama (1.19), Notre Dame (1.19), Ole Miss (1.20), Virginia Tech (1.20), Arizona (1.22), Ohio State (1.22), Nebraska (1.24), Duke (1.26), Indiana (1.26), Clemson (1.28), ECU (1.28), Mississippi State (1.29), Wake Forest (1.29), Temple (1.30), Kansas State (1.33), Bowling Green (1.37), Michigan State (1.39), Oklahoma State (1.44), Louisville (1.46), Florida State (1.47), Boston College (1.47), Auburn (1.49)

That these buckets are pretty crowded reminds us that some of these teams will likely even out this year. That might be a warning sign for a team like Clemson, I guess. Then again, the Tigers have Deshaun Watson at quarterback. They might be just fine.

By the way, I received this e-mail about Michigan State yesterday. It seemed applicable.

I'm writing you because of something I pulled out of there. It was borne from reading a silly question online, some Michigan State fan asking if Connor Cook was the best QB in the country (Bruce Feldman and Stewart Mandel helped by debating Cook on their podcast).

On first look, the question seemed obvious. He's not a star by most conventional metrics, and even with something such as QBR, he was outside the top 10. He seemed to play better than that, but I had a hard time saying someone could have bad stats, lead an offense that wasn't all that great and be considered in that rarefied air. Even if you factored out running, the passing game was outside the top 30 in S&P+, and it wasn't as if it were particularly efficient (crediting him for being methodical) or explosive (implying he was going deep more). But in looking through the numbers, I stumbled on an archetype and perhaps title that works for him.

The highest leverage quarterback in the country.

MSU's S&P+ is 104th on standard downs (100th success rate, 26th IsoPPP). That jumps to 14th on passing downs (ninth in success rate, though the explosiveness falls off to 92nd). This would have been interesting enough, but it wasn't as if you could credit a guy for his team struggling on standard downs. Then I looked up and realized MSU is 15th in standard down run rate, which of course is knocked down at least five spots by option teams.

Michigan State continues to be a run-first team even though the Spartans can't actually run the ball very well. They then ask Connor Cook to bail them out on passing downs, and he does so. That might be a little bit scary when it comes to the Spartans' prospects in 2016, when Cook is no longer around. For now, though, he gives them a Colt McCoy-ish weapon.


  • Defenses with 0.84 ratio or lower (28): Boise State (0.64), Fresno State (0.67), Toledo (0.70), Wyoming (0.71), Colorado State (0.72), Appalachian State (0.74), UL-Monroe (0.75), Connecticut (0.75), Bowling Green (0.76), Rice (0.78), Georgia State (0.78), Air Force (0.79), Old Dominion (0.80), Pitt (0.80), Tulane (0.80), UL-Lafayette (0.80), Miami (Ohio) (0.80), Washington (0.81), SMU (0.82), North Texas (0.82), Virginia (0.83), South Carolina (0.83), Arizona (0.84), Michigan State (0.84), Iowa (0.84), Alabama (0.84), Buffalo (0.84), Texas Tech (0.84), Baylor (0.84)

Again, a pretty big bucket there. It will likely thin out a bit in November.

Posted by: Bill Connelly on 30 Oct 2015