This article is the fourth in an offseason series investigating issues from Oregon’s 2018 football season; this will focus on the Ducks’ passing offense. I use statistical analysis of my play-by-play charts of each game to try to answer big questions in a methodologically rigorous manner. To that end, I’m excluding garbage time from each game, as well as the first three non-conference games and the Civil War as these were FCS-caliber opponents. For more information about the dataset I’m using, please see my February article on playcalling.
After excluding unhelpful data, we have 242 designed downfield passing plays to analyze. Oregon gained enough yardage, given the down & distance, for it to have been a successful play on about 53% of all such snaps, which is in line with their rushing and screen pass offense as well; compared to previous teams for which I’ve done whole-season film reviews, that amounts to conference-championship caliber efficiency but not playoff-caliber. About 20% of those 242 snaps were broken in some way and resulted in a scramble, sack, throwaway, or other improvised play - these are tough to analyze as to the play design and the broadcast camera angles quickly become useless, but fortunately #10 QB Herbert is a fairly good improviser and still put together a successful play a third of the time.
One of the datapoints I gather when charting games is the primary reason I believe the play succeeded or failed. This is necessarily subjective and I can’t claim it’s perfect, but I have been charting games for more than a decade now and I can at least say I use a consistent set of criteria. Here’s my breakdown of the cause of each failed downfield passing play in this data set:
Pass play breakdowns
|QB's poor decision||25||21.37%|
|Receiver can't get open||15||12.82%|
|Heroic defensive play||15||12.82%|
The data defy any easy, single explanation for why the Oregon passing game wasn’t more efficient. There’s a pretty even distribution between the three main groups of role-players, and I found no significant correlation between any pre-snap input (alignment, formation, play design, skill personnel) and success rate. It doesn’t appear that there’s any “magic bullet” to improving Oregon’s passing performance, other than an across-the-board development of talent and execution (some better injury luck would be helpful too).
There are, however, some interesting wrinkles within these categories to examine.
In the previous three articles, I’ve detailed how Oregon’s o-line injuries suffered during week 7 derailed the offense for a couple of weeks – during weeks 8 and 9 both the rushing and passing production drop dramatically as running lanes don’t get open and the QB is under significantly more pressure. Starting in week 10, different o-line configurations and a rebalanced gameplan (featuring more screen passes and offset RB formations) take hold and result in regaining their footing, although not at the same elite efficiency as in the first three conference games.
Interestingly, the downfield passing aspect of the gameplan during this “Phase 3” doesn’t make the same changes as the rushing game I examined last month; the distribution of pre-snap inputs looks almost identical throughout the entire season on these plays.
The other interesting factor is that while the effects of a weakened o-line are obvious, there’s a much bigger share of blocking breakdowns from the tight ends on passing plays compared to rushing plays, and the running backs in pass-protection contribute to a significant share of QB hurries. That is, the o-line -- even during the dark days of “phase 2” -- do an improved job of pass-pro, and the pressure that gets through is usually coming from the edges.
The well publicized plague of drops is borne out in the data, although while these are memorable and demoralizing plays, in terms of the hit to efficiency they’re hardly the whole story. Of Oregon’s five receivers with more than 10 non-garbage-time targets (#3 WR Johnson, #9 WR Schooler, #13 WR Mitchell, #27 TE Breeland, and #30 WR Redd), they all have a fairly similar drop-per-target rate a bit under 9%. The distribution over time is a little weird, though: almost all of Johnson and Redd’s drops come during the non-conference games and Civil War, and a third of Mitchell’s drops on the season come in the bowl game alone. They appear to be random events, and I can’t find any significant predictor of when they’ll happen. Subjectively, the common fan theory that Herbert throws an uncatchably hard ball does not match my observations, and at any rate the drop rate is too low for this to be a good fit for the data given that it would be a structural factor.
Of course, the big factor that jumps out in the receiver data is that Mitchell is the target on about 47% of all downfield passes. The suggestion that this is embedded in the playcall or Herbert is being instructed to throw it to him is risible -- that’s not how a read-progression passing system works -- and the minute differential in the various receivers’ drop rates doesn’t explain it either. Instead, this is pretty clearly an artifact of the receivers’ legs, not their hands – I recorded every other pass-catcher as getting open at about half the rate that Mitchell does. By far the most successful play Oregon has is the right-side comeback route to Mitchell, and it’s clearly because opposing defenses respect his speed so much they have to play an extra yard or two off him and that creates enough cushion when he makes his break to lock down the ball before contact.
The failure of receivers not named Mitchell to get open is also the best explanation for what I highlighted in February as the missing “oop” in the alley-oop of Oregon’s offense: the failure to complete passes at an efficient rate on 2nd & short. In such down & distance situations, Herbert’s decision-making and the drop rate actually improve compared to the global data, but two things get dramatically worse: the “nobody’s fast enough to get open” problem on all routes regardless of depth, and the receiver/quarterback miscommunication problem.
I examined the miscommunication issue in my weekly film reviews over the first four games and never came to a satisfactory explanation (I joked on a podcast that it might be something like Wazzu’s armband snafu; to this day I still don’t have a better theory). These plays make it impossible to tell if it’s the QB or the receiver who got it wrong, and the fact that during conference play they almost all happened on 2nd & short is even more puzzling.
I spent a good bit of time in my weekly film reviews discussing Herbert’s “misfire” issue, where for a stretch of games he’d rifle a ball at a receiver’s feet on a disturbing number of plays. Reviewing the data and film of each one I can’t find a correlation between them and don’t have any explanation for why they were happening, but it looks to me to be a physical issue – he often would look at his arm and make a gripping motion afterwards. I think he had a mild injury of some kind, complicated by a clear shoulder injury in the Civil War. Fortunately by the bowl game it seems to have cleared up.
The bigger issue is Herbert’s decision-making, and some of the problems here continue trends that I was writing about last summer in my review of the 2017 season – he holds the ball too long looking for the perfect throw and gets surprised by unorthodox blitzes. But there’s also been a peculiar reversal in something I criticized him for last year: then, he trusted his arm too much to blaze the ball through underneath coverage before the linebackers could get their hands up, and it culminated in several very close calls at the end of the season and a couple of devastating interceptions in the Las Vegas bowl. In 2018, he’s doing the opposite, illustrated at 4:11 of this video: skipping open passes to the middle of the field when there’s any backer in his field of view, in favor of one-on-one contests with DBs.
This tendency is no doubt influenced by some early bad experiences with drops and receivers not getting open, but there are still dozens of plays I have marked on my tally sheet where he’s either not seeing or not choosing wide open guys. If Herbert continues to consistently signal that opponents can scare him off with such light midfield defense it’s going to drag down the efficiency of every other part of the playbook, even if he does get an upgrade at receiver.
As readers will recall, Oregon ran virtually all its plays out of 10- or 11-personnel, used a very small rotation of receivers, and showed only one variation that had any correlation to either playcall or success rate - where the single set back was aligned, pistol or offset. Formation did not have any noticeable effect on play design or whether the pocket broke down for dropback passing plays – the playcall distribution is very similar regardless of how they line up.
There is a small, suggestive difference, however, in success rate: pistol-formation play-action passes are more effective, especially at intermediate and seam routes, but deep throws against man coverage are more effective out of the offset. Also, scramble plays are significantly more effective when the back was in the pistol and came up into pass protection – this might have affected the pass rush, but it’s still pretty weird that the jump is so big, and I suspect it’s coincidence. Overall, passing plays out of the pistol are about four percentage points more effective.
- I was somewhat disappointed that I wasn’t able to find “one weird trick” to improve Oregon’s passing performance. Looking at whether the Ducks needed to block better, upgrade at receiver talent, develop Herbert’s decision-making, or just get luckier, the only conclusion I could come to is, “yes.” Like last month, I have about a hundred cross-tabs of multivariate comparisons that show no meaningful correlations, and listing them would be terribly tedious, but I’m happy to answer any specific questions that readers have in comments.
- What do Duck fans think is going on with Herbert’s relationship with the middle of the field? The reversal between 2017 and 2018 is so stark that there must be something going on there, but I’m not sure what the explanation is.
- It’s reasonable to expect improvements in several aspects of Oregon’s passing game – the running backs in pass pro are no longer true freshmen, some of the incoming receivers look talented and ready to play, and it’s unlikely (knock on wood) that the o-line has two starters go down again. But I’d like to know what fans think about the prospects for improvement at tight end, both blocking and receiving. It’s great to see the return of #84 TE McCormick, and Breeland and #87 TE Bay are frequently the stars of Flex Friday videos, but is this really going to add up to a game-changing difference?