2018-19 Deep Ball Project
Guest column by John Kinsley
One of the more intriguing parts of an NFL offense is the deep pass. Just as they do for the three-point shot in basketball, audiences seem to go crazy the further a pass is completed, and why wouldn't they? The odds of completing a deep pass are smaller than completing a pass in the short/intermediate areas of the field, so seeing these explosive plays end up successful is captivating.
There are many ways to measure deep passing, and the most popular methods are either focused on completion percentage, yards, or touchdowns. There's certainly nothing wrong with these methods, but it should be noted that completion percentage does not equal accuracy. Drops and inaccurate completed passes are not factored into completion percentage.
That's where The Deep Ball Project comes in. You may recall that I previously wrote a piece on this study last year for Football Outsiders but if you don't, here's the rundown: The Deep Ball Project separates accuracy percentage and completion percentage, focusing on a quarterback's accuracy throwing downfield regardless of whether the ball was caught or dropped. Raw stats such as yards, touchdowns, and interceptions are included, but accuracy percentage is the main stat.
Before I discuss what's featured in the 2018-19 edition of The Deep Ball Project, you can check out this year's edition of it right here. In previous editions I had included all throws of 16-plus air yards, but this time I decided to only include throws of 21-plus air yards, so the accuracy numbers and raw stats are completely different from those of previous years.
In last year's edition, I included efficiency score as a way to reward quarterbacks that made accurate throws under pressure and into tight windows. Going into this year's edition, however, I thought efficiency score only rewarded longer throws instead of overall accuracy. So that was removed for the 2018-19 Deep Ball Project, and instead the main focus is accuracy percentage.
Accuracy percentage can differ greatly from completion percentage. Aside from straight up drops, accurate passes are counted on incomplete passes if the receiver can't keep two feet in bounds on a pass he should have had no problem hauling in, some Hail Marys, some pass disruptions (depending on the placement), etc.
On the other hand, completions where the receiver has to make an unnecessary adjustment are punished as inaccurate passes. Back-shoulder passes are not included in here for obvious reasons, but passes where the receiver has to go out of his way just to make a catch are. (Just watch a highlight reel of Odell Beckham, or DeAndre Hopkins before Deshaun Watson.) As such, accurate incompletions and inaccurate completions have returned for this edition, though the amount is drastically reduced since the 16- to 20-yard throws from the past are removed from the equation.
New stats include throws to the left, middle, and right of the field, and the distances themselves have been modified from the 2017-18 Deep Ball Project (20-24 to 21-25; 25-29 to 26-30; 30-34 to 31-35; 35-39 to 36-40; and 40-plus to 41-plus). Air yards and yards after the catch (YAC) have also been added, as have passes defensed (PD), which include interceptions and straight up pass breakups. And as was the case for the 2017 season, the playoffs are included.
For this writeup, I'm excluding the raw stats (including air yards and yards after the catch), but they can be found in the full article on my site. Instead, this article will focus on accuracy percentage to completely describe the context of the numbers for the quarterbacks charted.
This is not a perfect study. It has its flaws. This doesn't factor in decision-making, poise, audibles, time in the pocket, etc., as is the case for most studies of this kind. No football study is perfect, but I still enjoy doing the research for The Deep Ball Project and adding new stuff to each edition to make it better and before.
For this write up, I'll do my best to explain the context behind the numbers of certain quarterbacks. With that said, let's dive right into this. First, here's a look at the mean value (average) of the raw statistics in this study:
|2018-19 Deep Ball Project: Stat Averages|
A total of 709 accurate deep passes (ACC) were thrown by 35 qualifying quarterbacks in 2018, an average of 19.0. The total amount of attempts came to 1,520, while the average was 43.4.
The average accuracy percentage was 46.6 percent, and in this study 19 quarterbacks finished above that percentile. Keep this in mind as we look at everyone's total accuracy percentage and raw numbers on all throws of 21-plus air yards, as displayed below. NOTE: The rankings are decided by accuracy percentage, as highlighted in the black boxes. Yards, touchdowns, interceptions, air yards, and yards after the catch are not looked at for this writeup, as I wanted to focus on accuracy percentage.
|2018-19 Deep Ball Project: All Throws of 21-Plus Air Yards|
Average accuracy percentage: 46.6 percent.
There's a lot to take in when looking at this chart, so let's get to it.
Yes, the guy who couldn't physically throw a football the season prior finished as the most accurate deep passer in the 2018-19 Deep Ball Project. Andrew Luck has usually been sensational in deep passing when he has actually been healthy, and with the help of a real offensive line he was at the top of his game throwing downfield, finishing with a 60.0 percent accuracy percentage, leading all 35 charted quarterbacks.
In second place is Luck's 2012 draft classmate Russell Wilson, who just barely finished behind Luck in downfield passing accuracy (59.0 percent). Despite being in a run-heavy offense, Wilson actually finished with one more deep attempt than Luck. His style of play has always been fun to watch thanks to his rocket arm and incredible scrambling ability.
This is yet another study that has high praise for Matt Ryan's 2018 season. He was one of the Falcons' bright spots despite an injury-plagued year for the team. The Browns having a real quarterback in Baker Mayfield is not something I'm used to, but it's true. Mayfield had the best situation of any rookie quarterback in 2018 (mainly in the second half of the season when Freddie Kitchens took over as the interim offensive coordinator), but his deep accuracy was terrific even in the first half of the year, and with a supporting cast that has added Odell Beckham we could see his numbers skyrocket even further.
Surprisingly, Patrick Mahomes was only fifth in deep accuracy. His playmaking skills were insane, and as you probably expected, he was great under pressure. Fellow AFC West quarterback Philip Rivers had one of his best seasons to date, and while his accuracy fell off a bit in the second half of the year, it wasn't enough to take him out of the top six in accuracy percentage.
Marcus Mariota has received a lot of flak, but his deep accuracy in the last two editions of The Deep Ball Project has been really damn good. Jared Goff and Carson Wentz are two quarterbacks who took huge steps forward in deep accuracy in 2018, but Drew Brees, while finishing in the top ten in accuracy percentage, barely attempted any passes beyond 31 air yards.
Three quarterbacks -- Derek Carr, Lamar Jackson, and Nick Mullens -- tied for 11th in deep accuracy, though it should be noted that Jackson and Mullens threw only 14 passes each in comparison to Carr's 40, so their sample sizes are much lower.
I couldn't tell you how Blake Bortles of all quarterbacks finished in the top 15 in accuracy percentage, and as far as I'm concerned it's an outlier. He's a terrible quarterback who somehow managed to have enough accurate passes to finish above average in the 2018-19 Deep Ball Project. Even more bizarre is how he finished higher above famed Game of Thrones cameo actor Aaron Rodgers. Rodgers' accuracy did fall off in the last several games of 2018 after another impressive stretch early, but a knee injury suffered in Week 1 hampered him all season long.
Dak Prescott's accuracy took a huge dip after he finished in the top five in deep accuracy in 2017. Part of it had to do with being in a limited scheme that suffocated his skill set, but he was missing a larger amount of throws than he had in seasons prior.
Moving on down the list one disappointment, at least from my perspective, is Cam Newton's placement in the rankings. In seasons prior, Newton's downfield accuracy had been terrific in spite of playing behind some pretty lackluster offensive lines. Knowing what we know now, though, the injury to his throwing arm plagued his comfort, and as a result his accuracy took a huge dip in quality. Here's hoping he heals up in time to start practicing for the 2019 season and return to form.
Tom Brady is going to be 42 this season, so honestly any criticism of his deep passing would just be gratuitous. I just about expect any decline from a quarterback in deep accuracy when they get to Brady's age, and besides, his legacy is already set so it's not like this will be a strike on his career.
Fellow AFC East quarterback Sam Darnold was the least accurate deep passer of any rookie quarterback, but I'm not going to be too harsh on him. He only just turned 22. Sure, he had his share of issues, but underneath all the growing pains expected for a rookie I do think there is potential for him to be a star. It's just that I'm skeptical if Adam Gase is the right coach to help give him some help.
Ben Roethlisberger's deep accuracy was pretty bad in 2017 and only got worse in 2018. Ryan Tannehill, who was actually excellent in the 2016-17 edition of The Deep Ball Project, struggled massively to say the least. Knee injuries from the previous two seasons plagued him, and he was never the most outstanding quarterback to begin with, but at this point he was an even worse version of himself, so it was inevitable that the Dolphins would move on from him.
But it's Jameis Winston who really commands the bottom of this list as the least accurate deep passer of the 2018-19 Deep Ball Project. I'm not going to put every single woe the Buccaneers have experienced in the Winston era on him, as the team is severely flawed (excluding the receiving corps, of course). But Winston's inaccuracy has plagued the offense enough to the point where he was getting benched for Ryan Fitzpatrick.
Twenty-four quarterbacks from the 2018-19 Deep Ball Project were also in the 2017-18 edition. That edition features throws of 16 to 20 air yards, whereas the 2018-19 edition does not, so comparisons of accuracy are a bit misleading, but I thought I'd take this time to look at how these quarterbacks did in accuracy percentage in 2017 and 2018.
The below chart will look at accuracy differential. Positive differential (2018 accuracy percentage minus 2017 accuracy percentage, and also 2017 ranking minus 2018 ranking) is represented in green, while negative differential is highlighted in red.
|Deep Ball Project Accuracy Splits (2017 vs. 2018)|
|QB||2017 Acc%||2018 Acc%||Dif|
As you can see, Derek Carr had the highest accuracy percentage differential; he jumped from 30th in deep accuracy percentage in 2017 to tied for 11th in 2018. Deshaun Watson's deep accuracy wasn't great, but he finished above average in accuracy percentage despite playing without Will Fuller for nine games and behind one of the league's worst offensive lines.
On the opposite side of the spectrum guys like Aaron Rodgers and Cam Newton saw their accuracy rankings regress. In both cases injuries hampered their comfort levels, especially in the latter half of their seasons.
Ranking lower in accuracy percentage in 2018 than in 2017 isn't necessarily a bad thing. Marcus Mariota and Drew Brees saw their rankings decrease, but they still finished in the top ten for deep accuracy. Also, since the 2018-19 edition includes only throws of 21-plus yards, the stats are completely different.
With that out of the way, let's look at how every quarterback ranks in deep accuracy percentage across two distances: 21 to 30 yards, and 31-plus. (Side note: The order of the charts are decided by how the quarterbacks finished in total accuracy percentage from best to worst.)
|2018-19 Deep Ball Project: Throws by Distance|
|21 to 30 Air Yards||31-Plus Air Yards|
While Deshaun Watson struggled with his accuracy throwing 31 or more air yards, he was the most accurate passer on throws of 21 to 30 air yards. Again, this was done without Will Fuller for a majority of the season. Lamar Jackson only threw nine passes of 21 to 30 air yards, but he finished second in accuracy in this area. Interestingly enough, Andy Dalton also finished in the top five in accuracy in this area.
Russell Wilson surprisingly didn't do as well as expected on throws of 21 to 30 air yards, but he was the most accurate quarterback on throws of 31-plus air yards. On the flip side, Drew Brees finished in the top five on throws of 21 to 30 air yards, but second last on throws of 31-plus air yards. This is a little concerning in my opinion because of how excellent Brees had been as an all-around downfield passer in previous seasons.
Sam Darnold's accuracy on throws of 31-plus air yards was excellent, as he finished fourth. He flashed plenty of potential in the last quarter of the 2018 season, and I look forward to seeing how far he can progress in 2019. Kirk Cousins' accuracy in this area was also high, as he finished fifth.
Now let's get to the areas of the field, including left, middle, and right.
|2018-19 Deep Ball Project: Throws by Direction|
While Patrick Mahomes had his issues throwing to the right, he was the most accurate deep passer on throws to the left, and finished in the top ten on throws to the middle. Dak Prescott finished first in accuracy to the middle, though it's worth noting he only threw two passes here. Blake Bortles of all quarterbacks finished first in accuracy on throws to the right, a demonstration that a lot of strange things can happen in small sample size.
Joe Flacco was surprisingly accurate throwing to his left, finishing in the top five himself. Mitchell Trubisky also finished in the top ten in accuracy throwing to his left, which will no doubt be used to combat the narrative that he's incapable of doing so. Josh Rosen and Sam Darnold struggled throwing to the middle, while the right of the field was unkind to Cam Newton and Ben Roethlisberger, who tied for 33rd in accuracy.
Next up are throws in clean pockets and under pressure.
|2018-19 Deep Ball Project: Pressure|
|Clean Pockets||Under Pressure|
The top two quarterbacks in deep accuracy dominate the top two spots throwing from a clean pocket. Despite Cam Newton's struggles with injuries, he managed to finished third throwing inside clean pockets. Guys like Ben Roethlisberger and Jameis Winston, on the other hand, did atrocious throwing from clean pockets, but unlike Winston, things didn't improve that much for Roethlisberger under pressure.
Jared Goff's splits with and without pressure show how reliant he is on good pass protection in comparison with his peers. He finished tied for fifth in accuracy throwing from a clean pocket, but was only 28th throwing against pressure.
As you may have noticed, many of the top spots for accuracy under pressure are occupied by quarterbacks who can play outside structure, including Andrew Luck, Baker Mayfield, Patrick Mahomes, Dak Prescott, Deshaun Watson, and Mitchell Trubisky. Lamar Jackson finished first in accuracy under pressure but only threw three passes here. Kirk Cousins was surprisingly great in his downfield accuracy under duress, finishing second.
Finally, here are the accuracy numbers throwing into both open and tight windows.
|2018-19 Deep Ball Project: Open and Tight Windows|
|Open Windows||Tight Windows|
The quarterback class of 2012 dominated the top slots for tight-window accuracy, as Andrew Luck, Russell Wilson, and Kirk Cousins all finished in the top three. Accuracy into open windows is far less spectacular obviously, but Nick Foles was the only quarterback to be accurate on every single throw here, and Ryan Fitzpatrick finished second. Aaron Rodgers finished in the top five as well.
Jared Goff also placed in the top five in tight-window accuracy, while Josh Rosen was surprisingly good in this area as well. Carson Wentz did well in both areas, as did Russell Wilson, Matt Ryan, and Baker Mayfield. Patrick Mahomes was surprisingly low in tight-window accuracy, but his playmaking ability and overall quality outside easily offset this.
In terms of open-window accuracy, this was a place where guys like Marcus Mariota, Jared Goff, Kirk Cousins, and Dak Prescott all struggled. A few of the rookies struggled here as well, which is to be expected due to their lack of experience.
John Kinsley is on Twitter @brickwallblitz. He has also written for Music City Miracles on SB Nation, Dynasty League Football, and The Riot Report.
6 comments, Last at 29 Jul 2019, 3:21pm
#1 by RobotBoy // Jul 28, 2019 - 2:24am
The analysis is much appreciated and quite useful. I'm left with some questions though.
In the article, you state, 'Accuracy percentage can differ greatly from completion percentage. Aside from straight up drops, accurate passes are counted on incomplete passes if the receiver can't keep two feet in bounds on a pass he should have had no problem hauling in, some Hail Marys, some pass disruptions (depending on the placement), etc.
On the other hand, completions where the receiver has to make an unnecessary adjustment are punished as inaccurate passes. Back-shoulder passes are not included in here for obvious reasons, but passes where the receiver has to go out of his way just to make a catch are.'
Obviously, this leaves a lot of room for subjective judgement. I'm curious as to how you see your deep ball rankings as they stack up against other outlets. For example, your 2017 rankings differ quite significantly from Doug Farrar over at BR.
Deep Wall blitz agrees with you about Brady. PFF, however, ranks him 12 in downfield throws.
PFF also had Derek Carr ranked as #1 in deep throws.
Finally, you say back shoulder throws aren't included, does that just mean in terms of accuracy rating?
#2 by Will Allen // Jul 29, 2019 - 11:00am
As is it is so often the case in football analysis, the sample sizes simply aren't large enough to draw strongly confident judgements. I say that as someone whose subjective judgement agrees with the guys placed at 1&2.
#3 by nat // Jul 29, 2019 - 11:21am
I appreciate the work put into this. But I remain cautious about any subjective decisions about accuracy. Especially on long passes, "accuracy" is largely a function of the receiver's ability to see and adjust to the pass quickly.
I remember a pass from Brady to Moss in 2007 - it might have been their first game together. Moss was mired in triple coverage, and possibly running the wrong route. Brady chucked the ball deep to the far side of the field. To my eyes, it looked like a throw made in frustration either at Moss's ad libbing or the coverage. But Moss immediately saw the pass, made the necessary cut, streaked across the field, and caught the pass in stride for a quick few steps to the end zone.
After the game, Brady happily commented that the play "wasn't the way we drew it up".
Was it an "accurate" pass? Was it "thrown where only the receiver could get to it"? Or was it an "unnecessary adjustment"? Who the heck knows?
But if a pass can be that far away from a receiver and still be caught in stride, how can "accuracy" be consistently judged - and blamed on the throw - on such long passes?
#4 by sbond101 // Jul 29, 2019 - 12:24pm
I remember that throw against the Bill's, it's not even clear it was intended to be triple coverage on the Bill's side but the safety on the other side of the field enters coverage when he sees the ball and where Moss is going. I think it's correctly illustrative of the illusion that accuracy can be evaluated from the result of the play - the result of the play is a mixture of accuracy, how the defender(s) & receiver played the ball, whether the QB decision was intelligent etc.. It was so odd because I've seen Brady appear to throw a deep ball to kill a play a lot, to see it end up completed was so strange and was part of creating the feeling that the 07 Pats could do anything. I felt somewhat the same way watching Joe Flacco's receivers create something from nothing with deep balls over and over again (with a lot of ref help) in 2012, or watching Big Ben bounce off sacks for years; wondering "if that can happen, are any of my assumptions about football/watching valid".
All that said, I agree wholeheartedly with the skepticism of meaningfulness of the data presented here.
#5 by nat // Jul 29, 2019 - 2:06pm
The one I was thinking of was against the Jets. But I think there were a few like that in 2007, so I believe you that it happened against the Bills, too.
My point is not to denigrate this analysis. It's more to illustrate that what constitutes "accurate" can vary a lot from receiver to receiver. If Chris Hogan (for a random, not-too-bad player Patriots example) had been the receiver on those plays, (a) Brady wouldn't have made the throw where he did, and (b) if he had, Hogan would have not reacted in time to get within five yards of it. The same pass that is perfect for Moss is horribly inaccurate for Hogan. The same goes for receivers across the league.
There are certainly inaccurate QBs in the league. Just as certainly, there are receiving corps that make their otherwise-accurate QB inaccurate.