Why Not All Passes Are Completions

Mike Clay changes the game by digging into the 'why' behind each incomplete pass. Part 1 of his series focuses on the quarterback position.

| 2 years ago
Screenshot 2015-07-03 09.55.15

Why Not All Passes Are Completions

Screenshot 2015-07-03 09.55.15This article will change your life.

Was that too dramatic? Maybe, but if you’re a hopeless football nerd like myself, the data examined in this article will influence the way you evaluate players.

Enter “reason for incompletion.” Basic in term, significant in value.

At PFF, our analysts chart every aspect of every play of every single game. This provides our team with unique and innovative data we can use to take NFL evaluation to a new level. By defining who is at fault for each incompletion (and why), we can better evaluate the effectiveness of both quarterbacks and receivers. It’s often a near-impossible task to separate a quarterback’s play from the production of a receiver and vice versa. With this data, we’re closer than ever to doing just that.

Before I dive into the data, I want to give you a brief understanding of how the process works.

Screenshot 2015-07-03 09.58.03

NFL-wide Pass Attempt splits

There are a lot of questions we can answer by reviewing this data. This article will dive into some of the more important ones, and you can bet I’ll be referring to this data going forward – not to mention the impact it will have on our player projections and fantasy rankings.We start with each and every drop back. From there, we eliminate scrambles, sacks, and spikes, which leaves us with a slightly adjusted pass attempt number (only spikes are removed from the industry figure – there were 73 last season). We follow by splitting attempts into four categories: on-target throws, off-target throws (over and underthrows, poor reads, slips, etc.), incompletions forced by the defense (batted balls, passes defensed, throw aways, etc.), and other (miscommunication, Hail Mary). That’s where the process ends for passers, but we can break it down further for receivers. Each on-target throw is either caught or incomplete as a result of a mistake by the receiver (drop, fall, bad route, etc.). Check out Part 2 of this series for a full examination.

Our first set of player data looks at the quarterbacks who had the highest (and lowest) percentage of their targets disrupted by the opposing defense during the 2014 season. Playoff data is included.

Rk QB Att Defended   Rk QB Att Defended
1 Josh McCown 323 21% 43 Tony Romo 484 10%
2 Drew Stanton 240 20% 42 Ben Roethlisberger 643 11%
3 Derek Carr 593 20% 41 Mark Sanchez 309 11%
4 Ryan Lindley 121 20% 40 Jake Locker 146 12%
5 Geno Smith 366 20% 39 Kyle Orton 444 13%

Josh McCown tops the list, which is intriguing after he was so good on a small sample with Chicago in 2013, but struggled as the starter in Tampa Bay last season. Defenses were clearly tough on McCown, which led to an on-target percentage of 64 percent (sixth-worst). Of course, McCown’s off-target percentage of 14 percent was right at league average, which suggests he wasn’t as bad as his 56 percent completion percentage suggests. If we neutralize the defensive impact, McCown was more average than he was bad. This is a hint of good news for Cleveland.

On the other hand, Tony Romo had the benefit of very little defensive pushback last season. Romo’s 70 percent completion percentage was tops in the league among players who attempted at least 150 passes. Only 3.9 percent of his throws resulted in a “pass defensed” (NFL average = 6.0 percent), 3.1 percent were dropped (5.6 percent), and 2.5 percent were thrown away (3.6 percent). In 2013, Romo’s rates in those three categories were 5.2 percent, 5.8 percent, and 3.0 percent, so it’s hardly a surprise that his completion percentage was six points lower. It’s fair to expect some regression to the mean in 2015.

Rk QB Att On Target   Rk QB Att On Target
1 Drew Brees 653 76% 43 Ryan Lindley 121 56%
2 Tony Romo 484 74% 42 Brian Hoyer 439 61%
3 Ben Roethlisberger 643 74% 41 Drew Stanton 240 62%
4 Robert Griffin III 214 74% 40 Charlie Whitehurst 182 63%
5 Alex D. Smith 464 74% 39 Nick Foles 311 64%

Drew Brees’ imminent demise is a hot topic this offseason, but it’s hard to understand the chatter after such an effective season. An impressive 76 percent of Brees’ throws made it to the intended receiver. That mark is actually higher than his 73 percent rate in 2013. Brees over- or underthrew his target on 4.4 percent of his attempts, which is nearly half the 8.7 percent NFL rate. That is some serious accuracy. As a result, no one was in the vicinity of Brees’ NFL-high 456 completions.

Ryan Lindley was clearly way off target, but he’s unlikely to take another NFL snap. Instead, let’s take a look at potential Texans starter Brian Hoyer. With Cleveland last season, Hoyer overthrew his intended receiver on 39 of his 439 attempts. That 8.9 percent rate was well above the 6.6 percent league average. Hoyer actually benefited from a low drop rate (4.8 percent), but ruined it with poor accuracy across the board. Inevitable garbage time production aside, DeAndre Hopkins’ prospects will take a hit if Hoyer beats out Ryan Mallett.

Rk QB Att Off Target   Rk QB Att Off Target
1 Ryan Lindley 121 22% 43 Drew Brees 653 8%
2 Jake Locker 146 21% 42 Teddy Bridgewater 397 10%
3 Brian Hoyer 439 19% 41 Carson Palmer 224 10%
4 Charlie Whitehurst 182 19% 40 Matt Ryan 623 10%
5 Zach Mettenberger 179 18% 39 Philip Rivers 569 11%
6 Blake Bortles 474 17% 38 Alex D. Smith 464 11%

Blake Bortles had a rough rookie campaign, and it shows up in here in the off-target department. Bortles primarily struggled with overthrows (41, 8.6 percent) and batted passes (17, 3.6 percent), but did have some bad luck in the drop department (31, 6.5 percent). The 17 batted passes tied for the league lead, as did the 55 sacks he took. Unlike Lindley and Jake Locker, Bortles isn’t too far off the 13 percent league average here, so especially with a much-improved supporting cast, there’s some reason to believe he’ll improve as a sophomore.

Teddy Bridgewater had a terrific rookie season. One of the league’s most-accurate passers, 73 percent of Bridgewater’s throw made it to the intended receiver. He was well above average in on-target throws and well below average in off-target throws despite an above-average number of throws disrupted by the defense. Bridgewater underthrew his intended receiver only twice all season. Consider that Tom Brady and Aaron Rodgers paced the NFL with 17 each. Bridgewater did struggle a bit with short-range accuracy, but was nothing short of exceptional in the mid-to-deep range.

I’ve discussed two of the three notable rookie quarterbacks from 2014, so I might as well dig into Derek Carr. I’ve been hard on Carr this offseason, but a deeper look at his splits suggests plenty of reason for optimism. Oakland’s schedule was rough last season and that is reflected in our earlier “defended” chart. Focusing on accuracy (over- or underthrows, in front of or behind intended target), Carr missed 11.5 percent of the time. That’s actually below the 12.2 percent league average. Carr’s underwhelming completion percentage can be traced to a league-high 45 defensed passes (7.6 percent), as well as, a combination of higher-than-average rates in the throw away, batted pass, miscommunication, and close coverage departments. This obviously provides reason for optimism for both Carr and budding superstar Amari Cooper.

As valuable as this data is and will be for player evaluation, our database is only two years deep. With time, we’ll be able to determine how players progress (or regress) in each category. We’ll also be able to determine if this modeling is predictive for players like Carr, who showed fine accuracy, but had his numbers deflated by a poor situation and supporting cast.

Be sure to also check out Part 2 of this series, which focuses on the receiver side of the battery.

Follow Mike Clay on Twitter: @MikeClayNFL and get full access to data like this through the Summer of 2016 by signing up for Fantasy Gold

  • leftsharks

    Out of curiosity, Mike, can you adjust for/refine where on the field the underthrows occur? B/c I feel like the value of underthrowing w/n the 10 yd. line should be weighed less significantly; i.e., Brady & Rodgers ideally want to throw a low ball (esp. if coverage is tight) from the 10 in. Assumption is a low risk, low ball endzone target is more valuable long-term than a throwaway.

    • leftsharks

      Let me qualify that with front-end red zone underthrows. Same would apply to the overthrow at the back of the end zone.

      • http://www.profootballfocus.com Mike Clay

        In terms of quantifying specific errors like under- and over-throws, I get what you’re saying, but it shouldn’t impact on-target vs. off-target very much. At the end of the day, the QB throws a catchable ball or he doesn’t. So, in terms of accuracy or ability for the purpose of this article, ‘QB errors’ can all be thrown in a bucket. Of course, as noted, I do plan on spending some time adjusting for depth and if the results add a new dimensional, I’ll post a study.