Incomplete passes suck.
As an NFL fan or a fantasy football owner, you know this to be true. When you see the ball hit the turf, or see it land in the hands of an opponent, it's a moment of intense disappointment. Knowing "what could have been" on a passing play is similar to seeing your running back trip in the open field, except that it happens on 30% to 50% of the attempts your quarterback makes.
And from a statistical perspective, most incompletions are created equal. Using numberFire's Net Expected Points (NEP) data (which you can learn more about here), we can find the difference between a player's Reception NEP (expected points earned and lost on receptions only) and Target NEP (expected points earned and lost on every pass targeted at the receiver) to find out how many expected points were lost, in sum, on incompletions.
Since 2010, wide receivers who see regular playing time (defined as having more than 30 catches) have lost around 1.02 Net Expected Points when they're targeted on a play and do not catch the football. In other words, when a quarterback and receiver don't connect on a passing play, that team loses one expected point.
But what happens when we introduce the enemy of efficiency: volume? Highly-targeted receivers see more throws than the normal receiver, and see them against better defenders and in tougher situations. Brandon Marshall's targets against top cover corners aren't the same as a tight end who gets open in the middle of the field because no one covers him, but they count toward this average all the same. That's why it's surprising to learn that, in 2013, the average per-target loss among receivers with more than 140 targets (there were only 14 such receivers) was nearly identical (within one hundredth of an expected point) to the three-year average among all receivers with more than 30 receptions in a season. Marshall in particular was very close to the average, despite seeing tighter coverage, double teams, and being thrown the ball at an above average rate.
But as is the case with almost any statistic in football, there are outliers to be found here. Three receivers among the highly-targeted group stand out as having well above average losses per non-catch, while three posted very efficient numbers on the average incompletion. So which receivers hurt their offenses the most when they couldn't haul in the catch, and which limited the damage the most on a per-incompletion basis? Let's begin with the bad news.
The Force Fed Failures
Before I get too far, allow me to explain that the above average NEP loss on incompletions for these top receivers is not an indictment on the receiver's ability. No one is going to argue that A.J. Green, Andre Johnson and Calvin Johnson are bad receivers, but rather that the way they're used in their respective offenses has led to an inefficient trend forming in their targets.
The fact that those three receivers were the ones who saw the biggest per-incompletion drop in NEP is more damning to their offensive coordinators and quarterbacks than it is to them as wideouts. Green and Calvin in particular are seen as two of the best receivers in recent history, and yet their offenses finished 11th and 12th respectively in Adjusted Passing NEP in 2013. Andre's Texans finished 30th within that metric.
So what can we learn from this? It varies from team to team.
In the case of the Texans, it was simply an issue of feeding Johnson too heavily, and not relying on other targets. Johnson saw 90 more looks than the next closest receiver (DeAndre Hopkins), and his NEP loss per incompletion was 12% worse than Hopkins'. The other targets in Houston were all somewhat average on a per-incompletion basis, which meant the disaster that was the Texans' passing attack mainly happened as a result of throwing to their best player.
For the Lions, it's a somewhat similar story. Calvin Johnson and fellow wideout Nate Burleson saw significantly worse average losses on targeted incompletions than did Kris Durham or any of the tight ends or backs that regularly featured in Detroit. Matt Stafford's overreliance on Megatron and the occasionally healthy Burleson was his downfall, as even missing on a target to the drop-happy Brandon Pettigrew didn't hurt the team as much as throwing to the best receiver in the league.
For the Bengals, the differences between downfield targets and short targets is even greater. A.J. Green's per-incompletion NEP loss rate paled in comparison to the NEP lost on the average missed throw to Marvin Jones, while Andy Dalton's misfired attempts in the direction of Gio Bernard, Jermaine Gresham and Mohamed Sanu came in well below average compared to the rest of the NFL's receivers. But over 250 of the team's targets went to Jones and Green, and the results were fairly catastrophic.
The common thread among all three teams is that not every target in the passing game saw poor per-incompletion averages. The average incomplete pass thrown to any receiver not named Green or Johnson across the three teams lost exactly one expected point, nearly the same as the three-year league average. So for these three teams in particular, force-feeding the best player at receiver wound up hurting more than it helped.
Narrowly missing the top-three in this less-than-prestigious category were Vincent Jackson of the Bucs and Pierre Garcon of the Redskins, and for both of these players, the story remains the same. Bad passing offenses trying to force the ball to the only playmaker they had available, and with frustrating results.
So while it may seem like the logical thing to do, these numbers prove that simply throwing in the direction of the best receiver on the team over and over again can hurt more than it helps. A color commentator on Sunday may say that you want to "put it up there and let him go get it," but in the case of the Texans, Lions and Bengals, it would have been helpful to share the wealth a bit more. Especially in the Texans' case, where Andre Johnson wasn't even producing the most on a per-target basis on his team (Hopkins had more Reception NEP per target than Johnson in 2013).
As we'll see in the next section, there are some teams that can get away with (and even benefit from) targeting a receiver heavily. In fact, five of the top 10 passing offenses in 2013 (according to Adjusted Passing NEP) had receivers found among this heavily targeted group. But in four of the five cases, there were below average losses on the average incompletion.
But specifically in the case of the Bengals and Lions, this data shows that simply having the best receiver possible does not supersede inefficiencies in play-calling and quarterback play. Top 10 passing offenses in San Francisco and Seattle did so without a dominant force at receiver, finishing ahead of the teams with the two best receivers in the NFL.
So yes, incomplete passes suck. But not every incomplete pass is created equal. And in the case of these three players and teams, some changes are needed to get the most out of their dominant athletes at receiver while not sabotaging their offense to force feed a playmaker.
The Heavily-Targeted Hero
Now that the bad news is out of the way, it's time for some good news. There was one player who saw more than 140 targets in 2013 and stood out as having well below average NEP losses on the average incompletion. Julian Edelman led the highly-targeted receivers with the lowest average loss in NEP per incompletion. The average pass thrown to Edelman that didn't connect cost the Patriots over 0.3 expected points less than the average misfire to a receiver for any other team. Going back to 2010, three of the eight most efficient incompletion targets have been Patriots' possession receivers (Wes Welker in 2011 and 2012, and Edelman in 2013).
So why do players in this role not see a huge loss in NEP when targeted on an incompletion? Tom Brady seems to be the common thread and obvious answer, but depth of target and play-calling also factor in. Edelman and Welker were being used on every down and in every distance situation, and weren't heading deep downfield where there's more risk for interceptions. Instead, they provide a reliable set of hands on a short route, and even when they don't haul in the pass, the ball falls incomplete rather than heading to the opponent. They move the chains on the regular basis, which boosts their Reception NEP and Target NEP, so the lost NEP of an incompletion pales in comparison to the consistent gains they provide as a possession receiver and a first down earner.
Another reason for this is the generally low Reception NEP earned by these players on a per reception basis. They see a very high amount of targets, and make a high rate of catches, but don't earn a ton of Net Expected Points along the way. There's an old football joke that features a big, slow running back who, when asked to get three yards for his coach, gets three yards, but when asked to get four yards, still only gets three yards. The same is true of Edelman and Welker, as they're not gaining big chunks of yards, and have a smaller impact on their team's expected points (both positively and negatively) as a result.
What Have We Learned?
The passing game, by its very nature, is boom-or-bust. The average quarterback's NEP is much higher than the average running back's NEP, because the opportunity for a huge play that brings with it a massive NEP boost is higher in through the air than it is on the ground.
And based on the data we've considered here, this is proven at the receiver position, where big play receivers tend to produce the most, but also present the biggest risk as targets.
Edelman, Wallace, Brown and Thomas all saw a lot of shorter looks in the passing game, and while they did break some for big gains and scores, the average target for these players was lower risk than it was for Green, Johnson and Johnson. Edelman and Thomas also get the benefit of playing with two of the best quarterbacks we've ever seen, while Andre Johnson in particular had to deal with a laughingstock under center this past season.
But overall, the point of this research isn't to dissuade teams from targeting their best receivers. Instead, it's to consider the situations in which heavily targeting a receiver can be a good thing, and the times at which it should be reconsidered. Based on this chart, it seems as if force feeding a top target as a deep threat (C. Johnson, Green, Graham), or over-targeting a player with a bad quarterback (A. Johnson, Garcon, Jackson) will lead to a less efficient result, while heavily targeting players in the short passing game, or having a good quarterback (Thomas, Edelman, Bryant) increases efficiency.
Incomplete passes still suck. Teams are throwing away expected points by the handful every game via incomplete passes. But it's a game of risk and reward, and the reward of a successfully executed passing play outweighs the risk of an incomplete pass. Teams just need to balance their offense in such a way as to limit risk of interceptions and impossible down-and-distance situations to get the most out of their elite targets in the passing game.