A Data Driven Look into Offensive Efficiency in the Passing Game
I was recently considering how to use statistics to determine the overall skill and strategy of offenses in the NFL. As DFS players, we are very good at gauging what kind of opportunity a wide receiver, running back, or tight end will get on a given week through synthesizing through news, snap counts, and other relevant data. We also are pretty good at evaluating match-ups. We can tell through personnel and history, that team x might be bad against WR 1s, and that WR y will likely have a better game because of this. We can tell which offensive lines are going to open up holes for RBs, and for the most part, we can tell which QBs are good and which QBs are bad. But when it comes to overall offensive strategy and coaching, I get stuck. What is good offensive strategy? How do we know it’s good? How do we know whether the players are making an offense great, or the scheme is? These are questions that for me, especially for someone with 0 football background, are very tough to answer.
On my quest for more relevant and predictive stats for fantasy, I came across something that may have gotten me a little closer to this answer. A writer at RotoViz.com (one of the sharper DFS strategy sites around) created a simple but useful stat called Air Yards (which you can find on the aptly named airyards.com). The statistic is a crunch of two numbers, aDot (average depth of target) and targets, to synthesize a stat that shows the total value of targets a receiver is getting, something that is useful from a fantasy perspective. 1 target 1 yard down the field does not equal 1 target 50 yards down the field, separating these out is useful to predict how many yards a receiver may get in the future based on his targets. Air yards does just that.
One other useful stat the comes from air yards is something the creator of this website calls RACR. RACR is calculated by taking actual Receiving Yards and dividing it by Air Yards (Receiving Yards/ Air Yards), Essentially, it’s a stat that calculates the efficiency of each Air Yard. If the QB throws it 1 yard down the field, and the receiver takes it 5 yards for a 6 yard gain, that’s a very very efficient and easy play. If a QB throws 15 yards down the field on a comeback, and the WR grabs the ball, immediately gets tackled, and gains 13 yards total, that’s a much more difficult throw that and not as “efficient” (but, obviously, more productive).
Both of these stats lead me back to the question of what is good offensive strategy in the pass game and my theory about the answer. Is the best offensive strategy one that minimizes the yards thrown downfield and maximizes receiving yards gained?
Quarterback clearly matters quite a bit to answer the question about proper offensive strategy. If you have an elite quarterback, let’s say someone like Aaron Rodgers, who has a lot interception rate on balls thrown downfield and a high completion rate, then limiting him to passes only 5 yards down the field seems ridiculous. But if you and interception prone QB with a low success rate down the field, someone like Blake Bortles, his most efficient throws are going to be closer to the line of scrimmage, something like passing to a RB out of the backfield and having him get some YAC.
Clearly, this question about good offensive strategy is not simple. But with everything written above mind, I went on a quest using Python to discover which offenses we’re difficult to execute, which we’re most productive, and see if I could answer some questions about which offenses we’re truly efficient, and why.
One limiting aspect of air yards is that it only tracks vertical, down-the-field yards. It doesn’t take into account horizontal yards. Throws toward the sideline are going to be much harder to execute than throws down the middle of the field. I found a simple solution to this by weighting air yards depending on the position of the player thrown to. For a WR, our QB likely is going to have to throw more towards the sideline, so I adjusted those air yards by 1.3x (by the way, a better way to do this would be to weight slot receivers less and outside receivers more, but that was a bit too complex of an undertaking). For TEs, who for the most part will catch balls closer to the middle of the field, I kept air yards the same. And for RBs, who are almost exclusively getting thrown to in the middle of the field, on screen passes and as bail outs on broken plays, I weighted .7x.
Then, I grouped this new, truer, air yards stat by team and divided it by overall team targets, making a new stat that gauges how many yards in the air are thrown on each pass. The results are in the graph below.
The Top 5 air yards per target: The Jets, The Panthers, the Buccaneers, the Steelers, and the Bills. The Steelers have one of the best QBs in the NFL, so it makes sense to have an offense geared around letting him “air it out.” With the Jets on the other hand, with a mediocre at best QB in Ryan Fitzpatrick, this strategy caused them a lot of trouble. Fitzpatrick had a terrible year and the Jets passing offense ranked in the bottom 5 in many statistical categories, including 2nd to last in turnovers at 2.1 per game.
Interestingly enough, the Panthers passing offense and Cam Newton struggled quite a bit last year. The Bills were fine. And the Buccaneers were pretty good (Seeing them up here actually has me more bullish on Jameis Winston since he performed well in a tough offense, but that might be for another article). Overall, the results bode poorly for anyone who argues that having your QB throw the football far down the field on most plays will make a productive offense.
So who we’re the most productive passing offenses? I grouped the teams below by receiving yards per target.
That tall bar to the left is Atlanta, clearly a cut above the rest on offense last year in the NFL, which I was happy to see since it aligns with what we already know to be true. Atlanta was a historically good offense last year. But look at a couple of teams that round out the top 6: New England and New Orleans. Two teams in the bottom 6 of air yards per target. Essentially, these two teams had their QBs throw the least distance, but they gained the most yards per throw almost every offense in the NFL. Sounds like efficiency to me.
Now, I’m not saying that Tom Brady and Drew Brees aren’t amazing, because they are, but I think this goes to show that their coaches also put them in a very good position to succeed. Passing to RBs is a huge part of both these offenses, which makes the QBs job a lot easier. And both pass quite a bit to the TE, with Brady having a great weapon in Gronk to go to. Both these teams also clearly didn’t force it down field when they passed it to their WRs. Overall, it seems like a combination of great QB play, talent, and strategy makes these two offenses consistently two of the best in the NFL.
Contrast this with the Jets. The Jets had absolutely 0 production from the TE last year, so Ryan Fitzpatrick didn’t get to enjoy some of those easy middle-of-the-field throws at other QBs do. When they threw the ball to their WRs, they didn’t make it easy, throwing it far downfield. They did have production from Bilal Powell in the receiving game toward the end of the season, but it seemed like they only used him in this way when trailing. Yes, this offense did not have a lot of talent and their QB was not very good, but clearly the offensive scheme contributed significantly to their struggles.
One last thing I want to point out about NFL offenses and efficiency. I made the point above that passing to RBs and TEs is a lot easier for the Quarterback, but is it as productive? Again, this is a complex question to answer, but here’s a stat that helps the argument. Below is the RACR per target clustered by position:
RB: 2.4 RACR
TE: 1.15 RACR
WR: 0.8 RACR
RBs gain 2.4 receiving yards per air yard, while WRs only gain 0.8. Essentially, RBs gain 3 times the receiving yards per air yard as WRs, and the throw is much, much easier. Obviously, an NFL offense can not simply just pass to the RB every play, otherwise their opponents will just stack the box and these plays would become less productive. There are also some situations, say a 3rd & 10, where getting 5 yards on a 2 yard pass isn’t helpful, and we must throw it down field. But it seems like the data suggests that these high efficiency, easy throws to RBs likely should be used a little more, and the great offenses seem to already be doing it.View all posts by Max J Steinberg