NFL scoring is down. Here’s why: Part 2

Thanks to Pro Football Reference, Pro Football Focus, NFL.com, NFL Next Gen Stats and the nflFastR project for being awesome sources of data.

Previously, I analyzed the recent drop in NFL scoring, which peaked in 2020 and for three straight years has fallen back to levels that are now lower than in 2008.

I showed that this unprecedented drop in scoring is highly correlated with a reduction in passing efficiency, which is reflected in metrics like Net Yards per Dropback (NY/d).

In other words, scoring is down because passing has been getting worse. Notice that NY/d first peaked in 2018. That year will come up a lot.

In this analysis, I will dive further into the data to see if I can determine what is causing the decrease in passing quality. To do that, I will split NY/d into two mutually exclusive categories that measure distinct aspects of QB play:

1. Passing Ability – dropbacks that end with a passing attempt
2. Abandon Frequency – dropbacks that surrender the attempt (e.g. sacks, scrambles)

For this analysis, I will focus only on the passing ability portion and leave abandon frequency for part 3.

Yards per Attempt

This means I will be focusing on the portion of NY/d specific to attempts, which is Yards per Attempt (YPA), or at least a modified version of it. My calculation of YPA removes throwaways and adds in accepted defensive passing interference calls, so my numbers will differ from “official” numbers (mine are better).

• YPA = (Passing Yards + PI Yards) / (Attempts – Throwaways + PI Calls)

The overall YPA curve looks similar to NY/d, which is not surprising. However, there is a clear peak in 2018. So, even though scoring has been down for only 3 years, efficiency on pass attempts started to decline before then.

The YPA trend, by itself, does not tell me much, so I need to peel back the onion a bit more. Basic math tells me that I can split standard YPA into two components, Yards per Completion and Completion % so that:

• Passing Yards / Attempt = (Passing Yards / Completions) * (Completions / Attempts )

Therefore, to understand why YPA is falling, I need to understand how those components are changing.

Component 1: Yards per Completion

Since I am using an adjusted YPA, I will adjust my Yards per Completion too. I can do this by simply adding PI calls. Throwaways are incompletions by definition, so I can ignore them.

• Yds / Cmp = (Passing Yards + PI Yards) / (Cmp + PI Calls)

This is very interesting. Clearly, average completion yardage has been getting shorter for quite a while. So, this alone cannot explain why efficiency rose through 2018 and then suddenly fell.

This next chart splits completion yardage into Air Yards (passing depth) and Yards After the Catch (YAC) to show its composition over time.

Average YAC has fluctuated but is basically flat in the long run. So, the shorter completions are almost totally driven by passing depth.

Component 2: Completion %

For Cmp %, I include accepted PI penalties and exclude throwaways:

• Cmp % = (Cmp + PI Calls) / (Att – Throwaways + PI Calls)

Completion rates have improved dramatically over time. So, was this an improvement in QB accuracy? No.

Too often, Cmp% is associated with accuracy, but that is not necessarily so. Cmp% does not account for passing depth, down, distance, field position etc., which are all variables outside of accuracy that will impact the result.

Therefore, I will incorporate Expected Completion %, which is a measure that calculates the probability of a completed pass given the known variables that impact it.

The expected completion rate curve mimics actual completion rates for the most part, which means that variables other than accuracy are mostly causing the changes to Cmp% over time. In other words, QBs weren’t getting more accurate up to 2018, they were just throwing higher probability passes.

After 2018, expected completion % leveled off and so actual completion % flattened as well.

Combined Impact

So, I now know the drivers are passing depth and completion probability and that when combined the trends completely explain the sudden drop in efficiency.

Through 2018, QBs were making increasingly higher probability throws, so the loss of yards from shorter completion depths were more than offset by additional completions, which caused efficiency to improve.

However, after 2018, passing depth on completions kept falling, but not from higher probability throws. So, there were no additional completions to make up the yardage deficit and YPA fell.

So, the only real mystery here is, given shorter throws after 2018, why didn’t completion probability increase? Unfortunately, the exact formula for completion probability is a black-box machine learning algorithm that looks at multiple variables simultaneously, so I cannot point out specific calculations, but I can provide some examples of the variables used and show how they are trending.

While depth is a primary driver in completion probability, other variables like passing location also play a part. Even when accounting for depth, passes over the middle have historically earned a 2.5% higher completion rate than those outside the numbers. However, QBs have been making those center field throws at a much lower rate recently.

Down & distance has an impact as well. Here is a table that shows historical completion rates for downs with

Throws made with 10 yards to go on 1st or 2nd down represent 48.6% of all pass attempts since 2009. Those passes have higher completion rates, but offenses are throwing fewer of them, which shifts the league’s passing profile towards lower probability throws.

There have been significant changes to other passing variables as well, but whatever the exact mix of drivers, the bottom line is that YPA has been falling because passing has been sacrificing depth for lesser quality throws.

SUMMARY

Does that mean I am saying QBs have gotten worse? In some ways, yes. However, the initial driver for the drop-off in passing since 2020 is not QBs per se, but rather adapting defenses.

Most defenses have changed their tactics moving towards more zone coverage with two high safeties. According to a 2022 New York Times article (my emphasis added):

One big factor is that defenses have altered their coverage schemes, particularly by using two-high safeties, to counter the pass-centric offenses that were propelling those high scores. Their strategies are aimed at preventing deep passes that can turn a game on its head and expose teams who cannot run the ball or keep drives alive with shorter throws.

“I think the biggest thing is just trying to eliminate those explosive plays,” Jets Coach Robert Saleh said last month in a news conference. “And if you do that, you make teams go 10 plays over and over again, it’s just really hard to do, playing them game in and game out.”

The approach is working. Explosive pass play rate is down and QBs are not getting get first downs with their arm, so drives are stalling more often.

In Part 1, I mentioned how recently rushing efficiency has never been better, setting records in rush success rate. Well, that is because defenses are selling out to stop the deep pass by conceding runs and daring QBs to beat them underneath. Unfortunately for offenses, QBs have not countered the defensive changes.

The way to beat this type of defense is with timing routes and finding the open man underneath, often in the middle of the field between the dropping safeties. That was the bread-and-butter skill of the pocket passer, but the new batch of QBs cannot do it as they struggle to get through their progressions. Over the last 3 years, passers have taken a longer and longer time to make shorter and shorter throws that are less and less efficient.

That is a numerical picture of passers who cannot find open receivers.

In my opinion, all of this is a direct result of QBs coming out of colleges that focus more on athleticism instead of players that excel at passing. As our own Chris Shepherd puts it:

“Guys who are athletic rarely have to learn to progress through reads, they can take off when #1 isn’t there. Their college coaches don’t care that much because their goal is to win games, not prepare guys for the draft. If they do both, great, but they have to win to keep their job. So, college coaches are valuing athleticism at the position like never before and athletes often don’t have to learn to work reads/be a QB.”

I have been wary of “running QBs” for this very reason. If you can take a great passer and add running capability, then great, I will take that any day (Mahomes). But that is not what has been happening. Instead, we have QBs who are great runners, but not great passers. They can throw deep, but they cannot consistently move the chains and defenses have gotten wise.

Maybe these QBs will learn to be better passers and the issue goes away. Maybe. Or maybe if your only tool is a hammer, then everything is a nail. I will dive into this more when I address abandoned attempts in part 3.