NFL scoring is down. Here’s why: Part 1

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

In 2020, the NFL set a record high in scoring, averaging 49.6 points per game (PPG). However, since that peak season, scoring has retreated each of the following 3 years, falling to 43.5 PPG, a mark lower than 2008.

That may not seem that remarkable, but prior to 2020, the largest scoring drop was -2.1 PPG from 2016 to 2017 and this recent drop is almost three times as large (-6.1 PPG). Also, in the last 24 years, PPG had not fallen 2 consecutive years let alone 3. One year is not a trend, two years may be a coincidence, but three down years definitely has my attention.

So, what is driving this? Maybe nothing. 2020 was likely an outlier year of high scoring and 2023 was probably depressed due to a myriad of QB injuries. So, perhaps this is much ado about nothing. Hmmm, yes, perhaps.

Or maybe, like most things in life, the answer is far more nuanced with many different contributing variables, many of which are indiscernible. Instead of making up narratives and being philosophical about ultimate truth being unknowable, I am just going to calculate the objective measures of offensive production and see if that paints a picture.

To help frame the situation, I will go back to 2000, not because that year has any significant meaning, but rather because that is as far back as my play level data goes. This will allow me to break the data into different quarterback “eras” based on scramble rates reflecting the evolving role of the NFL QB. I am also going to convert Points per Game to Offensive Points per Drive (PPD) because that is a less biased measure of scoring (1).

2000 – 2009: The Pocket Passer

There were certainly athletic running QBs in the first decade of the century, Michael Vick, Donovan McNabb and Daunte Culpepper, to name a few, but the primary offensive weapon was the pocket passer. The best offenses were led by names like Manning, Rivers, Brady, Brees and none of them ran around much.

This chart shows that QB scramble rates (red line) in those years were low and flat over time. Points per Drive, however, trended up year over year as these passers got better at moving the ball.

2010 – 2017: The Read Option

With the rise in popularity of read option offenses, QBs like Keapernick, Newton, RGIII and even Tebow (ugh) had life in the evolving NFL. Eventually, the popularity of the read option faded away, but the recruiting of physical QBs that led those types of offenses did not.

With athletic QBs becoming more popular, the rate of QB scrambles jumped to a significantly higher level, but then curiously stayed relatively flat. Meanwhile, scoring made its steady march higher (that is until the aforementioned 2017).

2018 – 2023: The Running QB

The most recent years have seen run-away scrambling rates (pun kind of intended) and by 2020, league scoring hit a record high 2.35 PPD. That year, the league experienced:

• the best rushing success (adj Rush Success Rate)
• the 2nd lowest sack rate
• the 2nd highest passing yards per attempt
• the highest first down conversion rate
• the highest TD rate
• the lowest Turnover rate

It was everything an offensive coordinator could want. Athletic QBs were running away from sacks extending plays and turning them into first downs and points. They were making chicken salad out of . . . not chicken salad.

But then 2021-2023 happened and all the gains of this period seemed to have been “given back”, dropping from 2.35 PPD down to 2.01 PPD, lower than it was in 2016. It is this 3-yr period that I want to focus on to quantify where that 0.34 PPD went.

Play Efficiency

There is a strong correlation between points scored and the value of an offense’s average play. So, to find missing points, look to missing efficiency.

Years of analysis has led me to rely on Expected Points Added per dropback (EPA/d) to measure passing efficiency and my own adjusted Rush Success Rate (aRSR) to value rushing efficiency. This next graph shows those metrics since right before the last significant scoring drop in 2017 and compares that to PPD (gray dashed line).

To fit all 3 measures on a single graph, I converted the values to z-scores, which are the units displayed on the y-axis, but you can ignore those and just look at the trends over time.

The purple line is aRSR, which includes QB designed runs. The last 4 years were some of the most successful rushing the league has ever produced. 2021 was the 6th best aRSR ever, and 2020, 2022 and 2023 nab the top 3 spots of all time. In other words, rushing has never been more successful and that does not correlate well with a fall in scoring.

The green line is EPA/d, which includes QB scrambles. It hit a league high mark in 2020, but by 2023 dropped by -0.066 points to +0.049, which is the 2nd lowest value since 2010, the same year scrambling rates jumped.

Since the drop in EPA/d was large and the changes are in lockstep with changes in PPD, it is pretty clear that the recent decline in scoring is driven by a decline in passing quality. Now, assuming that 100% of the changes in scoring is due solely to passing is certainly a false premise. However, since 2000, the correlation between EPA/d and PPD has been +0.83 and over the last 4 years it was +0.99. So, my assumption may be a little bit wrong, but it is mostly right and I’m OK with that.

So, if I want to know where -0.34 PPD went, I need to know where -0.066 of EPA/d went. Keep that number in mind, because I reference it a lot.

Non-Yardage Events

There are play outcomes in football that have much more value than the yards used to measure them. For passing, two big ones are touchdowns and turnovers. The following chart shows the TDs and TOs per dropback since 2016.

The Turnover rate has crept up slightly in the last few years, but not by a lot and the 2023 number is still the 6th lowest value since 2000, so turnovers are not a big factor. In fact, the data show that for passing turnover plays (INTs + sack-fumble + scramble-fumble), the average EPA actually increased a small amount (+0.004) over 2020’s value.

On the other hand, the TD rate in 2020 was at a record high and since then has retreated to a 15 year low. So fewer TDs are assuredly a contributing factor. Isolating the play data for passing TDs (passes + scrambles), yields an average EPA value that is -0.014 lower than 2020’s number (0.121 vs 0.107). That -0.014 represents 21.1% of the missing -0.066 EPA/d drop.

This segues to another non-yardage event. It is tough to get to the endzone if you can’t get first downs. This next chart shows a decline in the rate of first downs over the last 3 years. The 52.8% mark in 2020 was a league record high, but since then it has dropped to the 5th lowest rate since 2000.

The average EPA difference between 2020 and 2023 on first down plays (TDs excluded) is -0.026, which is another 39% of the missing -0.066.

Passing Efficiency

Yardage efficiency is a big driver of scoring. The all-encompassing passing measure for that is Net Yards per Dropback and after rising steadily for years, it has recently dropped off a cliff.

The magnitude of the drop from 6.62 in 2020 to 6.24 in 2023 is unprecedented and I have to go back to 2007 to find a lower NY/d value.

Dropbacks include TDs, TOs and first down plays, so from an EPA-perspective I have already accounted for a lot of this decline. But if I remove those plays, the remaining “other” dropbacks have average EPA that is -0.030 lower than the same “other” plays in 2020 and that accounts for 45.7% of the average EPA reduction.

Conclusion

Here is a summary of the EPA impact:

TDs, TOs and first downs combined account for about half of the drop in scoring, with the remaining amount driven by reduced yardage efficiency on other dropbacks. In other words, QBs are averaging fewer yards on pass plays, which prevents them from moving the ball down the field and finding the endzone as often.

So, the general take-away is that scoring is down because passing is down, but that’s not a satisfying answer because it just leads to the question of why is passing down? Since dropbacks include attempts, sacks, scrambles, throw-aways and even pass interference penalties, there are a lot of variables to investigate that could better explain why efficiency is so depressed.

However, that will take far more than one article to explain, so I will address that in future installments of this series.