Has 2-High Safeties Ruined Football?

There has been a lot of talk (or unhinged yelling) about how 2-high safety looks have “ruined” football. So rather than just agreeing with Mel Kiper (even though he has never ever been wrong), I wanted to take a deeper look at how NFL defensive schemes have affected offensive production over the past 10 years.

The main offensive statistics I wanted to analyze are total points, total yards, turnovers, and explosive pass plays. I chose these stats because they have the strongest correlation to a team winning the game. My goal is to see how these offense statistics change when compared to 2 high, 1 high, man, and blitzes.

*Before I get into this, it’s important to note that this is very small sample size. This would be much more insightful if you compared these stats against each specific type of defense rather than league averages through the years. However, I don’t have access to that level of detail, and it would take me an immense amount of time to gather that data.

Total Scoring

As most football fans know, scoring has been down the past 3 years. From 2014 to 2021, the league average for points was 23 points. Compared to 2022-2024, which has an average of 21.6 points. Mel Kiper combined this information with the fact that 2 -high defensive looks are about 10% the past couple season to come to the harsh conclusion that we should ban 2-high defenses. He referenced the 60s-80s era of NFL offenses multiple times and how exciting those times were. Well during those “exciting times,” the league average for points was 20.6, a full point less than where we are at now.

I gathered some defensive scheme data from PFF and used that to run a multiple linear regression model to see how defensive schemes affect scoring specifically.

The regression model had an adjusted r-squared value of .18. This means that the defensive scheme only explains about 18% of the variance in the points being scored.

2-High (Coefficient: -0.142 & P-Value: 0.366)- For every 1% increase in the number of 2-high defenses being run there is a .142 decrease in the number of points being scored. However, with a p-value significantly above .05, this isn’t statistically significant.

1-High (Coefficient: -0.07 & P-Value: 0.66)- Again, for every 1% increase in the number of 1-high looks there is a 0.07 decrease in the number of points, but the p-value suggests this isn’t significant.

Blitzing had the greatest impact on the offenses scoring, with a coefficient of -0.22 and a p-value of .13 suggesting that there is a stronger (still not significant) correlation to blitzing and scoring than other defensive schemes.

Total Yards

We saw similar statistical figures here as we did with total points. There was an adjusted R-Square value of .16, signifying that defensive schemes only explain 16% of the variance in a team’s yards. We also saw slightly negative coefficients across the board and large p-values, which again suggests while there is a decline in yards against these schemes it is not strongly correlated to the defensive play call itself.

Explosive Pass Plays

Another one of Mel Kiper’s frustrations is we don’t see the deep ball anymore. Well unfortunately for Mel, the numbers don’t support that either. From 1960-1989 (because these were the years Mel described in his rant) the average net air yards were 5.83 per attempt. Over the past 10 years it has been 6.24 air yards/attempt. I was able to get explosive pass play data over the past years. I will be using a stat called deep completion ratio (I may have made this stat up.) It is the number of completions over 20+ air yards/total number of completions.I think this will suffice for this project. The average DCR (deep completion ratio) over the past 10 years is about 6.6%. The more interesting thing is that there has been no change in this number as 2-high looks have increased or as points have decreased. Meaning that offenses are finding ways to hit deep shots no matter what.

Turnovers

This part was more for me. I was curious to see if interceptions have increased with the rise of this immovable object Mel Kiper calls “Cover 2.” Turns out, they haven’t. Interceptions have actually decreased…by a lot. Using the same time frames as before (1960-1989), the average interceptions per game was 1.4. Today, they are 0.8, that’s a 57% decrease. When looking at the linear regression model we see a higher adjusted R-Squared value at .375 (or 37.5% of the variance), but we see similar coefficients and p-values across the board. Looks like the defensive schemes don’t tell the full story here either.

So Why Has Scoring Decreased?

As our regression model showed, there are most likely a lot of reasons why scoring is down. One reason could be due to the increase in starting QB injuries. QBs missed 102 games in 2023 and 88 games in 2022. In fact, in 2023 16 of the 32 starting QBs went down with an injury.

Another reason could be the increase in RPO’s. The passing concepts off RPOs are typically screens and slants, which are limited air yard pass attempts. Naturally, this leads to less explosive plays which leads to more failed drives. Check out this PFF article to learn more about that.

Maybe it’s because of the increase in the use of 5 defensive backs and more athletic linebackers. The defense has more athletes on the field than ever before. According to MatchQuarters.com, “in 2019 teams used 5 defensive backs 55% of the time; now this number sits at 67%.”

Latly, it could just be that we are reverting closer to the mean. The average points per game since 1940 is 20.9, you could argue that the years when we were at 23 or 24 points were just outliers.

Whatever the reason is, there is no data that suggests that the 2-high safety defense is causing this short-term decline in scoring.

Nonetheless, football is in an amazing place. Viewership, revenues, and talent is at an all-time high. If a specific defensive coverage is our biggest worry, then the sport is doing just fine.