Last Season, The Colts Had a Top 10 Defense . . . But Not Really

Thanks to the nflFastR project, PFF, and NFL NextGen Stats for being awesome sources of data.

---

The job of a defense is to prevent the other team from scoring and in the 2021 season, the Colts did that pretty well, giving up only 21.5 points per game, which was the 9th lowest of any team. This achievement lined up pretty well with many analyst top 10 rankings. Pro Football Focus graded the Colts D as the 9th best, while Football Outsiders was even more bullish with an 8th ranked defensive DVOA.

I, however, beg to differ. Underneath the “Top 10” veneer of the Colts defense was a team that just wasn’t very good and I doubt they can repeat their 2021 success without a significant upgrade in overall play.

Let me frame my argument with some numbers to explain what I am talking about. This chart ranks last season’s defenses by points per drive against (1) and also provides some additional opponent drive level information.

This shows that the Colts gave up the 11th lowest PPD to opponents, which is pretty good. However, it’s the other supporting numbers that concern me.

---

FIELD POSITION

Each green bar in the chart represents the average opponent drive data and the left edges of those bars are their average starting field position. Field position is fluid and I would be lying if I said as the game wears on that a defense has zero impact on opponent starting field position.

However, the results of special teams and the offense are by far the biggest contributors of where your opponent starts their drives and last year the Colts’ opponents had the 5th worst starting field position, beginning their drives a full 2 yards further back than the average offense.

Now, that 2 yard advantage may not sound like a lot, but the impact is actually pretty significant. Here are the average points scored from starting field positions over the last 10 years.

While there is some variance of results in there, the trend is clear: every yard of starting field position is worth about 0.04 offensive points on average. This means the Colts opponents had about a 0.08 point per drive disadvantage before they even started their drives.

Again, that may not seem large, but if I adjust the defensive PPD by starting field position (ADJ), the Colts defense falls 6 spots to 17th and suddenly this top 10 defense looks just average.

---

EFFICIENCY

As some of you may know, Drive Success Rate (DSR) is one of my favorite stats. It represents the probability of an offense converting a series to a new set of downs, so it does an excellent job of gauging how well an offense moves the ball (or how well a defense prevents it). In 2021, the Colts defense had a DSR of 74.8%, which means they let opponents convert nearly 3/4 of their series. That ranks 22nd in ability to prevent offenses from moving the chains. Simply put, the defense was significantly below average at preventing first downs.

Their inability to limit drives was also reflected in their yardage efficiency. On average, the Colts gave up 33.2 yards per drive, which ranks 17th and on a per-play basis it is even worse with a 23rd ranked yards per play against.

---

TAKEAWAYS

So, if they were so bad at stopping drives, then how did they limit points? Well, as already discussed, starting field position was a variable, but the much larger factor was turnovers.

Last year the Colts defense had the 2nd most takeaways with 30 (special teams excluded). Those takeaways break out to the 3rd most INTs (19) and the 2nd most fumble recoveries (11), so they were proficient in both offensive phases.

Takeaways are great. They keep your opponent from scoring and usually set your offense up in very good field position, so they are often the most impactful plays in a game. The problem, however, is that they aren’t really a repeatable skill.

I can already hear the groans. “Sure, Darius Leonard’s punch-outs aren’t a skill. Whatever stat-boy.” Obviously, some players are better than others at ball hawking, so what I mean when I say “not repeatable” is that there are many other variables involved beyond player skill that prevents takeaways from consistently happening.

This can easily be seen using year over year correlations. Here is an explanation from Football Outsiders:

The idea behind year-to-year correlations is that if a particular measure closely represents actual ability, then that stat will hold relatively consistent from one year to the next. However, if there is a lot of variation from one year to the next, that suggests that performance measured by the statistic is largely due to luck or other factors. Thus, the lower the correlation, the less predictable it is, and the more it can be attributed to luck.

— Jim Armstrong

Over the last 10 years, the year over year correlation of defensive statistics like points per drive or yards per drive has been fairly strong (for football), but for turnovers that relationship has been very weak.

In other words, there is almost no relationship between the number of takeaways a team gets one year to the next. While that may seem hard to believe, here are some common sense explanations why this is.

1. Interceptions are more a function of the QB that throws them than the defense that catches them. Some QBs are just more careful with the ball and/or throw less often. Defenses face a varying suite of QBs year to year leading to varying interception opportunities.
2. Interceptions tend to be situation dependent. Teams that trail, throw more than 2.5 times the number of picks as teams that lead. Consistently getting an opponent in a hole will likely increase INT volume. A defense can limit opponent scores but game score differential also depends upon their offense, which the defense has no control over.
3. Forcing fumbles is a skill but fumble recoveries are luck. Historically, the defense recovers around 50% of forced fumbles and the year over year correlation of recoveries is basically 0 (2).

If you don’t buy these explanations, then fine, but the historical results are unequivocal. If you believe the Colts are going to be top 2 in takeaways next year, prepare to be disappointed.

---

THE OTHER 87%

A good defense will have the ability to stop drives when turnovers don’t happen and in 2021, the Colts were not that defense. If I remove all turnover drives from last year’s games, then the chart I originally presented looks like this.

Without drives ending in turnovers, most defenses still retained their PPD ranking. Only 2 teams from the original top 10 dropped out of that list (MIA 7th to 15th, TEN 9th to 12th).

But the Colts have the most drastic drop of any team, falling from 11th place to 20th. This illustrates how reliant they were on takeaways to stop drives. Here are their other non-turnover drive rankings:

• 22nd Adj PPD
• 21st DSR
• 20th Yds/drive
• 26th Yds/play

Seeing as 87% of opponent drives don’t result in turnovers, they might want to fix that.

---

CONCLUSION

I want to stress again that I am not saying that ball hawking is not a skill, but historically, there just has not been much persistence in takeaways regardless of skill level. They are uncommon events with high variance that just can’t reliably be schemed around.

It is highly unlikely that the Colts will have the same takeaway success next season as they did in 2021. As such, if they are to maintain their “Top 10” defensive status, they will need to be a lot better at stopping opponent drives.

---

Footnotes

1) PPD are from regular season games that exclude drives that end due to time expiring. Point totals are reduced for safeties and opponent scores off of turnovers.

2) The defensive recovery rate from forced fumbles between 2016 - 2020 was about 47%. This is very similar to a 2000 - 2011 study that found a 50.1% overall recovery rate. From 2012 - 2021 the year to year correlation of forced fumble recovery rates was -0.01. Unforced fumbles have a much lower defensive recovery rate than forced fumbles but they also occur much less often.