It’s not a good time to be a New England Patriots fan, but there’s glimmers of hope. One of them is Drake Maye, our new rookie quarterback, and the other one is Christian Gonzalez: Bill Belichick’s final draft pick with the franchise. Broadly speaking, Gonzalez has been tasked with guarding the best receiver every single game, and while he’s struggled here and there, he’s shown a lot of promise.
Seeing Gonzalez succeed gives me and other New England fans flashes of the past, when strong cornerback play shaped the team’s defensive identity, often in very different ways. It also made me curious as to the nature of cornerback play itself. What makes for a good cornerback? Is it forcing turnovers? Creating pass breakups? Eliminating targets? Can we quantify great cornerback play?
In this piece, I’ve examined 18 seasons worth of cumulative premium statistics on prominent cornerbacks from Pro Football Focus (2006-2023). My goal was to generally explore, analyze, and visualize the statistics between different styles of cornerback play – specifically focusing on “shutdown corners” and “ball hawks” – in modern NFL history.
Methodology
The first task I had was to gather up all the spreadsheet files of PFF’s coverage statistics. I decided to keep these constrained to only completed seasons for the sake of comparing historical data. One element I learned from this process is that cornerback play itself is quite volatile. It’s not uncommon for cornerback statistics to wildly skew from season to season; however, by taking a decade plus approach, we can hopefully discover some stability.
After collecting all the relevant CSVs, I merged them into one spreadsheet accounting for all 18 seasons. I then created a separate sheet accumulating a specific set of statistics for every single player rather than all of them, ranging from position to coverage snaps, receptions, targets, PFF coverage grades, interceptions, yards, and yards after the catch.
NOTE: When it came to career statistics, I took a strictly sum-based and cumulative approach. In other words, for ‘averages’ stats per season, I did not factor them for career stats; I only averaged them after accumulating totals.
The last data-filtering task was only accounting for cornerbacks with over 3,000 snaps. For the purpose of this project, I wanted to prioritize the range of players whom we could broadly describe as long-term and reliable starting cornerbacks over the last 18 years.
Shutdown Corners
The number one job of a cornerback is to disrupt the passing game. To oversimplify, this is done through two types of coverage: man coverage, in which the cornerback follows their assigned receiver and sticks to them like glue, and zone coverage, in which the cornerback patrols a specific area of the field. In either way, a shutdown corner sticks to their assigned area, or receiver, and makes it difficult for the quarterback to throw the ball in their direction.
Pure shutdown corners are very tricky to assess. PFFs’ process for grading cornerbacks primarily accounts for plays made around the point of a catch, such as if the receiver catches the ball (bad), or if the defensive player intercepts it (good). However, the most successful outcome of a play that a shutdown corner is typically that the quarterback throws the ball somewhere else. Ideally, this happens when the cornerback is assigned the opposing team’s toughest receiver.
This leads me to the primary metric I’d like to bring up for this category of play: yards allowed per 35 coverage snaps. Although it’s a little different than PFF’s more direct “coverage snaps per reception,” I believe yards allowed per coverage snaps is a nice shorthand for encapsulating a cornerback’s ability to limit yardage from their respective opponent in an easily understandable format. There are times when a cornerback may not be targeted much, yet may give up a surprisingly high number of catches when targeted (like slot corners or number two corners).
Unsurprisingly, Nnamdi Asomugha – someone who once allowed 13 catches in an entire season – is at the top of the list. Not far behind him is the predominant zone corner of the 2010s in Richard Sherman. After that, you have your selection of Pro Bowl contenders and, in the case of Champ Bailey, Darrelle Revis, and likely Patrick Peterson, future Hall of Fame players. The only name I was genuinely surprised to see in this list was Tre’Davious White; at the same time, his struggles since his 2021 ACL tear have sadly overshadowed the elite caliber of play he previously showcased.
In terms of single-season marks, there’s nothing too surprising here, save for the appearance of Brian Poole, whose time as a key slot corner for the Jets in 2019 is represented here. However, it’s interesting to note a bit of a resurgence of strong lockdown corners within the 2020s. This is largely due to league-wide shifts away from the Cover 3 zone defenses that ruled the previous decade, and a larger use of split coverage across the league.
Ball Hawks
If playing lockdown defense and preventing plays from happening at all in the first place is the dirty work, breaking up passes after they happen and taking picks to the house is the money-maker. This is what most people think about when they imagine the best cornerbacks to ever play the game, and it’s not a surprise that many of them fit this mold to a tee. Unlike shadowing receivers, it’s far easier to isolate a cornerback’s success on a play based on what he does around the point of a possible reception.
When it comes to assessing this part of CB play. I took an approach not too dissimilar from what PFF already does with sacks, hurries, and QB hits – I combined pass breakups and interceptions into one category I called “Ball Hawk” plays. Though it’s technically more accurate to a cornerback’s value to weigh these two statistics differently, for now, it’s beyond the scope of what I’m exploring.
These are players who nonetheless have made reputations for themselves as some of the best play makers in league history. In the case of Asante Samuel, his time with the Patriots and Eagles coincided with him often being near the top of the league in interceptions. Conversely, in the case of Malcolm Butler’s time as a starting cornerback, he was always a reliable pick for someone who would break up a pass. This whole group is a mix of physical zone corners, ball hawks, and press coverage fiends of just about the last two decades.
Here’s where I was most surprised: at the presence of Asomugha atop the list for highest single-season ball-hawk-play-rate in recorded history. For as much as Asomugha’s reputation among football fans has been in the realm of practically erasing one corner of the field for opposing passers, it’s easy to forget that he had to earn that reputation for himself. In his legendary 2006 showing with the Oakland Raiders, he broke up 13 passes and had eight interceptions in only 430 coverage snaps.
For reference, Champ Bailey that same season – a First Team All-Pro player – had five pass breakups and ten interceptions in 598 snaps. Remember: this is only one particular statistic; it doesn’t necessarily mean that Asomugha and Bailey were of the same exact caliber or that Asomugha was better that year. However, it is worth noting as a point of emphasis for Asomugha’s legacy – that he actually was quite a physical corner capable of making plays when quarterbacks were not deathly terrified of throwing the ball in his direction.
Grouping CBs via Clustering
In the final portion of this piece, I wanted to test out an unsupervised machine learning method (K-means clustering) in order to automatically cluster or classify the cornerbacks captured through the entirety of my data set. To begin my process, I did the Elbow Method of trying to see how many clusters I should create for my analysis. As you examine the below graph, keep in mind that ‘inertia’ broadly describes the distance squared between each data point in my two primary categories (yards allowed per 35 coverage snaps and ball hawk plays per 35 coverage snaps).
Based on this chart, I ended up choosing four clusters. I could have done more, but I decided not to due to wanting to keep most of my analysis general and easily understandable. Had I gone with more than four clusters, the inertia between data points would have decreased, but we’d run the risk of over-fitting our data. Anywhere our unsupervised method of analysis falls short, I can point out.
Before anyone gets mad at me – I do not think this is entirely a one-to-one ranking of the best coverage cornerbacks. However, it does come fairly close in terms of the ranges created. The Asomugha to Revis to Peterson triangle essentially encompasses a group of players from franchise Hall of Fame contenders (Jimmy Smith, who might have had Pro Bowl potential were it not due to missing several games throughout his career) to NFL Hall of Fame locks (Richard Sherman).
Meanwhile, the second cluster broadly features strong role players (Desmond King II) to Pro Bowl or Defensive Player of the Year contenders (Stephon Gilmore), and the third cluster features players who range from reliable journeymen (Brandon Carr) to specialist role players (Dominique Rodgers-Cromartie). Lastly, the fourth cluster is almost entirely journeymen (Kyle Arrington) to reliable starters with flashes of brilliance (DeAngelo Hall). There’s people on the outskirts of either group you could move up and down into a different tier – I obviously do not think Jimmy Smith is a better all-time corner than Champ Bailey – but generally speaking, I think this was about right.
NOTE: I would take some of the spots for older players, like Charles Woodson, Champ Bailey, and Ronde Barber with a grain of salt. These statistics don’t account for seasons before 2006.
PFF Grade Correlation
Along with their immense database for coverage statistics, PFF also grades cornerbacks. Out of curiosity, I decided to measure the correlation between my two statistics – yards allowed per 35 coverage snaps and ball hawk plays per 35 coverage snaps – and PFF’s coverage grades.
The negative correlation (-0.2379) between YA/35s and PFF coverage grades initially seems to illustrate the obvious – a player’s PFF coverage grade tends to increase the fewer yards they allow. However, the relationship is extremely weak. This is likely due to the fact that number one cornerbacks generally take tougher assignments than number two cornerbacks. Guarding Tyreek Hill is not quite the same thing as guarding an average number two wide receiver.
In similar fashion, the positive correlation (0.1742) between ball hawk plays and PFF coverage grades indicates that cornerbacks with many interceptions and pass breakups tend to have higher PFF grades. At the same time, this is an even weaker relationship than the previous yards allowed. Though interceptions and pass breakups remain important, there is a clear element of variance to them, and not every interception or pass breakup is created equally.
Next Steps
Although I’d like to consider myself far more knowledgeable about the specifics of cornerback play after this research, the truth is far more humbling. As my correlation analysis shows, assessing a cornerback in coverage goes far more than assessing the yards a cornerback gives up or the number of ball hawk plays.
One particularly big statistic I didn’t incorporate due to time constraints was penalties and penalty yardage allowed. Furthermore, while a cornerback may or may not give up many yards, it’s not all encompassing. A player could theoretically allow many receptions and first downs, yet not be targeted very much by raw numbers. In a future analysis of cornerbacks, it may be worth assessing first downs allowed, touchdowns allowed, and factoring penalty yardage into any analysis.
The last part of any cornerback analysis that I believe may make a fun post another time: impact on team defense. Although it’s difficult to isolate cornerback statistics from the statistics of safeties and linebackers – let alone the broader pass rush of any team – I would be curious to assess the correlation of strong cornerback play in my two metrics to team defense statistics, such as pass defense EPA or pass defense DVOA.
For now though, I hope those of you interested in professional football and cornerbacks enjoyed this post. Maybe you learned a bit more about what makes for great cornerback play in the NFL.