When studying pass rushers in the NFL, the first statistic that often comes up is sacks. After turnovers, sacks have the largest impact on a drive for a defense. In 2018, offenses lost an average of 6.7 yards per sack, as well as losing a down on that series.
The problem with using sacks to evaluate pass rushers is that they are dependent on much more than just the play in the trenches while lacking year-to-year consistency. As far back as 2009, evidence suggests that sacks may be as dependent on the quarterback as either the offensive or defensive lines. In 2013, Football Outsiders showed that sacks have a better year-to-year correlation than any other quarterback stat except completion percentage. Jason Lisk did a similar study and found the same results in 2018.
What statistics, then, should we use in our evaluations of pass rushers?
Pressures and pressure rate may paint a better picture. According to Pro Football Focus, pressures are more than twice as stable year-to-year as sacks among players who rush the passer on at least 150 snaps per season. My biggest issue is that most football fans (myself included) have no baseline for what should be considered a *good* pressure rate.
If I told you that Player X finished the 2018 NFL season with a 16.6% pressure rate, would you know how well he performed? You’d likely have to ask what the average pressure rate was last year, or where he ranked leaguewide. In contrast, if I told you that Player X had 11 sacks last year, you’d probably have a general idea of how good he is.
That is what brings us to our new metric: Expected Sacks (xSacks). The goal of xSacks is to bridge the gap between pressures and sacks for pass rusher evaluation. The basic idea is to take game-by-game pressure rates for pass rushers, factor in the opposing quarterback’s sack rate, and spit out a number that should be in the same range as sacks.
Using data collected from Pro Football Focus, I calculated the sack rate for every quarterback for the 2018 season specifically on dropbacks where they were pressured. Among players to start at least 5 games, Marcus Mariota rated the worst in this category, getting sacked on 29.8% of plays where he felt pressure. Andrew Luck was far and away the best at avoiding sacks, taking one on only 9% of pressured dropbacks (Editor’s Note: I refuse to acknowledge this).
I then took week-by-week pressure numbers for all players listed as either an EDGE or interior defensive lineman and multiplied with the opposing quarterback’s season-long sack rate on pressured dropbacks to get the number of sacks we’d expect to see from each player, each week.
Here are the top 10 pass rushers by xSacks:
Remember that Player X from above? That’s the Philadelphia Eagles’ Fletcher Cox. Finishing the year with double-digit sacks is a great accomplishment, but it wasn’t enough to finish in the top 15 for the year. And yet, Pro Football Focus gave him the second best pass rushing grade of 2018, just a shade below Aaron Donald and his 20+ sacks. According to this xSacks model, Cox should have finished with 19.5 sacks, second to Donald. It was a dominant season, but if you don’t have a chance to watch him on tape, and you only see his sack numbers, it can be difficult to see why PFF saw him as the second best pass rusher in the NFL last year.
When we compare the xSack totals to PFF’s season-long pass rushing grades, we can see that there is a correlation between a player’s grade and his xSacks. (r² = 0.445)
The initial goal of this metric was to quantify pressures in a way easier to digest at a glance for any football fan than something like pressure rate. Ideally, we should see a distribution of xSacks that looks similar to that of actual sacks.
Player xSacks tend to be just over 1 sack higher than their actual sacks. Intuitively, this makes sense because any player that had at least one pressure will have an xSack total greater than 0. This is not true with sacks, as many players had multiple pressures without ever getting to the quarterback. Both xSacks and sacks follow a very similar distribution and max out right around 20 sacks, indicating that we can use our preconceived baseline for what a *good* number of sacks is when looking at xSacks.
That’s all well and good for fans debating which player is better, but it needs to do more. How can teams utilize this information to gain a competitive advantage? This statistic doesn’t add much value when game planning, but it could absolutely help in roster building. If we go by Sean Clement’s idea for a General Staff structure of an NFL organization, this data would be best utilized by the Future Operations division. Below, sacks and xSacks are plotted against one another. The top 10 players in both sacks and xSacks have been highlighted.
Any player above the red line finished the season with fewer sacks than expected. Conversely, players below the line finished with more than expected. Let’s take a closer look at two players that signed new contracts this offseason: Jerry Hughes and Dee Ford.
|Player||Age||APY||Total Contract||Contract Length|
|Dee Ford||28||$17 million||$85 million||5 years|
|Jerry Hughes||30||$10.75 million||$21.5 million||2 years|
Ford had a fantastic season last year and was rewarded with a big 5-year deal. His 15 sacks almost perfectly matched his 15.1 xSacks. Hughes had a perfectly respectable year, finishing with 7 sacks. But his 13.5 xSacks was just 1.6 behind that of Ford. Age is, of course, a factor here, but if Hughes can keep up with Ford at this pace, he’ll be providing 90% of Ford’s productivity at 60% of the cost. With teams committing so much cap space to quarterbacks these days, they are forced to find cheaper options at other positions, and this is a great place to start.
Here are the top 10 players with the lowest sacks over expected (that is, they performed better than their sack total suggests):
This is a great place to start if you are a team looking to find some cheaper options for pass rushers.
Likewise, we can check out the players whose sack totals may have overexaggerated their abilities by looking at the top 10 highest sacks over expected:
Kyler Fackrell is entering the final year of his contract in Green Bay, coming off a career year where he put up 11 sacks, after getting just 5 in his first two seasons combined. The Packers now have to decide if Fackrell’s 2018 was a breakout year or an anomaly. Should they pay him like a double-digit sack artist? Based on his xSacks, the answer is a clear no. He outperformed his expected production by far more than anyone else last season.
When we look at xSacks, it’s important to note that this statistic, like sacks, is cumulative. Players with more snaps are generally going to accumulate more xSacks. If we want to better evaluate players who missed time due to injury, or younger players that aren’t getting many snaps quite yet, we can turn this into a rate statistic by calculating the number of pass rush snaps per xSack.
These are the top 10 players who had at least 150 pass rush snaps in 2018. Hughes and Cox look great here again, as well as the usual suspects like Aaron Donald, Khalil Mack, and Justin Houston. But we also now see a couple rookies in Uchenna Nwosu and Kemoko Turay stand out. By this metric, Nwosu and Turay performed as well as any of the elite pass rushers in the league on their limited opportunities. Based on this metric, coaches should make an effort to get them on the field more often throughout their sophomore campaigns.
Issues and Improvements
The sack rate for opposing quarterbacks when pressured that is used in this formula comes from every passer’s 2018 season totals. This gave a decent sample size for each, while also focusing on their current playing style. In the future, we will have to determine which sack rate best represents each quarterback, each week. Will their full career totals be the most representative of their propensity to take a sack when under pressure? While that may be more characteristic of the quarterback, it won’t accurately capture their current play. After all, a 40-year old Russell Wilson might not be the escape artist he was at 25. I believe the next step would be a rolling average of the previous (x) games. More work is needed to figure out just what (x) should be.
There was also the issue of what to do when multiple quarterbacks played for one team in the same game. For example, there were multiple games in which Jameis Winston and Ryan Fitzpatrick both played for the Buccaneers. Winston had a pressured sack rate of 18%, while Fitzpatrick held his below 14%. Since we can’t know which quarterback was on the field for any given pressure, I took a weighted average of the two quarterback’s pressured sack rates based on their total dropbacks in the current game.
As with any statistic, the numbers can’t show everything. Both Aaron Donald and J.J. Watt finished the season with more sacks than expected, but they are definitely not overrated pass rushers. Jerry Hughes had a fantastic year, but is his age going to catch up with him? These numbers are best utilized as a starting point for player evaluation. Find the players who look underrated here, then review their tape to find what it is that makes them stand out. As always, the numbers and tape study should work in tandem to tell a player’s story.
Listen to our newest episode of Tasteful Profanity right here!