The Dallas Cowboys’ defense has surrendered 35 or more points three times in five games. They rank 29th in points allowed. The defense is on pace to set a record-low 10 turnovers over a 16-game season.
In short, the Cowboys’ defense stinks right now. It’s a major problem and if the team has any real chance of making 2017 season a success changes need to be made. But what changes? If you follow social media you would find an overwhelming consensus on two particular players who have not satisfied the eye test of fans:
Jeff Heath is 60th in tackling efficiency against the run among 62 safeties who have played more than 50 % snaps #Cowboys— John Owning (@JohnOwning) October 2, 2017
Dear Jeff Heath, please do this before each game and have Jerry bench you. Thank you pic.twitter.com/FIz29bcqMp— Cowboys Life (@DezCowboys88) October 9, 2017
A strong consensus has emerged among fans that neither Jeff Heath nor Jaylon Smith are playing well (that’s putting it kindly). Both are feel-good stories that had fans rooting for them going into the season. One an undrafted free agent who’s slowly worked his way into a starting position. The other a highly-pedigreed elite athlete who suffered a devastating injury in his final college game that had many wondering whether he’d ever play again.
Fans love to root for such stories. But great stories end when performance is deficient and the team is losing as a result. But is our perception really true? Does the eye-test match what the numbers and advanced metrics tell us? Let’s find out.
Defensive players are hard to measure statistically, but we can use “splash plays” to get at least some gauge on how much impact defensive players have on games. Splash plays include sacks, interceptions, fumbles caused, fumbles recovered, tackles for loss and passes defensed. They’re called “splash plays” because each makes a significant impact on that particular play.
Not surprisingly we find Demarcus Lawrence far ahead of every other Cowboy in generating splash plays. Also not surprisingly, rookie Jourdan Lewis ranks second with eight splash plays. After that we quickly degrade into a fairly large group with four or five splash plays and then a third group with fewer. Five games into a season if you’re a defensive player with three or fewer splash plays you’re not making plays. One splash play per game would make you a solid player; less than that you’re a part-time contributor. So everyone who has three or fewer can be considered a contributing piece. Just to note, Jeff Heath and Jaylon Smith have combined for exactly three splash plays in five games. So, from that standpoint the social media judgement rings true. Those are remarkably low numbers for players who’ve been on the field as much as Heath and Smith.
All splash plays are not equal, however. The following is a “splash points” grid. I learned that official NFL stats count sacks as both a sack and a tackle for loss and an interception as both an interception and a pass defensed. This grid reflects that new-found knowledge.
You’ll see interceptions and fumbles recovered score the most points, which makes sense since those take the ball from the opponent and give it to the Cowboys. Dallas has generated only three turnovers so the splash points are not much different from splash plays:
A couple observations:
- Yes, Demarcus Lawrence has been a beast this season.
- Jourdan Lewis is the best Cowboys’ pass defender right now and it’s not really close. He’s gained notice from others as well:
Tre'Davious White has led an impressive class of rookie CBs pic.twitter.com/uRjkUZUp04— Pro Football Focus (@PFF) October 7, 2017
- David Irving, after 60 minutes of football, ranks sixth in splash points on the season. I’m not sure if that’s good because it shows he had an immediate impact or whether it’s bad because so few of his defensive teammates have been able to top his numbers in five games.
- I’ve read many claim the struggles of the defense are predictable because the Cowboys front office has failed to invest “resources” on that side of the ball. That’s not exactly true. Consider, five of the teams last six first- or second-round draft picks have been on defense. Those five players are contributing very little:
The problem hasn’t been a lack of investment on defense. The problem has been investments in players who simply haven’t been as good as expected thus far. Jones has been acceptable but not the dynamic player many hoped for. The Gregory pick is a complete bust. Smith is nearly a year and a half into his rookie contract and has made one play. Charlton is a first-round pick who’s having problems getting on the field and Awuzie seems to have a perpetually pulled hamstring.
Your defense will not succeed if the high draft picks are not producing.
A look at how much each player is contributing to the team’s overall share of splash plays show the outsized contributions from a small number of players:
Seven players have accounted for 73% of all splash plays (and one of those players has played all of one game). That means the other 12 players who have seen time on defense have contributed only 27% of splash plays. That can’t continue; the rest of the defensive roster simply has to start making plays if the defense is going to show improvement.
One positive to take away from this is the youth on display. The top impact players by age and experience:
Other than Crawford, all are 25 or under and three are rookies or in their second season.
I said above that all splash plays are not equal. Dallas has done a good job generating sacks (sixth in the league; on pace for 51 on the season) and tackles for loss but the defense simply can’t generate turnovers. Two interceptions and one fumble recovery through five games simply won’t cut it.
Some like to claim that turnovers, especially fumble recoveries, are basically random. And there’s some truth to that. But this is the third consecutive year with Dallas generating very few turnovers and failing to recover 50% of fumbles caused. At some point it’s not “randomness” and is instead a feature of the program.
Just for fun, here are the splash plays by game:
Eighteen splash plays per game would make a defense elite; twelve per game makes you average. Dallas has been average for the most part and had an elite game against Arizona. The problem has been the type of splash play, as noted above. Tackles for loss and passes defensed are good, but they’re not the same as turnovers and until the defense can start taking the ball away from opponents they’re going to struggle.
Looking at splash plays by position reveals a couple of really interesting insights:
- The linebacker position has generated only seven splash plays all season
- The safety position has generated only six splash plays all season
That’s barely one splash play per game from two of the five position groups on defense. Those are atrocious numbers for positions that should be making double that number on an average defense.
So, thus far we’ve basically captured production numbers from each player. And the numbers support what the eye test tells us which is that we have problems at safety and linebacker and Heath and Smith are the prime culprits.
But every player doesn’t play every play; some have more opportunities to make plays while other have less. Let’s check the snap counts:
Wow. Anthony Brown and Byron Jones having the most snaps on defense doesn’t surprise me. Jeff Heath being third does surprise me. I knew he was playing a lot; I didn’t realize he’s basically playing every down. That means he has lots of opportunities to make plays. And right behind him is Jaylon Smith. The two rank third and fourth in defensive snaps played. Compare those numbers with Lewis (more than 60 fewer snaps than each), Woods (165 fewer) and Irving (240 fewer).
A good comparison is to look at snap counts (line) versus splash plays (bars):
And here we get to the nut of the problem. The Dallas Cowboys have two players who are getting the vast majority of defensive snaps while not making many plays. The numbers completely support the fan judgement that Heath and Smith are weaknesses on defense.
Using the above numbers we can come up with a “splash play rate” metric which shows the percentage of plays on which a player generates a splash play:
David Irving, with two sacks in his only game, ranks highest. Demarcus Lawrence is right behind him. Wreaking havoc 8% of the time or more is an elite number. But we see the numbers quickly drop down below 2% which is a below average number. Most disturbing is the fact that Heath and Smith are both around 0.5%...which means they both require 200 snaps or so to make a play. This means they’re basically making one play every three games and that simply isn’t good enough for a starting player.
Notice the other names below 1%, they’re all backup or role players. Only Heath and Smith are starting while also generating sub 1% numbers.
Pro Football Focus grades
I noted earlier that measuring defensive players based upon statistics can be difficult. Pro Football Focus reviews every play by every NFL player and assigns each a grade. It’s not a perfect approach, and you can quibble with this grade or that. It does, however, provide us another data point to measure defensive players. Let’s check out their findings:
Hmmm...these numbers largely match up with the splash play metrics. Lawrence is the best defensive player on the Cowboys. Lewis ranks second, Tyrone Crawford has also been a solid contributor.
Now, the PFF numbers are not counting numbers, so a player with very few snaps could rank high in PFF grades while not playing very often (and thus not accumulating big splash numbers). Thus we see both Xavier Woods and Chidobe Awuzie with relatively high grades despite playing sparingly. Still, we see the same two regular players jump out with poor grades: both Jeff Heath and Jaylon Smith receive “poor” grades from PFF.
Adding snap counts well illustrates the issue facing the Cowboys:
I haven’t mentioned Maliek Collins up to this point. But the above chart really illustrates three problem players. Heath, Collins and Smith grade poorly from PFF yet are getting a huge majority of defensive snaps. Collins was considered an upcoming star going into the season and got off to a hot start against New York. But I can’t recall his name being called much...and he’s no doubt been a part of the defense surrendering 160 yards rushing three of the last four weeks.
As for Smith and Heath...these grades simply confirm what our eyes and the splash statistics told us already: they shouldn’t be on the field at the rate they have been. Neither is a capable starting player on an NFL defense.
I understand why the team has put Smith out there; they don’t have any other options. First Anthony Hitches was injured, then Sean Lee got hurt. Smith was supposed to be eased into a significant role as he rebounded from his injuries. Instead, he’s been asked to start and play practically every play from day one. The task has proven too much; he’s not ready. It’s not an indictment on him but right now there’s no question he’s a vulnerability opponent’s are exploiting.
As for Jeff Heath...I have no idea why he’s playing so much. The Cowboys have not one but two alternatives. In short opportunities Woods has played very well:
These rookie cornerbacks have been tough to find success against pic.twitter.com/hK5AWH1wUq— Pro Football Focus (@PFF) October 10, 2017
Woods looks exactly like the hard-hitting, playmaking safety he was in college. He’s under-sized and his tackling technique leaves something to be desired but how can he possibly be worse than the abject failure the Jeff Heath experiment has been? I absolutely do not understand how fans, stats and advanced metrics see an obvious truth the Cowboys’ coaching fails to see.
There are rumblings, though, that if a replacement is to be made it will be Awuzie taking Heath’s spot. I’m fine with that too because I feel anyone else offers promise that doesn’t exist with Heath. I feel Woods is the better centerfield-type safety but Awuzie is the more pedigreed player so I get it.
Thus, we get to the whole point of the article. Using PFF grades we can put together the lineup that has generally been run out there every week thus far, color coded by elite / above average / average/ poor:
Players in red are poor and we can see the areas of concern. Four starting positions have ranked poor thus far. Luckily, there’s solutions for most of them. Heath should be replaced by Woods (or Awuzie); that’s the no-duh, obvious thing to do. Smith should also take the fourth linebacker spot with Lee, Hitchens and Wilson taking the starting roles.
Irving has already proved disruptive and has assumed the starting tackle position.
While Collins has been graded poorly he has played well in the past and thus gives hope he can improve moving forward. A potential new lineup
PFF doesn’t actually have a grade on Wilson (which I don’t understand) but I rated him “poor” because the linebackers overall have been terrible. Suddenly, instead of having four of 11 positions manned by “poor” performers the defense has only one (Collins) and he at least has the pedigree to offer promise for improvement.
At this point, the best move for the Cowboys may be to go all in on the young talent. Make the starting nickle (the real base defense in the pass-centric NFL) Awuzie, Lewis, Woods, Byron Jones, and Anthony Brown. Let them learn together as a unit and hope they can continue what has been a good start. Based on what we have seen from the rookies, that may become one of the better secondaries in the league. Scandrick, Jeff Heath, and Kavon Frazier should be relegated to backup and/or special teams duties, which frankly is what their play has earned them so far this season.
I’m pretty certain the starting linebackers against San Francisco will not include Jaylon Smith. However, I’m very curious to see if Woods or Awuzie is starting in place of Jeff Heath, and who’ll be making up the team’s nickle and dime units. This seems like a simple decision to me, but thus far the coaching staff sees it differently. I imagine the braying from fans and media will intensify if we don’t see changes.