Subscribe to Blog via Email

Enter your email address to subscribe to this blog and receive notifications of new posts by email.

Join 6 other subscribers

MN Vikings Tweets

Bleacher Report – Vikings

Arif’s deep stats dive: Vikings’ excellent defense propping up worrisome offense right now

By Arif Hasan

After every game, we’ve taken a look at the advanced statistics that helped drive the win, what can be counted on going forward and what are areas of concern to be fixed before the next matchup.

There’s a good explanation of all of the advanced stats in the Week One article. We cover a number of statistics not generally covered in the box score or don’t get the attention they deserve for giving us a better or different context for how games unfold—or how they may unfold in the future.

Team Level

We fundamentally discuss two statistics with regards to team-level performance: points per drive and drive success rate. Because the fundamental unit of football play is drives, not plays (drives must end in scores or turnovers, but plays can continue), it makes sense to evaluate how teams perform on a per-drive basis.

That’s especially true as teams can modify their pace or be subject to the pace of their opponents and that will have more bearing on who wins.

The Vikings offense was bailed out by a safety and a special-teams score once more, and they scored 12 points total, excluding after-point attempts. They might deserve a little bit of extra credit for scoring a two-point conversion, but points-per-drive does best when measuring holistic play and not set-up play from the goal line; it measures the offense’s ability to get in scoring range and finish.

This week, the Vikings scored 1.20 points per drive, compared to a league average of 1.63 this last week. For the week, the Vikings ranked 22nd of the 30 teams that have played so far. That’s a modest improvement over what they’ve done this season, in part because they had more drives in the other games and didn’t score on those drives.

For the year, the Vikings haven’t improved much relative to their opponents. They rank 27th of the 32 teams, only above the Titans, Browns, Rams, Texans and Bears.

Defensively, the story is nearly the opposite. Last week, they ranked fourth in points per drive given up, with Carolina grabbing 0.58 points per possession.

On Sunday, the Vikings played against the team that scored more points per drive than any other team in 2015. On the prior Sunday, the Vikings played against a team that went on to score more points per drive than any other team in Week 3 of the NFL season.

Despite that, the Vikings rank second in points per drive given up for the season, behind only the Eagles—who have played the worst team in PPD, the 29th team and the 17th-ranked team in PPD. Once adjusting for opponent (something we’ll be able to do next week), the Vikings are probably the best team in defensive points per drive.

The Vikings’ net points per drive in Week 3 was much better than it usually is, and at +0.62, was 11th in the league. That doesn’t quite match the Vikings’ likely status as a top five team by most estimates, but it does indicate that the Vikings probably have some unsustainably good luck bolstering their play—like unusually excellent special teams work on both sides of the ball, and the ability to play with a lead despite the offense not themselves grabbing the lead.

That matches the most concerning trend the Vikings have—their third-down rate. I typically like to use drive success rate to measure the ability of a team to create new downs (after all, converting a new set of downs on first down is good and not measured by third down rate), but sometimes it demonstrates a concept perfectly.

The Vikings offense is converting third downs on 31.7 percent of plays. That’s the same rate as San Francisco and Chicago, two of the worst offenses in the league.

Still, if the Vikings are getting in scoring position by converting more often than not on first and second down, it wouldn’t matter. That’s why we use drive success rate—which measures, generally, how often teams move the chains and score touchdowns—over third down rate.

Defensively, the Vikings have a fantastic drive success rate. Keeping the Panthers to 0.620 (Carolina only created a new set of downs or scored a touchdown on 62 percent of their opportunities) is extremely impressive; the 30th-ranked offense in DSR last year produced a success rate of 0.633. It is difficult to overstate how incredible that is against an offense that ranked third in the statistic last year.

For the season, the Vikings defense has limited opponents to a rate of .630, which currently ranks seventh. As mentioned above, an offense last year with the same rate would have been the third-worst in the league.

Against the Carolina Panthers, an admittedly stout defense, the Vikings once again had issues generating first downs. Their drive success rate was 0.590, which is awful and ranks 25th among the 30 offenses that performed outside of Monday Night on Week 3.

Through 2016, the Vikings offense has produced a drive success rate of 0.603. The only offenses with a worse drive success rate are Houston, San Francisco and Los Angeles.

A net drive success rate of -0.027 is predictably around average and ranks 21st.

Again, this doesn’t mean that the Vikings are the 21st-best team in the league. These statistics don’t account for the ability to avoid turnovers or generate them. But it likely means the Vikings will fall back to earth; Sam Bradford threw two dropped picks against the Panthers and is relatively fumble-prone for a quarterback.

In Jerick McKinnon’s college career, he fumbled on 4.3 percent of carries. For context, the average running back (at least 500 total carries since Adrian Peterson entered the league) has fumbled on 1.2 percent of carries. The most fumble-prone among those running backs (Darren Sproles) fumbled on 3.1 percent of carries. After that, backs like Reggie Bush, C.J. Spiller and Ricky Williams averaged 2.8, 2.6 and 2.5 percent of the time.

Fumble-prone Adrian Peterson? 20th-most of 74 running backs, at 1.6 percent—though in his worst years averaged 2.5 percent.

McKinnon is carrying the ball differently now than he did as a quarterback, but it’s a reasonable concern going forward that he is more likely to fumble than the average back, especially with his smaller hands and issues at times catching the ball.

The Vikings will probably maintain their positive turnover differential, but a plus-eight over three games (42.7 over 16 games) is itself unsustainable. Only one team since the merger has had a turnover differential above 30 (the 1983 Washington Redskins, at 43, interestingly enough). The defense might be able to produce turnovers at this rate but the offense will not be able to avoid them. Quarterback sack-fumbles, receiver fumbles, interceptions and running back fumbles are inevitable.

On average, they’ve won their games by eight points—one possession. It’s not difficult to imagine that a team winning 2.7 turnovers a game suddenly losing similar games if their turnover differential inches closer to +1 per game.

That’s not to mention that return touchdowns are also unsustainable. As FiveThirtyEight put it:

In terms of fluky scoring, so far this season, Minnesota has scored as many return touchdowns (three) as it has offensive touchdowns, and that trend is unlikely to continue. Even good defenses and special-teams units can’t produce return touchdowns every week. Last year, Seattle and Arizona were the only teams with at least three return touchdowns through three games, and they combined for just five more the rest of the season. Digging deeper into history, from 1990 to 2015, there were 22 teams with exactly three return touchdowns after three games, and those teams averaged only 3.3 more return touchdowns the rest of the year.

The Vikings are clearly better than their 21st overall ranking in net drive success rate and are performing like a top-five team. Even better, after accounting for the fact that two of their three opponents were division favorites.

A strength-of-opponent adjustment and some accounting for what will likely remain a positive turnover differential should boost the team’s overall showing but unless the Vikings can more consistently generate offense (for example, having the running game perform like they did in the first half against Carolina and Bradford perform like he did in the second half), they will drop more than a few games.


There’s a huge disagreement among different advanced statistics for how the passing game performed. Some of it has to do with the fact that Bradford threw two dropped interceptions early on. Some of it has to do with the fact that different statistics assign different credit for sacks. ESPN QBR and ANYA (adjusted net yards per attempt), for example, largely treat sacks like quarterback statistics, while AYA (adjusted yards per attempt) does not.

The Vikings were abysmal on third down when passing; two sacks and four incompletions as well as three completions short of the sticks hurts Bradford. The Vikings passing game only converted three third down attempts through the air, compared to nine failures. That’s big for the formula that ESPN QBR uses, which is dependent on a concept called “expected points added,” which you can find explained here. Generally speaking, third down failures have negative expected points added, while conversions have positive expected points added.

Functionally, QBR eliminates the 16 yards that Bradford gained on third down completions that didn’t convert a new set of downs.

Below is a table telling us how Bradford ranks in various metrics. The left column is the name of the metric (again, a review of what they mean here in the Week One piece), followed by the rank that Bradford achieved that week among other passers in the same week. The next column compares his weekly score to the full 2015 season to give us a better context for how that performance looks in general. The final column on the rate looks at Bradford’s year-to-date performance in those categories, so how his two games stack up to the two or three games other passers have put together so far this year.

Metric This Week Rank vs. 2015 Total 2016 Rank
PFF 11 11 3
Implied YPA 26 30 13
Passer Rating 11 15 5
ANYA 18 26 10
ESPN QBR 24 32 20

Generally speaking, these metrics argue Bradford was not accurate enough for how deep his typical throw was downfield and that he generally did not produce that many yards per passing attempt, though the touchdown was nice. The Pro Football Focus grade essentially says that on a play-to-play basis, after accounting for some context, that Sam Bradford did better than the box score would suggest. That abysmal first half (4/11 for 26 yards, or 2.4 yards per attempt, and a passer rating of 44.9) really dragged him down.

We still have an enormous disconnect on his season-long grades, with the lowest at 20th and the best at third. While we know his accuracy-for-depth (as represented by Implied YPA) has been average, he gets credit for avoiding turnovers. Those sacks are really hurting him in ANYA and QBR, however, and some of them are probably his fault—especially in the first half of the Green Bay game.

If you gave him some credit for how poor the offensive line has been and therefore don’t weight the statistics that penalize his sack rate—but don’t think his relatively average accuracy and depth of target are too worthy of high praise, he’s been performing like a top ten (but not a top five) quarterback thus far.

That’s certainly not bad.

As a note, Bradford was once again under heavy pressure, though with only two early sacks it’s easy to argue the offensive line performed well. On rewatch, I haven’t been so sure and even T.J. Clemmings—whose name wasn’t called that often by the broadcast—performed very poorly given his assignments.

Pro Football Focus has Bradford under pressure on 40 percent of his snaps in the game, which is better than the Green Bay score, but still a bottom-five pressure rate in general. Had the Vikings allowed pressure on 40 percent of snaps last year, they wouldn’t have ranked last like they did, but still would have been among the worst pass protectors in the league.


The primary receiving statistic we’ve been tracking is yards per route run. This is the most reliable and role-independent measure of receiving play. After Stefon Diggs’ pedestrian 40-yard outing against the Panthers, it makes sense that his YPRR will drop but it’s still pretty excellent. First, this week:

Rank Player YPRR
43 Stefon Diggs 1.40
47 Adam Thielen 1.35
84 Charles Johnson 0.00
1 Cordarrelle Patterson 8.00
5 Kyle Rudolph 2.69

Rudolph’s rank is relative to other tight ends. Cordarrelle Patterson ran one route, so he has that going for him.


Rank Player YPRR
1 Stefon Diggs 3.37
46 Adam Thielen 1.64
97 Charles Johnson 0.20
1 Cordarrelle Patterson 4.00
10 Kyle Rudolph 1.89

Patterson has run two total routes, so he would rank first if he qualified. He doesn’t so it’s Stefon Diggs, with a still-absurd 3.37 yards per pass route run rate.


Watching the game, it’s easy to argue that Jerick McKinnon and a somewhat new-look offensive line produced a far better rushing attack than in the previous two games. This could be built off of an initial impression, however. In the first half, McKinnon had a very high success rate of 50 percent, 6.0 yards per attempt, and 1.6 yards over expected per carry (a measure built off of giving credit for yards over the average running back in the same down and distance situations).

He continued this trend into the third quarter but tapered off into the fourth when asked to run the clock off for the Vikings after they developed a two-score lead.

In the end, he got stopped for a loss or at the line of scrimmage so many times in the fourth quarter that his overall rates went down. Let’s look at how McKinnon did in the first three quarters in these two statistics and compare it to the fourth quarter:

Quarters Yards Over Expected Success Rate
1-3 0.41 56%
4 -4.3 0%
1-4 -1.7 29%

To get a good understanding of “yards over expected,” just add 4.3 to the number and you’ll get an approximate equivalent running back yards per carry. So McKinnon, after accounting for down and distance, functionally averaged 4.7 yards a carry in the first three quarters and 0.0 yards per carry in the final quarter, for a total functional average of 2.6. That’s actually pretty close to his real yards per carry of 2.8 in the game.

A success rate of 56 percent would be the highest in the NFL over the course of a season, and ninth among all running backs with at least ten carries in Week 3. A yards-over-expectation of 0.41 ranked 18th.

But his final totals of 29 percent and -1.7 ranked 30th.

Matt Asiata didn’t have an interesting quarter split, and his expected yards over attempt came out to -1.4, while his success rate was 17 percent—one successful run of six attempts.

Jerick McKinnon kept up his solid yards after contact, with another 2.2 average (the 2015 average). Asiata plugged in a pedestrian 1.5 yards after contact, but that’s not too different than what he’s put in before. That means that before contact, McKinnon only had 0.6 yards while Asiata had 1.1 yards. Again, some of this speaks to blocking on the part of the line, while some of it speaks to the vision and agility of the running backs.

Here are their season ranks among all running backs with at least 15 carries (54 running backs):

Player After Contact Rank Before Contact Rank Success Rate Rank Yards/Expected Rank
Jerick McKinnon 44 45 51 49
Matt Asiata 52 43 54 50

Though McKinnon’s yards-after-contact rank looks awful, a lot of running backs have been generating more yards after contact than usual this year. As the year goes on, that will die down. If McKinnon bucks that regression, then he’ll rank solidly in the 20s. The rest of the numbers look pretty bad.

Based on what I’ve seen, I’d argue that most of the issue comes from the offensive line, though the fact that six of McKinnon’s 19 carries came in run-out-the-clock mode probably plays a significant role as well. Without them, his success rate overall ranks 40th and his yards over expected number ranks 24th.

Run Defense

The Vikings did a fantastic job stopping the Panthers from making big gains on the ground running the ball, but Carolina did do a good job getting the minimum yardage needed to keep the offense on schedule. Their success rates weren’t stellar, but not too bad:

Player Success Rate Yards over Expected
Cameron Artis-Payne 41.7% -0.0
Fozzy Whittaker 40.0% -1.9
Cam Newton 80.0% -1.1
Total 50.0% -0.77

While stopping Cam Newton was difficult, the Vikings did a generally good job of containing the run and preventing either Artis-Payne or Whittaker from punishing them on the ground.

Who was involved in it? We’ve talked about stop rates before, and essentially it disconnects the relatively useless “total tackle” count and replaces it by counting “good tackles” instead of rewarding players who make the tackle after they allow something in coverage or after recovering off of a block they got pushed downfield for.

Player TOT SACKS Stops Stop Rate
Chad Greenway 3 0 3 100.0%
Shamar Stephen 1 0 1 100.0%
Linval Joseph 7 1 5 71.4%
Anthony Barr 7 1 4 57.1%
Eric Kendricks 12 0 5 41.7%
Danielle Hunter 5 1 2 40.0%
Harrison Smith 7 1 2 28.6%
Everson Griffen 4 3 1 25.0%
Andrew Sendejo 8 0 1 12.5%
Captain Munnerlyn 9 0 1 11.1%
Terence Newman 4 0 0 0.0%
Trae Waynes 4 0 0 0.0%
Brian Robison 2 1 0 0.0%
Anthony Harris 1 0 0 0.0%
Tom Johnson 2 0 0 0.0%
Emmanuel Lamur 1 0 0 0.0%
Mackensie Alexander 1 0 0 0.0%
TEAM 78 8 25 32.1%

Chad Greenway did extremely well, while linebackers Anthony Barr and Eric Kendricks both did better in the run game, seemingly, than I gave them credit for.

Linval Joseph is continuing to prove that he’s, statistically, the best nose tackle in the NFL and it’s not particularly close. This week was even better for him statistically, as he added an absurd five stops and pressured the quarterback on ten percent of dropbacks. Given that the top pressure rate for nose tackles last year was 10.3 percent—and it was by a nose tackle not particularly adept at stopping the run (Eddie Goldman)—that’s astounding.

Among nose tackles this year, Joseph ranks first in stop rate and seventh in pressure rate. He is one of only three nose tackles to place top ten in both categories. Of the 17 nose tackles with a stop rate above eight percent, only one has a sack—Linval Joseph, who now has three.

Pass Pressure

Aside from Joseph, the Carolina game was predictably a pretty good game for the Vikings in terms of creating pressure. Here are the pressure rates for the defensive linemen and how they compare to the rest of the league at their position (edge defender or interior defender) for the week:

Player QBP Rank
Shamar Stephen 20 2
Linval Joseph 10 18
Tom Johnson 9.1 24
Everson Griffen 23.1 5
Brian Robison 11.8 24
Danielle Hunter 16.2 14

And here’s how they compare to their peers based on their year-to-date pressure rates:

Player QBP Rank
Shamar Stephen 4.2 54
Linval Joseph 7.5 31
Tom Johnson 11.5 9
Everson Griffen 12.6 26
Brian Robison 8.8 56
Danielle Hunter 14.3 15

Everson Griffen had a great week, but he’ll need to be more consistent in order to score higher as a pass-rusher over the season.

Pass Coverage

Football Outsiders has begun its charting project, and it’s well worth the price of admission. Week 3 statistics aren’t up, but going through how the cornerbacks have done when targeted over the past two games can be illuminating.

FO agrees with PFF that Trae Waynes has been targeted more than any other cornerback in the league, and teams seemed to pick on him through two weeks. That didn’t happen this last week, and it will be worth looking at how the cornerbacks looked after a unique gameplan to deal with Kelvin Benjamin and functionally ignore other receivers. So, while knowing that we don’t currently have Week 3 statistics, we can look at how the defensive backs did in Weeks 1 to 2.

Player Targets Yd/Pass Rank
Terence Newman 12 4.3 20
Captain Munnerlyn 3 4.7 22
Trae Waynes 26 7.3 55

That’s out of 97 qualifying cornerbacks.

In Week 3 specifically, Waynes was targeted four times and likely retains his lead as the most picked-on corner in the NFL.

Player Cover Snaps Targets
Captain Munnerlyn 22 5
Trae Waynes 26 4
Terence Newman 25 3
Xavier Rhodes 27 2
Harrison Smith 36 2
Andrew Sendejo 36 1

If one estimates how Pro Football Focus would divvy up the targets from watching the game quickly, one can come up with the overall coverage statistics for those corners:

Player Targets Total Cvg Snaps Yards Per Target Snaps/Target Yards per Cover Snaps
Terence Newman 15 61 101 4.07 6.733 0.60
Captain Munnerlyn 8 79 77 9.88 9.625 1.03
Trae Waynes 29 220 94 7.59 3.241 2.34

In yards give up per target, Newman would have ranked first in 2015. Munnerlyn would have ranked 110th (of 131) and Waynes would have ranked 65th.

But yards per target is probably the worst of all the statistics because good corners aren’t targeted often—so when they are targeted, they might have made a rare mistake and have an inflated score. So the combination of target frequency and yardage given up is better.

For example, Munnerlyn’s 9.9 coverage snaps per target would have ranked first in 2015. It’s significant that he’s not targeted that often and that’s because teams don’t target slot receivers when he covers them. Newman’s 6.7 snaps per target ranks 28th of 131 CBs while Waynes would have ranked dead last out of all 131 corners.

In total yards given up by snaps in coverage, there’s a happy medium: deterring targets and making the targets not count for much both give information.

Newman’s 0.6 yards per snap in coverage given up is better than any corner in 2015, 2014 or 2013. It will come back down to Earth, but quarterbacks really don’t like targeting him and he’s made them pay when they have. Not noted above are turnovers and Newman’s interception changes the calculus completely.

Munnerlyn’s 1.03 yards per snap in coverage would rank 36th and that’s more than respectable for a slot cornerback. Trae Waynes, on the other hand, ranks 130th. Those two interceptions certainly make up for it from a productivity standpoint, but two interceptions isn’t worth over 200 yards given up in three games (not to mention a touchdown given up to Jordy Nelson in the prior week)—especially as one interception occurred on the final play of a half on a meaningless deep shot.

The Vikings should be concerned about their offense. Every indicator we have tells us that even with this level of defensive play, winning is not sustainable with that poor of an offense. But the defense is fantastic, anchored by an incredible defensive line that might be playing better than any other defensive line in the NFL. Add to that, solid run support from the linebackers and great coverage by safeties means the Vikings are almost impossible to attack.

The post Arif’s deep stats dive: Vikings’ excellent defense propping up worrisome offense right now appeared first on 1500 ESPN Twin Cities.

Source:: 1500 ESPN Sportswire

Leave a Reply

You can use these HTML tags

<a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <s> <strike> <strong>