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

Hasan’s deep stats dive: Week 1 dud casts concern on Vikings running game

By Derek Wetmore

NFL: Minnesota Vikings at Tennessee Titans - Sun, 11 Sep 2016 15:14:29 EDT

BY ARIF HASAN

Follow Arif on Twitter

Every week, we’ll take a look at some of the advanced statistics behind the most recent Minnesota Vikings game and catch readers up on the cumulative statistics that give us context for the season.

None of this is to say that football is played on paper, but rather that the statistical tools we typically have—passer rating, completion rate and so on—don’t usually give us the full picture.

Team-Level Statistics

Before looking at individual player statistics, it’s often useful to take a step back and look at overall team performance. There are some better statistics than points scored, points allowed, or total yards gained that give us a clearer understanding of how the team may perform in the future.

Some teams compile a lot of yardage but still maintain a poor offense because they play a lot of snaps in a no-huddle system, or throw the ball much more than they run it. They still might be inconsistent or turnover-prone, however, and might still be poor units.

Instead, points per drive tells us how successful a team is on a per-drive rate that accounts for the fact that teams like the Vikings play a lot slower than a team like last year’s Eagles. For example, last year the Jaguars played 190 offensive drives while the Vikings only played 168.

Much of that might be due to the Jaguars’ quick-strike offense, but a lot of it had to do with turnovers and a defense that allowed teams to score quickly, putting the offense back on the field. From a total-points basis, the Jaguars earned more from their offense than the Vikings, with 318 points to the Vikings’ 306.

Jacksonville ranked 13th in offensive points (for points, we only use rushing touchdowns, reception touchdowns and field goals while extra points. For drives, we don’t include those ending in a kneel) while the Vikings ranked 16th. But in points per drive, the Vikings ranked 11th in the league while the Jaguars ranked 26th.

Generally speaking, any points per drive number above 2.0 is excellent and top-five offenses generally produce this much. Anything below 1.5 is worrisome territory, in the bottom third of the NFL. Below 1.4 and you’re dealing with Browns-level problems.

The Vikings scored 12 offensive points on nine drives, for a less-than-stellar total of 1.33 points per drive. As an average, that would have ranked only above the 49ers, Rams and Titans last year.

On defense, they allowed 1.36 points per drive, but the last one was in garbage time. For the purposes of this column, we won’t exclude it in the cumulative standings, but it is important to note each week whether or not any specific scores will skew the statistic.

With a PPD differential of -0.03, the Vikings played at about the talent level of their opponent. That’s probably not encouraging when talking about a team that has only one home games against the Jaguars over the last several years, but it’s also not too surprising after what we saw.

Incidentally, defensive scores definitely count when figuring out who won the game, but don’t count for this statistic because it’s simply not a consistently viable method of producing points. Only five teams since the merger have score more than half a defensive touchdown a game—five team-seasons out of 1,321 possible team-seasons, less than half a percent.

Those teams together ended up winning exactly half of their games. So the Vikings will have to be consistent on offense and defense if they want to argue that they are a playoff team.

Another method that takes a more granular look at how the teams perform is called Drive Success Rate. It looks at how often teams have to “give up” and punt while still accounting for field position. The Vikings defense forced the Titans to punt on their first drive, but there was some trepidation from Vikings fans because Tennessee found ways to get first downs early.

A statistic like Drive Success Rate captures that and incorporates it into its measure. The calculation is simple—first downs converted plus touchdowns divided by first down opportunities.

Every time a team gets a new set of downs, that’s another first down opportunity—so a team that converts three first downs and punts would have a DSR of 0.75, whether or not it took them three plays to grab those conversions or seven.

It does an even better job than points per drive in predicting team success and it isolates those teams that can move down the field efficiently instead of relying on one-shot plays that they cannot repeat. Something like third down conversion rate doesn’t capture that because it doesn’t account for teams the consistently convert on first or second down, like DSR does.

The top offensive DSR last year belonged to the New Orleans Saints, followed by the Arizona Cardinals, both with a DSR a hair above 0.75. At the bottom were the 49ers, Rams and Titans once more, with DSRs ranging from 0.629 (San Francisco) to 0.598 (Los Angeles née St. Louis).

The Vikings converted 15 first downs on 9 non-kneeling drives, with no touchdowns. That makes their DSR 15 divided by the sum of 15 and nine, or 15/24 (0.625). As you can figure out from the teams languishing at the bottom, that’s a poor DSR.

The Titans converted 17 first downs and two touchdowns on 11 drives, for a DSR of 0.678. That’s pretty poor, but not as poor as the Vikings’ DSR. The net of -0.053 would have ranked 29th last year.

Passing

Generally speaking, passing statistics are supposed to give us an idea of how good a quarterback is. In reality, they generally give us an idea of how the passing offense performs—the quarterback, receivers and offensive line. Some allow you to separate performances, like rating under pressure, while others attempt to account for the talent of receivers.

There are two measures that attempt to account for receivers. The first is ESPN’s QBR, which actually performs extremely well when attempting to predict wins. It attempts to account for a variety of variables, like pressure, difficulty of throw and receiver drops.

Shaun Hill’s QBR of 63.0, significantly higher than NFL average generally, and specifically 15th of the 28 quarterbacks who had played entering Monday.

One thing Hill was good at on Sunday was sack avoidance in the face of pressure, and a statistic that might be easier to understand is adjusted net yards per attempt. All it does is assign values to touchdowns and interceptions—20 for TDs and negative 45 for INTs—and add sacks and negative sack yards to the equation. Quarterbacks tend to be much more responsible for sacks than analysts like to admit, so this captures QB play better than a lot of people would initially think.NFL: Minnesota Vikings at Tennessee Titans - Sun, 11 Sep 2016 13:39:00 EDT

In 2015, the top ANYA belonged to Carson Palmer, with 8.41 adjusted net yards per attempt. The bottom ANYA belong to Peyton Manning, with 4.52 adjusted net yards per attempt. The average team in the NFL had 6.3 adjusted net yards per attempt.

Because Hill took no sacks and threw no interceptions (also throwing no touchdowns), his ANYA and yards per attempt are identical, and at 7.2, it’s a fair bit higher than most quarterbacks in the NFL. Passer rating, which emphasizes completion percentage more than other quarterback metrics, has him lower overall, with a below-average 77.3.

Also rating him below average was Pro Football Focus, where he graded as the 26th-best passer out of 28 from last weekend. Like ESPN’s QBR, PFF is supposed to take into account context like down-and-distance and teammate performance, so seeing a wide spread is somewhat unusual. I suspect the biggest difference is that PFF downgrades Hill for the dropped interception more than anyone else does.

One final statistic is “implied” yards per attempt, which looks at a quarterback’s accuracy on throws that were not drops, throwaways, spikes or thrown while being hit, and multiplies that accuracy against their average depth of target—which allows you to compare gunslingers who throw deep but not for completions very often to game managers who have a high completion rate but don’t take risks.

Just like yards per attempt, the average for an NFL quarterback is 7.3. Over the course of a year, the best performance will be around 8.5, but individual games can rise above 10.0. At the bottom, teams can average 6.3 over a season and can drop below 4.0 in individual games. Shaun Hill put up a respectable but not great 6.95, ranking 18th of the 28 passers from the weekend.

PFF credited Shaun Hill’s receivers with three drops and excused Hill for one throwaway, but did not exclude any throws he made while hit, though I think the near-interception deserves more scrutiny from those guys.

Here’s how the passing game under Shaun Hill looked by each of the metrics

Metric Rank
ANYA 12
ESPN QBR 16
Implied YPA 18
Passer Rating 21
PFF 26

The best way to interpret all of these advanced statistics is to look at how they measure the same performances differently. Because Hill’s play-to-play grade (PFF) is low and his passer rating is low, it stands to reason that he wasn’t consistently accurate, as both preference completions.

His relatively average implied YPA points to the fact that his completion rate was low in part because of how deep his passes were, while his high ANYA and QBR point to the fact that despite missing more than a few passes, his dropbacks didn’t result in big mistakes.

Receiving

There aren’t a lot of receiving “advanced statistics,” though there are a few pieces of data worth looking into. The first thing is to check out drops from the receivers, tight ends and running backs in the passing game. Among wide receivers, PFF logged no drops. Kyle Rudolph had one drop, while both Jerick McKinnon and Adrian Peterson each recorded one drop as well.

A lot of fantasy statisticians like to look at yards per target, but I don’t like that because targets themselves are an indicator of quality. I’ll let Adam Harstad explain:

@ketom1220 @Marcus_Mosher Targets are an indicator of quality. Say 5 WRs run a route, one gets a target, and the ball falls incomplete.

— Adam Harstad (@AdamHarstad) September 11, 2016

@ketom1220 @Marcus_Mosher Who was the best WR on the field? Probably the guy who got the target.

— Adam Harstad (@AdamHarstad) September 11, 2016

@ketom1220 @Marcus_Mosher You get targets when plays are designed to go to you, or when you get open, or when you win contested balls.

— Adam Harstad (@AdamHarstad) September 11, 2016

@ketom1220 @Marcus_Mosher So the guys getting a lot of targets are the guys who are really, really, really good.

— Adam Harstad (@AdamHarstad) September 11, 2016

A statistic that improves once a player does something that indicates lower-quality play is not ideal and one reason statisticians have had issues with passer rating.

Instead, we should look at the yards they get for every opportunity they have to get yards. Opportunities are defined by passing plays, not targets—because targets are just evidence that the first half of the job of a receiver (getting open) was accomplished.

In this case, we’ll look at yards per route run. This way, we account for the fact that the offense doesn’t always throw the ball and accounting for how often a receiver is on the field. It’s not everything—the top receivers will see their yards per route run drop simply because defenses key in on them, and receivers on good receiving corps suffer because their teammates take yards.

The best YPRR last year belonged to Julio Jones, followed very closely by Antonio Brown at 3.04 and 2.93 yards per route run respectively. Generally speaking, elite receivers will get above 2.5 yards per route run, though some low-sample receivers (like Rex Burkhead) or situational receivers will get up there, too. On average, receivers will generate 1.5 yards per route run and poor receivers will drop below 1.0. Mike Wallace last year was last on the Vikings with a YPRR of 1.1.

Because individual games will have greater variance, individual receivers can end up with very high scores. J.J. Nelson only had one reception but because he only ran two routes, his YPRR led the league with 5.5. You’ll also find a fair number of receivers with 0 YPRR, like Charles Johnson, that should get resolved over time. Here are how the Vikings receivers did:

Player YPRR
Stefon Diggs 3.08
Cordarrelle Patterson 2.96
Adam Thielen 2.38
Charles Johnson 0.00

That should be extremely pleasing, even with Charles Johnson’s dud. Stefon Diggs had a hell of a game despite a backup passer slinging the ball to him. Cordarrelle Patterson did well too, with a critical catch in a normal receiver situation.

Adam Thielen’s performance was naturally more impressive than Patterson’s, but these things happen in small samples.

NFL: Minnesota Vikings at Tennessee Titans - Sun, 11 Sep 2016 14:17:50 EDT

Tight ends naturally have a lower YPRR as they are out there on receiving plays almost as often as wide receivers are but get far fewer yards. Last year’s best in YPRR was Jordan Reed, with 2.44. The worst was Garrett Graham, at 0.17. The average was 1.44 yards per route run. Typically, elite receiving tight ends get more than 2.25 yards per route run and poor ones dip below 1.0.

Kyle Rudolph was the only tight end to run routes in passing situations for the Vikings and had a very respectable 2.17 yards per route run.

Running

Running means both the ability of the running back to carry the ball and the ability of the offensive line and play design to allow those things to happen.

The available evidence we have is that all facets of play failed against Tennessee. The line blocked poorly, while Peterson played poorly. Usually, when only one of those is the culprit, a mediocre running line is produced—something like 2.2 yards a carry. When they conspire together, you can see a result like 1.6 yards a carry, what Peterson produced behind the Vikings offensive line Sunday.

There are a few tools we can use. One of them relates to how the running back deals with contact—either missed tackles per touch or yards after contact. Last year, Le’Veon Bell led the league in yards after contact per carry, with 3.4. C.J. Spiller disappointed with a league-low 1.4.

The highest yards-after-contact per attempt naturally belongs to a back with one carry, Jalen Richard (25 yards after contact, on a 28-yard carry), but the highest among those with at least ten carries belongs to David Johnson, with an eye-popping 4.1 yards after contact—not a bad number in total to be honest.

Peterson languished with 1.3 yards after contact (which is distinct from the 0.5 yards after contact per carry ESPN Stats & Info credited him with). For context, between 2007 and 2013, Peterson demolished the league in yards after contact with 3.2 over the course of seven seasons and never fell below 2.9. No other running back accomplished anything remotely close—even as backs like Bell and Spencer Ware accumulate single-season accolades in the statistics, only Peterson has sustained.

This was true between varying degrees of quality across his offensive lines. Between 2010 when his run-blocking line was the fourth-worst in PFF’s run-block grading system, and the very next year when the line led the league in run-block grading, Peterson has seen it all and created yards after contact regardless.

Hopefully, this bit of poor play doesn’t continue.

NFL: Minnesota Vikings at Tennessee Titans - Sun, 11 Sep 2016 12:38:07 EDT

But it’s not all on him, either. It is rare to see a running back with so few yards before contact, too. Generally, people will make the mistake that yards-before-contact gives you an idea of how well the offensive line is doing, but skilled running backs will consistently avoid contact as well. Like regular yards per attempt, yards-before-contact is a mixture of both the offensive line and running back play.

Anecdotally, Peterson looked hesitant when attacking runs on the edge and perhaps too aggressive to seek contact when attacking runs on the inside, so it certainly makes sense that he had issues generating his own yardage that contributed to his low yards before contact.

But it’s difficult to conceive of a good offensive line only allowing 0.3 yards before contact.

Only two running backs had fewer yards before contact than Peterson: Terrance West with the Ravens and Devonta Freeman with the Falcons. Tied with Peterson were T.J. Yeldon of the Jaguars, Doug Martin of the Buccaneers, James Starks of the Packers and Thomas Rawls of the Seahawks.

He tied for 27th among 30 qualifying backs.

This will rebound; no running back last year had fewer than 0.6 yards before contact and the league average was 1.75, but it’s certainly concerning.

We can certainly attribute a large share of that to the offensive line, as we saw a number of snaps where Alex Boone or Andre Smith gave up the block that led to the tackle or flushed Peterson into the waiting arms of another defender.

Aside from just looking at yards before and after contact to figure out the truth, we can look at the success rate of runs. Research indicates that run success rate, not yards per carry, correlates with winning future games and increased passing success.

It makes sense intuitively, too—defenses do not have a running YPC clock in their heads, but they do have a general sense of how often a running back has been successful against them. Figuring out how often a running back does a “good enough” job can help determine their value. Typically, runs are successes on first down if they get 40 percent or more of the required yardage to convert a new set of downs. On second down, that percentage jumps to 60 percent and on third and fourth down, it jumps to 100 percent.

Running backs are generally successful between 34 percent of the time (LaRod Stephens-Howling last year) and 62 percent of the time (Thomas Rawls)—though 62 percent is a bit high, as the high in the previous several years were lower. In 2014, Lamar Miller led the league with a 57 percent success rate. In 2013, Danny Woodhead led the league with a 60 percent rate and in the year before that, Willis McGahee led the league with a 58 percent rate.

On average, running backs seem to be successful 47 to 48 percent of the time.

Adrian Peterson’s success rate in the game was 26.3 percent.

Game to game success rates can vary, of course, but that’s extremely poor and it speaks to the state of the running game overall, whether or not it’s Peterson, play design, the offensive line or a combination of all three.

Success rate isn’t everything, though. It also matters that yards are gained in critical situations over general first-and-ten scenarios, and explosive yards are good, too. A statistic called yards over expectation adjusts running backs’ yardage for the fact that some of them get to run in favorable situations, like second-and-ten (where the league average is 4.77 yards a carry) and others need to run it on third-and-two more often (where the league average is 3.68 yards a carry).

It also helps us resolve comparisons between players like Jerick McKinnon and Adrian Peterson, who had comparable yards per carry numbers last year, but ran in wildly different situations.

The best yards over expectation over the last four years belonged to Jamaal Charles, who ran 1.04 yards over expectation per carry. The worst belonged to recently-cut Andre Williams, who ran -0.95 below expectation per carry. Because of the nature of the metric, 0.0 is the league average. Adrian Peterson between 2012 and 2015 placed fourth, at 0.78 yards over expectation per carry for a total of 664.1 yards over expectation—the second highest total over that span (again, behind Jamaal Charles).

On Sunday, Peterson’s yards over expectation were -2.47 per carry and -47 yards total.

Certainly, there’s a lot to be concerned about here.

Coverage

On defense, there are two things to be concerned about, generally speaking—the run and the pass. Knowing how often a player is targeted is pretty critical information on its own and though we don’t have all the target information on hand, we do know that Trae Waynes was targeted more than any other defensive back who played at least 50 percent of his team’s snaps, soaking up 30 percent of Marcus Mariota’s targets.

Most of those were to Tajae Sharpe, who only had one reception to any other cornerback (Terence Newman), and Waynes gave up a reception to Andre Johnson as well. Waynes allowed seven receptions to those two receivers for 80 yards on 12 targets, for a yards per attempt of 6.7 and a passer rating of 85.

Compare that with Newman, who grabbed nine targets and allowed five completions for 47 yards (5.2 yards per attempt, rating of 70.1).

Without more charting data, we can’t break everything down completely, but remember that the top defensive backs in the NFL allow a passer rating below 50, and in some years below 45.

Pass Rush

The Vikings boasted a powerful pass rush last year, but didn’t have any individual players dominate in pressure rate—with Everson Griffen as the best of the bunch, converting 14.3 percent of his pass rush opportunities into pressure and ranking 11th in the NFL.

This year they seem to be operating at the same pace, with a number of players all boasting a relatively high pressure rate, but none of them dominating over the others. For context, the top pressure rate last year belonged to J.J. Watt at 15.2 percent, while the worst pressure rate among frequent players was 2.4 percent for Nick Hayden.

For edge-rushers specifically, the low-water mark appears to be Mario Williams’ 7.3 percent pressure rate.

The highest pressure rate belonged to situational rusher Danielle Hunter, who got to the quarterback 11.5 percent of the time he rushed the passer. Behind him was Everson Griffen, whose 9.1 percent pressure rate wasn’t great but it wasn’t necessarily awful, either. Situational three-technique Tom Johnson was next at 8.3 percent—a very high rate for a defensive tackle—and Sharrif Floyd wasn’t too far behind at 7.7 percent.

Linval Joseph’s 4.8 percent is relatively high for a nose tackle, but low for him. Last year, he got pressure at an unbelievable 9.4 percent rate. Dontari Poe, one of the best nose tackles in the NFL, gets pressure at about a 5.6 percent clip.

The only truly worrisome rate comes from Brian Robison, who used to rack up a ton of pressure in the NFL as a 4-3 DE but never earned the sack numbers to back it up. Against Tennessee, he didn’t even have that, only getting pressure 5.6 percent of the time—same as nose tackle Dontari Poe last year, and lower than the league low for 4-3 defensive ends.

There were also quarterback hits by Danielle Hunter, Linval Joseph, Everson Griffen and Anthony Barr.

Run Defense

Luckily, there’s another aspect to playing along the defensive line that has nothing to do with rushing the passer and that’s stopping the run.

Overall, the Vikings were a successful run defense unit against a growing Titans unit that has two talented running backs, a running threat at quarterback and an offensive line rapidly building on its reputation as a mauling presence.

The Titans averaged 3.0 yards a carry despite some bigger runs in the first half, and their running backs only managed 2.5 yards a carry.

The only caveat is that though the Titans didn’t get a lot of yardage per carry, they still found time accomplish their goals. DeMarco Murray averaged a success rate of 46.2 percent and Derrick Henry had a success rate of 40 percent. With Mariota’s rate of 75 percent, the sum total of the running game led to a success rate of 50 percent overall.

At least the Vikings did a good job against the running backs.

One good way to tell a solid run defender is to see if they created tackles for offensive run failures, referred to as “stop rate” by PFF. It’s useful to look at totals (total stops, which includes tackles for loss) as well as percentages.

A player with a high number of tackles but a low number of run stops might be doing a poor job because they allow runners to go too far before bringing them down or allowing receivers to get behind them and tackling them after they make a reception.

For example, Trae Waynes led the team with ten total tackles. Only one of them was a run stop (and an impressive one at that). His stop rate was only 10%. On the other hand, Andrew Sendejo had four tackles, but two of them were run stops—a stop rate of 50%. Here’s the stop total by player and the stop rate:

Player Total Tackles Sacks Stops Stop%
Danielle Hunter 4 1 2 50.0%
Andrew Sendejo 4 0 2 50.0%
Brian Robison 2 0 1 50.0%
Anthony Barr 2 0 1 50.0%
Harrison Smith 8 0 2 25.0%
Linval Joseph 5 1 1 20.0%
Eric Kendricks 6 0 1 16.7%
Terence Newman 7 0 1 14.3%
Trae Waynes 10 0 1 10.0%
Everson Griffen 3 0 0 0.0%
Adam Thielen 2 0 0 0.0%
Marcus Sherels 1 0 0 0.0%
Jerick McKinnon 1 0 0 0.0%
Jayron Kearse 1 0 0 0.0%
Emmanuel Lamur 1 0 0 0.0%
TEAM 57 2 8 14.0%

It looks like Danielle Hunter was the only player who was consistently a force in both the run and passing game, though metrics like this don’t account for other important factors, like if a defensive end overruns the play too often. They may get some stops, but they could also be a bigger liability than an asset.

Still, this provides us with a more complete picture.

Pro Football Focus

In recent years, PFF has gained prominence nationally as a football grading organization that attempts to grade every single player on every single play. While it isn’t perfect, their constant revision process and communication with the NFL improves their process. The fact that they have former NFL and college coaches on staff has certainly helped, too.

The grades they provide on players are a controversial, sometimes wrong, but often valuable tool that helps us understand how detailed evaluators looked at the game. Below are the general tiers (Elite, Very Good, Good, Average, Below Average and Poor) they rated last week’s starters:

Player Position Tier
Shaun Hill QB Poor
Adrian Peterson HB Poor
Kyle Rudolph TE Average
Stefon Diggs WR1 Very Good
Charles Johnson WR2 Poor
Adam Thielen WR3 Average
Matt Kalil LT Poor
Alex Boone LG Poor
Joe Berger C Average
Brandon Fusco RG Poor
Andre Smith RT Poor
Everson Griffen RDE Poor
Sharrif Floyd UT Poor
Linval Joseph NT Average
Brian Robison LDE Below Average
Terence Newman LCB Poor
Trae Waynes RCB Poor
Captain Munnerlyn SCB Below Average
Anthony Barr SLB Poor
Eric Kendricks WLB Average
Harrison Smith FS Good
Andrew Sendejo SS Average

That is probably too harsh an assessment, especially of players like Sharrif Floyd and Everson Griffen, but some of the stalwarts, like Anthony Barr, deserve to be dinged. The clearly best-graded player was Stefon Diggs, followed closely by Harrison Smith. PFF didn’t give Kendricks too much for his interception and clearly thought other parts of his game were lacking.

They praised Joe Berger in their review of the game, and that kind of praise seems more worthy of an above-average player than an average one.

Next week, the article will be significantly shorter as we won’t have to explain every term and just get into the details immediately, but this should provide us with a lot more context to the games we’ve been watching than simple box score numbers like passing yards and total tackles.

The post Hasan’s deep stats dive: Week 1 dud casts concern on Vikings running game 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>