People are not very good at numbers.
That’s something that’s true about humans in general, and sports fans in particular. That’s fine; I’m bad at understanding numbers as well, but for the sake of accurate commentary I try to overcome it as often as possible.
The reason I bring this up is because I’ve been seeing a lot of it with regard to Teddy Bridgewater and Shaun Hill.
The loss of Teddy Bridgewater as a result of a gruesome knee dislocation has sent Vikings fans and the beat scrambling for a replacement at quarterback that can dispossess Shaun Hill of the starting job, in part because of how awful Hill looked in training camp.
It has also generated the opposite response: that there will be virtually no difference.
The Loss
https://twitter.com/FantasyBeast15/status/770792429251362816
Things don’t change too much when we take into account the fact that numbers from when Shaun Hill was with the 49ers at age 28 are not relevant with Minnesota in he’s 36. Hill’s numbers over the last three years (mostly in St. Louis) are still in some respects comparable. A completion rate of 62.3%, a touchdown rate of 3.4% and an interception rate of 3.0% and a passer rating of 82.5 seem pretty close to Teddy’s numbers above.
Their yards per attempt were pretty similar, too (Hill’s 7.1 is not too different from Bridgewater’s 7.2).
And, for a long time, the best numerical tools we have would argue that Shaun Hill is not too far off from Bridgewater, with an adjusted net yards of 5.59 dropping to 5.45 (or, to use 2015 numbers, 5.70 to 5.45).
That is a great tool, but it can only provide approximations. Most of the time, they give us a pretty good accounting of how well a quarterback plays. Adjusted net yards per attempt (ANYA) provides 70% of an accounting for why a team wins, and gives us nearly a quarter of the information we need to predict wins.
But at the moment, we have better numerical tools.
Not only does more information give us different angles through which we can solve the same puzzle, some of it is simply more predictive. When all of those tools agree, then there is a strong possibility that what they imply is true. A quarterback who is at the bottom of many of those metrics is likely bad, while one at the top of many metrics is good.
When those metrics disagree, sometimes one set is correct, while sometimes the truth is in the middle.
In the Chase Stuart piece linked above, he found that ESPN QBR was more accurate than ANYA, and in that Bridgewater ranks 13th of 33 qualifying passers (with a QBR of 62.7). Shaun Hill’s QBR in 2014 (41.4), where he threw 229 passes for the Rams, would have ranked 31st in 2015.
My research has corroborated the predictive power of ESPN’s QBR, and found two other measures that are even better: depth-adjusted accuracy and Pro Football Focus grades. Simply put, completion rate is an almost useless statistic because it doesn’t account for the difficulty or impact of throws. Once accounting for that, as well as getting rid of throwaways and dropped passes, one gets a more representative picture of a quarterback.
PFF grades are controversial but historically effective at predicting wins.
In depth-adjusted accuracy, Bridgewater ranked sixth. After adjusting for field placement, his rank stays the same, with an implied yards per attempt of 7.80. Shaun Hill’s implied yards per attempt in 2014, 6.89, would rank 29th in 2015 and he would tie with Blaine Gabbert.
As for PFF grades, it’s not too different. Per 1000 snaps (a typical season for a quarterback), Shaun Hill’s 2014 would rank 31st of the 34 quarterbacks who played at least 350 snaps last year. Bridgewater ranks 13th.
And none of this accounts for the strides that Bridgewater was making as a quarterback this year.
Does that mean the Vikings should find another solution at quarterback? Can they find a starter who can manage to create wins for the team without setting the franchise back in the following years?
Despite how bad Shaun Hill performs in these grades—and frankly, how he looked worse in camp—I don’t think there are many reasonable options at the moment to replace Hill, especially given how he improved over the course of camp and looked somewhat sharp in his preseason game against Seattle.
Think of it this way, Hill was a lock to make the 53-man roster before this Bridgewater injury, even if Taylor Heinicke had been fully healthy and able to replace Shaun Hill at the #2 spot. That puts him in a different class of quarterback than those that are available now—those that could not crack a 75-man roster, or in some cases, a 90-man roster.
In a quarterback-thirsty league, that’s telling.
That’s because most quarterbacks are bad.
It follows that quarterbacks who are easily available are worse even than the quarterbacks who are already bad but on rosters.
The Replacements
Another way to look at it is by checking out the rumor that former Falcons/Eagles/Jets/Steelers quarterback Michael Vick might be visiting the Vikings. He’s an exciting name, and there’s hardly a football fan that forgets his run in 2010, perhaps one of the most dynamic and incredible quarterback seasons we’ve seen.
UPDATE: this is no longer the case–
Vick’s implied yards per attempt over the last three seasons—with Pittsburgh, New York and Philadelphia—is 6.5. That would have ranked 34th last year alone. And excluding his final year in Philadelphia and looking solely at his more recent work, he drops to 6.3.
His box score numbers look just as bad, with a worse adjusted net yards per attempt than Hill—5.06 over the last three years and a shocking 3.71 over the last two. For context, no quarterback has thrown as many passes (181) and produced a worse ANYA since 2012. Four quarterbacks have done it since 2010: Jimmy Clausen, Blaine Gabbert, John Skelton and Brady Quinn.
That’s not all; Michael Vick’s PFF grade over the past two seasons prorated over 1000 snaps was -44.3, worse than any qualifying quarterback in the NFL last year by a significant margin. Colin Kaepernick, PFF’s worst quarterback, clocked in at -30.4, which was four points worse than second to last, Andrew Luck.
The average difference separating individual quarterback ranks is two points, so for Vick to come in significantly behind a group of quarterbacks who couldn’t crack the top 32 is telling.
This shouldn’t come as a surprise. Michael Vick was benched for Landry Jones despite playing behind PFF’s seventh-best pass protection unit and working with Le’Veon Bell and Antonio Brown—the best running back and best receiver in the NFL.
Vick’s ESPN QBR tells the same story. Over the last two years, it’s at 25.1. That, again, is lower than any quarterback put up in 2015 of 35 quarterbacks. In 2014, the lowest was 25.2. In 2013, the lowest was 21.3. And when the best receiver is Stefon Diggs instead of a player like DeSean Jackson, the problem could compound.
Last year, Pittsburgh allowed 32 sacks over 589 attempts—a rate of 5.15%. But, in only 11% of attempts, Vick racked up 31% of the sacks. Take Vick’s sacks and attempts out of it, and the Steelers give up a sack rate of 4.19%. Vick himself had a sack rate of 13.2%.
Last year the Vikings gave up more pressure than any other team in the NFL in the last decade of data that either Pro Football Focus or STATS, Inc. have.
Give a fumble-prone quarterback (11 fumbles in 454 passes, sacks and runs over the last three years) with a predilection to take sacks a much worse pass protection unit and watch the turnovers multiply.
Vick was just one example, too.
The vast majority of quarterbacks you could find without giving up any assets are not too different, just less well-known.
Vikings fans may remember Austin Davis as the quarterback benched for Shaun Hill. They will surely be familiar with Tarvaris Jackson and Josh Freeman. Yates, of course, lost a quarterback competition that included such luminaries as Ryan Mallett, Brandon Weeden and B.J. Daniels.
There are other avenues. Trading for a quarterback makes sense to some degree, if you’re convinced that a quarterback new to the system and to the players on the offense can surpass Shaun Hill within a few weeks.
Those few weeks are an important distinction. Carson Palmer had two Pro Bowl seasons before being traded to Oakland and one after. In his career, he’s only had one game with a net negative adjusted yards per attempt: the game he played coming off the bench five days after learning the playbook, against Kansas City. Unsurprisingly, it was also his lowest individual game QBR (2.5).
That year he had flashes of excellence—a 14/20 299-yard performance with two touchdowns and only one interception comes to mind—but he had as many games below a QBR of 40 as above it.
Josh Freeman had some terrifyingly bad games in his career, but it should come as no surprise that his outing against the New York Giants on Monday Night Football twelve days after arriving was his worst in terms of QBR since his rookie year, where only a three-interception loss to the Jets ranks below.
Give it a couple of weeks and a quarterback of real quality may be able to make an impact on the roster, even as merely a capable backup in case Shaun Hill gets killed behind the line of scrimmage.
There are some caveats to that, though.
The first is that the Vikings may already have that quarterback on the roster in Taylor Heinicke. I genuinely believe he can develop into a quality passer and word is that the Vikings expected Heinicke to take the backup spot over Shaun Hill in training camp had it not been for his baffling injury.
Last year’s preseason looked pretty good and I had to watch each snap a few times in order to generate a complete scouting report that captured his negatives. He was frighteningly accurate and decisive, and his quick release is the first clue that he might be able to avoid some of the issues that Hill (and sometimes Bridgewater) run into. Full scouting report, from the guide I wrote before camp:
Strengths: While not a particularly competitive award, Heinicke is clearly the most athletic quarterback in camp and displayed excellent quickness and speed at ODU and in the preseason. He’s a smart quarterback that seems to be making all the right reads; in the preseason he may not have made a single wrong decision with the ball (he recorded one INT, but it wasn’t his fault). He’s generally quite accurate, especially over the middle of the field and demonstrates an understanding of timing that allows him to throw into tight windows. He has a quick release and usually has rock-solid mechanics. Relatively young (23) for a second-year quarterback.
Weaknesses: If there are two fatal flaws for Heinicke, they would be his arm strength and response to pressure. His arm strength issues are similar to Christian Ponder’s; if his mechanics are perfect, there’s not much of a problem, though routes to the outside can cause issues. Otherwise, the ball will float. This didn’t impact him too much in the preseason but it would impact him if he became an NFL backup. Under pressure, his mechanics fail, and too often. He doesn’t step into throws, and the ball takes too long to get to its target. He might be a short, more athletic David Fales.
Those negatives sound very bad—and make no mistake, they can kill quarterbacks—but they didn’t present themselves as strongly in the prior preseason game as they did when watching him at Old Dominion, as Mark Schofield at Inside the Pylon demonstrates.
A comparison to Ponder isn’t a high note to start off with, but consider that Heinicke consistently—maybe unerringly—made the correct decisions with the ball in his brief NFL play, I think the two players are already continents apart.
Maybe it’s the confidence I have in an unknown quarterback and the false confidence NFL observers in general project to undrafted free agents—who by rule are longshots—but I think the early indications of his play as well as the Vikings confidence in him give us reason to think he could be the quarterback people are clamoring to trade for anyway.
There are a brief list of other quarterbacks to trade for.
It has been suggested that the Vikings trade for Philip Rivers, who has all of his contract guaranteed this year, and three more years after this of contract—with some of the next year guaranteed.
Rivers will take up $18 million of cap space in 2017.
The Vikings could do nothing but pay for the next four years of the contract in a trade. But they wouldn’t use those four years because it’s not as if they would bench either Bridgewater or Rivers. It’s a non-starter that would deprive the Vikings of a first-round pick or more.
A first-round pick isn’t an abstract high-value concept. It’s something that translates into players like Anthony Barr, Teddy Bridgewater or Laquon Treadwell. What Vikings fan would give up a decent shot at another Anthony Barr in order to get one year of Philip Rivers while sucking up cap space that could be used to fill holes in free agency in 2017.
Similarly, there are suggestions to trade for Sam Bradford. Like Rivers, he’s a starter. His contract is shorter, but his cap hit in 2017 is $22.5 million. While the Vikings could theoretically cut him before then, he would still cost $9.5 million in cap space.
Also, again, the Vikings would be paying to acquire a starter, while only using him for part of one season. As much as people do not like Bradford, the Eagles view him as a starting-capable player and would require as much in order to pry him from the roster.
There are some young backups in the league that would not be bad to have on the roster. A.J. McCarron is not a very good quarterback, but he’s also not the abysmal kind of quarterback that compose the range of options otherwise available on the free market.
The Bengals are well aware of the value of a decent backup, and played McCarron in a playoff game. Being reminded that quarterbacks can go at any time is a pretty good time to hold on to your valuable player more, not let him slip from their grasp.
Mark Sanchez has improved dramatically since his time in New York, as much as Philadelphia fans disagree with me, and he showcased quicker decisionmaking, improved accuracy and better pocket awareness there than he did under Rex Ryan. To keep referencing QBR, it was a Shaun Hill-esque 38.8 last year and a shocking 59.4 the year before.
The Broncos are trying to deal Mark Sanchez, and the Vikings are reportedly interested. I have no idea what compensation is being discussed, but it is noteworthy that Sanchez only has a one-year contract with the Broncos and though the cap number is unpalatable for Minnesota, it is also not unreasonable for an almost-starter ($4.5 million).
I suspect the projected compensation for a one-year rental (and again, you would wait a few weeks before he’d be viable) would be too much. Anything beyond a mid-round pick seems like an immediate “no” for my money, but I’m not opposed to acquiring Sanchez in general.
It is possible they could pick him up after cuts. I would not be opposed to that.
Aaron Murray was someone who caught my eye, but after reading the scouting report I put together on him a few years ago, I’m not too interested in him as a backup if he fails to make a squad where the third quarterback is either Tyler Bray or Kevin Hogan (or nobody).
Geno Smith has been mentioned by a few people. He’s on a roster where the primary quarterback has been settled (Ryan Fitzpatrick), where a backup seems to have emerged (Bryce Petty) and a recent high draft pick was invested on a third (Christian Hackenberg). He likely wouldn’t cost much, and his general traits seem to fit.
He’s easy to dislike, but the decision is between unworkably bad (Michael Vick) and pretty bad (Shaun Hill) and I think Geno Smith is only pretty bad. And by that, I mean he’s in the category of passer that could develop further in a different environment and displays the fundamental traits that fit a Norv Turner offense.
Smith has effective arm strength and a solid understanding of timing concepts. His accuracy was overrated coming out of West Virginia—it wasn’t very good in the NFL—but it has improved. He’ll still suffer from decisionmaking errors, and takes too long to make decisions, but that’s the reason he’s available at all.
In the universe of available quarterbacks, he can accomplish some of the features of the offense and can take care of the ball a bit better than his raw numbers indicate (a career interception rate above 4% is very worrisome), but it remains an issue. If one wants to use ESPN QBR, his 2013 and 2014 QBRs were 38.6 and 44.3 respectively, so in Shaun Hill territory.
I’m not opposed to limited compensation to grab Geno Smith with the understanding that he’d be a backup to Hill. But he’s not a savior, and I’m not sure he can develop into a player that will display consistent starting qualities. But in a few weeks, he may be able to provide the quality Shaun Hill does in a very different package.
A few people think Mike Glennon could help. He had a good touchdown-interception ratio as a rookie, but there are two problems.
The first is that his price would be pretty high. The Buccaneers view him as a valuable backup quarterback and have resisted several trade offers in the past.
The second problem is that he’s bad. Adjusted net yards per attempt is designed to give big bonuses for touchdowns and bigger penalties for interceptions. What does it say about a quarterback who threw 19 touchdowns to only 9 interceptions that his ANYA was close to Chad Henne, Christian Ponder and E.J. Manuel?
His QBR is better—48.0 his rookie year and even better the following year at 53.6, and his numbers are better than a lot of quarterbacks’ numbers that are being bandied around, but he doesn’t have the intrinsic qualities to be a quality backup to me. He takes too long to make decisions, is not often correct on those decisions, and has spotty accuracy. Glennon does not possess the mobility or the arm strength to make up for those traits.
I’m running against my own standard on this to make a qualitative judgment over a quantitative one, but I don’t think he fits the bill.
What’s Next
Not every problem has a solution.
But it might be possible to cobble together the best available offense given the constraints. Naturally, the offense will involve much more Adrian Peterson than originally planned.
Beyond that, it’s exciting to think about potential tinkers or even wholesale changes to the offense. Tony Sparano installed the Wildcat in Miami before the 2008 season, and it was devastating. There doesn’t seem to be a better candidate for a Wildcat offense than Jerick McKinnon, an option quarterback from Georgia Southern.
It’s not a bad idea to incorporate those elements, but the Vikings cannot install a new offense immediately before the season. Most of the plays will have to be standard dropback passes from a quarterback, especially if the team finds itself down. If lucky, a team can survive on two gimmick plays in the game, 25 runs from a running back and 15 passes.
That seems to be an extreme case, however. Generally, the Vikings will definitely choose to run the ball more often, and I’m not sure that Peterson will be able to respond to that the way he did in 2012.
Yes, option plays with McKinnon, Peterson and Patterson are incredibly tantalizing, but without an offensive system in place, they can’t generate a lot of value. Without the ability to counter off of common defensive responses to the plays, the ability to create similar plays from different offensive looks without ruining footwork, or a set of kill/audible responses at the line of scrimmage if the defensive look is bad, the offense dies.
Offenses are more than playbooks: they are a complete set of plans to respond to every available contingency on the football field, accounting for opposing personnel, down and distance and a bevy of defensive looks. Three plays is not a playbook, and a playbook is not an offense.
Sure, Colin Kaepernick torched the Packers on a small package of read-option plays, but it doesn’t sustain itself for a season without an entire system in place, like RGIII his rookie year or the Wildcat in 2008.
They should compensate for their loss in quarterback quality by running the ball more and implementing some unconventional strategies, and that will improve their odds in any gvien game. But they will still need to succeed at all the traditional things a team needs to succeed at in order to win.
So, can the Vikings win more than a few games in this scenario?
Bill Barnwell at ESPN seems to think so.
He functionally presents four arguments: 1) The Vikings defense can take the next step. 2) They’re a run-heavy offense. 3) The drop-off to Hill from Bridgewater is not large. 4) Hill may be a better fit for Norv Turner’s system.
In order to evaluate the first argument, I looked at the most successful teams who had a mediocre offense and a very good defense. Since 2000, they were the 2000 Baltimore Ravens, 2002 Tampa Bay Buccaneers, 2003 Carolina Panthers, 2003 New England Patriots, 2006 Chicago Bears, 2008 Pittsburgh Steelers and 2015 Denver Broncos.
Those teams all appeared in the Super Bowl, and one thing you’ll note is that none of those teams were competitive after that without significantly improving the offense (namely, the Patriots and for a short time the Bears). Another thing to note is that the Broncos are the only genuinely recent version.
All but one of those teams (2003 Patriots) had a top ten unit in the kicking game, punting game or return game. Four of those seven teams had a top five unit in one or more of those categories and three of them had the top kicking unit.
Essentially, an elite defense needs some help or extraordinary luck to overcome a bad offense and succeed. Moreover, three of those defenses weren’t just good, they were historically great.
While I know the Vikings have a top-tier return unit that is probably the best in the NFL, they are the opposite when it comes to punting and a bit of a wash when it comes to kicking.
Moreover, none of those teams had a bad offense. The worst was the Broncos, ranked 25th in Football Outsiders’ DVOA. Then there’s 22nd-ranked Baltimore, 21st-ranked Pittsburgh, 20th-ranked Tampa Bay, 18th-ranked Carolina and 14th-ranked New England. The Vikings, by going to Shaun Hill (or a worse quarterback) threaten to be one of the bottom five offenses in the NFL, even with Adrian Peterson.
That is incidentally Barnwell’s second argument. But it’s an argument he doesn’t really believe. Not only was the running game not determinative at all when looking at those seven teams with good-to-great defenses, but he has argued several times that passing is much more important than running the ball even when you run the ball more often.
The running game will also be impacted by Bridgewater’s absence. Inside the Pylon, this time at the Washington Post, detailed how defenses will respond (once again) to a powerful running back with no passing game to worry about. As you’d expect, more eight-man boxes, players at the line of scrimmage, and so forth.
The difference this time is that it is incredibly unlikely that Peterson turns this into a record-breaking season. More likely, he won’t hit the 4.5 yards a carry he averaged last season, even with a marginally improved offensive line.
We’ve already addressed Barnwell’s argument about the dropoff from Bridgewater to Hill, but the final argument about system fit is worth noting.
His argument makes two arguments to create its setup. The first is that Turner has been flexible and adapted the offense to feature shorter passes to compensate for Bridgewater’s weaknesses. The second is that he addressed the offensive line issue by not just running the ball more, but using misdirection in the form of read-option passing.
The second argument is less important, but I still feel compelled to point out how wrong it is. His premise is that the shotgun-style mesh gives the Vikings the opportunity to have Bridgewater read a defender and either pull the ball out and pass it, or give it to Peterson.
But given that Peterson took very few runs out of shotgun, it’s clear the RPO was not a common deployment for the Vikings.
The second argument relies on a common misnomer for Norv Turner. Barnwell speaks of how often Rivers would sling it deep in Norv’s system.
The last three years of the Turner offense with Philip Rivers saw him throw the ball over 25 yards (Teddy’s weakness) 6.5 percent of the time. Bridgewater threw it deep 5.8 percent of the time. That’s three passes per year. Rivers actually threw it intermediate (15-25 yards) as often as Bridgewater: 10.5% to 10.2% of all pass attempts.
Turner has always been an intermediate schemer.
Barnwell also uses Shaun Hill’s career deep passing rate to further his point. If he was correct about deep passing, it still would not be a great argument, as it is not useful to compare a 36-year-old Hill and a 28-year-old Hill. Without a large sample size of recent passes, the best we can do is reference Hill’s 27.3 percent completion rate on deep (25+ yard) passes with the Rams, which is lower than league average.
Moreover, the clear decline in Hill’s arm strength over the last two years makes it an even worse proposition.
Turner’s system requires accuracy, timing, pre- and post-snap reads and arm strength. Bridgewater has Hill beat on all four, especially because Hill’s throwing motion is inconsistent and changes his timing on a lot of throws.
The idea that “deep ball passing” can be the sole determinant of fit in a system when it only encompasses six percent of passes is stretching the truth at best.
And so
The Vikings will be worse off. They’ll lose more than one game they would have won. They have a talented roster, and the defense will provide some wins on its own—the Vikings aren’t at either end of the extreme.
At one end, there was the talent vacuum that was the 2011 Indianapolis Colts. They went from 10-6 (.625) to 2-14 (.125). That’s an 80% loss in quality. But they were driven solely by a quarterback and a defense that was thin but designed to attack offenses who were behind, notoriously weak against the run but acceptable against the pass.
The Vikings probably aren’t the 2008 New England Patriots, either. And make no mistake, that team was primed for more than it was; it went from 16-0 (1.000) to 11-5 (.688). That’s an over 30% loss in quality.
That Patriots team was stacked at nearly every position, with an incredible receiving corps, a fantastic offensive line, an underrated run defense and a decent pass defense, allowing the eighth-fewest points in the NFL. They also had the fifth-easiest schedule in the NFL, per Football Outsiders.
If the Vikings are somewhere in between because of a very good but not great defense, an underwhelming offensive support system and a somewhat tough schedule, one could estimate a five-win team (a 55% loss in quality). I think the prognosis is not nearly as bad as that, but the reality is such that even a team with a stout defense can lose a lot of games.
They can stem the damage with offensive tweaks and creativity, but probably not with new options at quarterback.
Having a bad quarterback doesn’t mean there aren’t a dozen worse options out there, and until we get to the 53-man cut, we won’t see a QB close to the caliber of the poor QB they have now.
Even then, bringing in a quarterback means they’ll have to learn a system and won’t be available for a few weeks—more time than Taylor Heinicke will be unavailable. Heinicke looks good and may be viable as an unknown in a way that all of these known quantities do not.
For now, the Vikings make small changes and wait.
Not every problem has a solution.
