Arif Hasan’s mid-season review (Quarterbacks)

EDIT: This is an outstanding job by Arif, so I'm moving it to the front page. - Chris

Well, we’re halfway through the season and onto our BYE week. I haven’t done this before, but I’m going to act like I have – I’ll go through every position on our roster (starting with the offense, in particular the quarterbacks) and evaluate our progress, based on what I see and on the analytics I trust. I like using a combination of advanced statistics and grading as well as what I can see, and I think that using both is really what’s necessary for effective analysis. Numbers can do a lot – they will, for example, capture the intangibles in a much more real way than people will believe, simply because the intangibles result in real results. Perhaps they will not assign intangibles to particular people on the field, but we, with our eyes, can. I think a combination of both builds an effective narrative that allows us to move forward with the best understanding of our team in mind. Fair warning, this is about 4000 words.

Note: I had a brainwave in the shower and now have to edit something near the bottom.

QB: Christian Ponder – Hot Young Prospect. Hailed as the most NFL-ready of the draft, Vikings fans weren’t able to see the readiness of this prospect as easily as many other fans of different teams were able to.  Foremost of his criticisms coming out of (and going into) the draft was his arm-strength. His greatest strength was his accuracy. How does he match up?

With a completion percentage of 51.9% over three games, one can imagine that there is significant room for improvement. Indeed, if one were to solely evaluate quarterbacks by their completion percentage, Christian Ponder would look absolutely abysmal (Tebow-esque? Not really, Tebow is at 46% with some garbage time help). His passes miss his receivers an astonishing 10% more than the league average. 6% of his passes are dropped, which is slightly higher than league average, but not by much. Compare this to Matt Moore, Charlie Whitehurst, or Sam Bradford, all of whom have 10% of their passes dropped, and Ponder seems to be gifted (certainly more than he was at FSU, if the limited tape I saw means anything). If you factor in dropped passes, spikes, and throwaways, Christian Ponder’s completion percentage is 55.5%, against a league average of 69%. That certainly isn’t encouraging. People with a lower adjusted completion percentage? Tim Tebow (52.94%), Carson Palmer (47.6%), and Bruce Gradkowski (41.67%). That’s it.

 This isn’t altogether discouraging, though. Well, it is a little bit, but not as much as you’d think. Rookie quarterbacks who go on to be successful vary in their completion percentage from league average in their rookie year. Rookie quarterbacks with similar completion percentages from league average include John Elway, Don Majkowski, and Kerry Collins.

 Better predictors of rookie success are average yards per attempt and touchdown percentage. They are not linearly predictive (that is, one cannot expect, generally speaking, for rookies who perform worse or better in these stats to necessarily perform proportionally worse or better as their career progresses), in two different ways. Rookies who average one yard less per attempt than league average will almost always do poorly in their career. Anything above that, and it is hard to tell. Rookies who average the same or better than league average do tend to do well, but it’s not as hard-and-fast as you think. Touchdown percentage (touchdowns per pass attempts) is similarly finicky – poor TD % is never predictive, but a good TD% is quite predictive. Don’t ask me why, I do not know.

 Christian has 7.19 yards per attempt on the season, and the league average is 7.2. Successful rookie quarterbacks with a differential similar to 0.1 YPA (which is very good for a rookie) include Drew Brees and Troy Aikman. For nearly every quarterback, it is more productive and predictive to use Adjusted Yards Per Attempt, where one subtracts 45 yards for every interception and all sack yardage from the pass yardage total before dividing from the number of attempts. The reason I say "nearly" every quarterback, is because interception percentage has nearly no bearing on future rookie performance. Nevertheless, it is easy to calculate and compare. The league average on AYPA is 5.6 and Christian Ponder sits right below it at 5.5. That is extremely encouraging. Only one first or second year QB is above Christian Ponder in this category (Cam Newton) and Sam Bradford, Josh Freeman, Colt McCoy, Andy Dalton, Tim Tebow, and Blaine Gabbert all sit below him. QBs with more experience that sit below him include Jay Cutler, both ‘Skins QBs, Tarvaris Jackson, Mark Sanchez, and our own Donovan McNabb.

Something astonishing pops up though. When playing with the numbers – subtracting dropped passes, throwaways (which are on the whole good plays), sack yardage, and adjusting for interceptions while totaling completions per attempt, Christian Ponder is 7th in the league. League average for what I’m going to call Adjusted Yards Per Adjusted Attempts (AYPAA) is 5.68, with jokers like Kyle Boller, Luke McCown, and Charlie Whitehurst hanging out near the bottom. As AYPAA doesn’t value completion percentage as much as the NFL Passer Rating (a system where completions for negative yards are positive plays), it doesn’t correlate well with the NFL Passer Rating’s system. Aaron Rodgers is still on top (more on that later), but instead of Tom Brady, Eli Manning is number 2. Tom Brady is behind Drew Brees at number 4. 9 of the top 10 QBs in Passer Rating have winning records, while 8 of the top 10 in AYPAA have winning records. Nevertheless, I think AYPAA is a better predictive measure than Passer Rating, which is probably better at evaluating the current or past performance of a QB (as it incorporates touchdowns). At any rate, Christian Ponder is producing minimizing losses and maximizing gains, beating out Cam Newton in this category by a third of a yard.

Christian Ponder’s TD % is 4.08%. League average is 4.25%. This compares extremely favorably to rookies who went on to do well, who are generally 0.5% behind the pack. This is less true recently, however.

 We saw 3 rookie QBs in 2008 who have a good sample size. Matt Ryan (-0.1%), Joe Flacco (-0.5%), and Chad Henne (-3.88%) – who threw 0 touchdown passes. Matt Stafford threw -0.7%, Mark Sanchez threw -0.85%, and Josh Freeman threw -0.7% compared to league average in 2009. In 2010, Sam Bradford threw -1.36%, Colt McCoy threw -1.71% (Ugh), John Skelton threw -2.82% (Oof), and Jimmy Clausen threw -3.41% (Oh my God, why). This certainly bodes well for our budding QB.

 And this last statistic I think captures a lot of what we like about Christian Ponder. He doesn’t seem to complete every pass, but he sure makes them count. Against Green Bay, 12 of 13 of his completed passes were for a first down or a score. Against Carolina, 14 of 18 were for a first down or a score. He completed 9 of 10 passes on 3rd down, and converted 7 of them. Some of us (mostly me, I think) referred to 3rd down as "Ponder down" because of his great ability to convert on 3rd and long. He’s converted 53% of his third downs, and the best team on third down in the league is converting 48.8%. That’s out of control.

 He has an awareness of the field, whether it’s the down marker or the pressure within the pocket. In fact, Christian Ponder has done something incredible, that we do not talk enough about. He ranks 5th in the league in sack percentage (sacks per dropback), behind the same line that Donovan McNabb accrued 16 sacks, or the 35th rank in the league (above such noteworthies as Tarvaris, Whitehurst, Moore, Blaine, Bradford, and Tebow). Not only that, Christian Ponder leads the league in having the lowest yards lost per sack, at 1.5. League average is 6.5. McNabb ranked 7th to last in that category too, with 7 yards lost per sack. Part of this is because McNabb ranks 12th lowest in throwaways per dropback, and Ponder ranks the absolute highest.

 So if you combined it all together, low completion percentage and high-value plays, what do you get? Well, something that seems a hell of a lot better than most QBs in the league right now. That is, a Win Probability Added of .11 per game. That is, Christian Ponder’s plays add an 11% likelihood, per game, to our wins. This is why his completion percentage matters less than his completion percentage on third down. Because he makes plays when it counts. The average QB in the league adds 7% per game to their team’s win. Donovan McNabb added the exact opposite of Ponder, with -11% added to the win, per game. That’s right – his play was so awful that our team had to win in spite of him. Is this predictive? I don’t know – in fact, I kind of doubt it. But because it seems to speak to an ability that Christian Ponder has across the board, I’m willing to believe that it is in this case.

Given that he has the second highest yards per completion (13.85) in the league (second to Aaron Rodgers, at 13.91 and eking out Schaub, at 13.84), one can probably conclude that arm-strength isn’t a problem. However, knowing his adjusted completion percentage is 14% below the league average gives us pause to question his accuracy. My, that seems odd. This should be enough for us to temper our enthusiasm, but perhaps not by much. Quarterbacks don't often improve their accuracy by more than 2 percentage points after their first 16 games, and I know that in order for Christian Ponder to be the QBotF, he needs to work on this. I know he has only started 2 games, but I'm grading off what is available to me right now, and that means we know that all of this can be subject to radical change (good or bad). Given his high yards per completion, his accuracy is not as much of a concern as it could be, but it does mean that fewer plays of his are successful and that standard has its own value (it’s not always about averages).

Grade: B+

I should add that a QB who we should expect to be average (based on experience and his supporting cast, not rep), who then performs average, would get a C. I would likely give this to Joe Flacco or Josh Freeman at this point in the season. Rodgers would get an A+, Newton an A, etc. Peyton Manning gets an in absentia A+ and Valedictorian status.

QB: Donovan McNabb – Veteran Stop-Gap. A Quarterback one year removed from a Pro Bowl season, brought in to both fulfill a Vikings tradition of being a last gasp train station for retirees and to maintain competitiveness in a year with QB questions. Well-known for his low interception-to-touchdown ratio (or high touchdown-to-interception ratio) and with reasonable expectations of some mobility, Donovan McNabb was seen as a safe option to bring the Vikings out of the cellar and back to respectability come 2011.

Well, Donovan McNabb has almost exactly league average 60.3% completion percentage (average is 60.4%) and an adjusted completion percentage of 66%, compared to league average of 69.1%. This will surprise some of you, given that McNabb has an affinity for overthrowing or underthrowing the intended receiver, but this is generally the bias we all have for remembering the atypical over the typical. Also, remember he technically participated in many of our first halves.

While outdoing Christian Ponder in completion percentage by a significant margin, Donovan McNabb ranked much lower in Yards Per Attempt, a fairly significant statistic for determining a quarterback’s likelihood of winning in the future. He gained 6.58 passing yards per passing attempt, beating Josh Freeman and Kyle Orton (as well as several others) in doing so. Under the league average (7.24) by two thirds of a yard, he was simply not performing up to league standards. If one were to adjust completions for interceptions and sacks (and dropped passes, throwaways, and spikes), Donovan performs much worse – despite generally avoiding the huge penalty assigned to interceptions. His poor decisionmaking on broken plays and even poorer scrambling has left Donovan McNabb with 5.6 adjusted yards per adjusted attempt, falling short of the league average of 6.1 by half a yard. The reason this is worse given the adjustment is because the variance around the mean is so small that performances 1.5 below or 1.5 above are so noteworthy – grounds for MVP candidacy or immediate removal from the NFL. With vanilla YPA, one can see that a variance of 2 is more likely meets that level of significance.

I’ve already written about how Donovan McNabb performs in "pressure" situations. His awareness of the field – pocket collapsing or first down markers (or where his receivers hands were) has been abysmal. Despite having an offensive line that has allowed the 6th most QB pressures this year (Sacks + Hits + Hurries), McNabb has the 7th least amount of throwaways. He also takes more yards per sack, ranking as the 8th highest yards lost per sack – beating the league average by a yard and a half. He ranks 8th in total sack yards lost per game, having half a sack fewer per game than everybody above him on the list.

Out of the 33 QBs listed in NFL.com’s stat page (which isn’t enough and does not include QBs like Christian Ponder, the McCowns, or Tim Tebow), McNabb ranks 19th in 1st down %, converting exactly 1/3rd of his passes for a first down. In fact, under McNabb, the Vikings converted 35% of their third downs. Under Ponder, we have converted 53% of them. If we maintained the 53% average from the beginning of the season, we would be ranked 3rd in the league, behind San Diego and New Orleans and ahead of Pittsburgh. If we continued at 35%, we would be 20th in the league, behind Cincinnati.

Donovan McNabb has hurt his team’s chances to win by 11% a game. That’s .67 games lost just from his performance. More importantly, he doesn’t contribute to winning, which is his job. When you compare to the mean (7%), Donovan has contributed to our games a net negative 18%. Per Game. That is, an average quarterback would increase our chances to win over Donovan McNabb by an astounding 20%. If we had an average quarterback (say, the pretty inconsistent, but always passable Joe Flacco), they would have "won us" 1.2 more games. Perspective: in 2009, Peyton Manning contributed 6 wins over the course of the season (remember, this is by performance alone, not playcalling, morale, leadership, or, apparently, coaching) and elite quarterbacks tended to contribute 4 wins. That is, Peyton contributed 4.8 wins over average and elite quarterbacks contributed 2.8 wins over average. In 2010, a much more historically typical year, Matt Ryan contributed 4.75 wins and elite quarterbacks contributed about 3.75 wins over the year, or Ryan led the average QB by 3.5 wins and elite QBs beat out average QBs by 2.5.

If this performance was consistent, Donovan McNabb would have lost us 3.2 games vs. an average quarterback. He is worse this year than elite or MVP quarterbacks are good in most years.

His front-level stats are fine. Completion percentage and yards per attempt weren’t good, but they weren’t obscene by any means. He protects the ball from turnovers and didn’t seem to perform badly given his role – at first glance (if that glance wasn’t while he was playing). If you dig a little deeper on paper or happen to watch the games, you can see that McNabb was simply abysmal. Do I think there are worse QBs playing this year? Oh yes. John Beck, Tim Tebow, and Kerry Collins were all worse in nearly every measurable way. But that might also not be fair. Kerry Collins technically retired, and Tebow’s a running back. Given that he could perform worse, and Kevin Kolb technically did perform worse, I won’t give McNabb the worst grade, even though these are curved towards expectation.

Grade: D-

Seriously Kolb, get it together.

 QB: Joe Webb – Misfit Wunderkind. Declared in the draft as a QB, but drafted to fulfill a WR role, Joe Webb hasn’t had a solid place on the team. Having played designed runs, returned kicks, and lining up in the slot, the Vikings are struggling to find a way to make this playmaker make plays.

This is a surprisingly easy write-up, considering that we haven’t seen him on the field as a QB yet. I’m just going to compare Joe Webb’s first 180 snaps to Christian Ponder’s first 150. That’s about a half a game difference. These happened in a different league year, yes, and with the addition of receivers like Sidney Rice, but it is certainly food for thought. The biggest difference here is the change in offensive coordinators. Both have been performing in "lost seasons" and both want to prove that they are franchise QBs. Each of them suffer the same WPA penalty of entering a game that’s already lost, so I won’t adjust for that.

It’s easy to start with something that both QBs are lauded for – extending plays and their athleticism. How good has Webb been in avoiding sacks? Joe Webb had better pass protection (13th worst in the league in pressures allowed… weird, I know) than Christian Ponder, but still accumulated 8 sacks, to be sacked 7.1% of the time he dropped back. Ponder has been sacked 6.7% with the 6th worst offensive line in pressures allowed. His yards lost per sack are below league average as well – 6 yards per sack.

Again, some context. The second lowest yards lost per sack this year is 3.8 and the next step up is 5 (which is still impressive and quite a bit lower than typical). It’s not that Joe Webb’s sacks are bad, it’s that Ponder’s are remarkably good (relatively speaking) Joe Webb holds on to the ball a little bit too long, especially compared to Ponder, but it doesn’t hurt him as much as it does most QBs in the league. That said, Joe Webb only threw the ball away once. This speaks to his awareness, I think – he’ll find a way out when there’s pressure, but he is merely average at detecting it.

Evaluating his performance by the metrics above, one can see that Joe Webb was a much more accurate passer (by 3%) than Christian Ponder was in his first 150 snaps, to match the 2010 league average at 61%. His TD% is unsurprisingly low at -4% compared to league average, if you include all touchdown rushes and run attempts (for all QBs). His YPA is -1.68 from league average, which is condemningly bad. He did make more than a couple plays with his feet, though – excitingly enough for 6.67 yards. How does he compare if you adjust his yards/completion for runs, run yards, dropped passes, throwaways, interceptions, etc.? Well, the league average for all of that in 2010 was 5.49. Christian Ponder’s was 6.46 (League average in 2011 is significantly higher in 2010, but I do not have all QB run data entered in at the moment – it seems that there are fewer QB runs this year per game, too – but not by much) for what probably beats league average by probably a good 1.1 yards.

Joe Webb’s was 3.8. The difference here is stark – I am willing to concede the argument that Musgrave’s offense was better than Bevell’s both in terms of playing to QB strengths and in terms of doing more to win, but being below league average by 1.1 yards is not good. I’m willing to bet that if Joe Webb had first team reps, and a solid offseason, and an offensive playbook tailored to his strengths (the long ball, sneaks, designed runs, etc.) he could be better than league average, but to me there’s no question that Christian Ponder is the better quarterback, especially in this offense.

Joe Webb has a pretty good completion percentage, and when factoring dropped passes, ranks at exactly league average, which is certainly impressive for someone who has had very few first team reps, was drafted to fulfill a WR role and was a 3rd string QB. Does this mean he should continue being a QB? Not necessarily, but I think having Joe Webb as a backup would be pretty fantastic. He’s already a better QB than some starters in the league, and I think the only backup QB that I can think of that is better than him is Vince Young. As a developmental QB, we could do far, far worse.

His strengths are accuracy, particularly on the long ball as well as his athleticism. His biggest weakness is his peripheral field awareness. He’s 12 for 30 on 3rd down conversions, but has more 1st downs per snap than Ponder, who is not as good on 1st and 2nd down as we’d like. Joe Webb converted 37.6% (That’s insane. Only 5 QBs – Rivers, Schaub, Rodgers, Brady, and Roethlisberger – were higher) of his snaps into first downs, Ponder - 31%. Does that mean we have Joe Webb play the first two downs and then sub Ponder in on 3rd down and in the Red Zone? Well, no that’s not how offenses work. But I think if Joe Webb knew when to run and where his receivers need to be to convert, he would become an incredible quarterback. His weaknesses are coachable and his strengths are not (it is extremely unlikely for QBs to shift their completion percentage much year-by-year).

He is hit or miss. Either his plays do exactly what they need to (high conversion percentage) or the do exactly what you don’t want them to do – and they’ll do the second more than the first (low AYPAA).

Who should start? Christian Ponder. Is Joe Webb a QB? Absolutely. Just give him time, because he can be coached to do what can be coached and he already knows how to do what can’t be coached.

Grade: C+

Addendum: Would Webb make a good WR? I don't know. He has the biggest hands of his draft class and an extraordinary vert. His 40 speed is great, but I don't know how long it takes to coach discipline on routes. Also, I'm of the firm belief that WRs should block, especially if we're going to run reverses, sweeps, and screens (we are) - can he do that? Probably not nearly as well as the other WRs we have on our roster.

EDIT: Ponder's 1st Down % has been edited - it didn't seem right, so I went through the play-by-play. Almost all of his negative 1st Down % comes from 20 plays run in Chicago, where he converted 25% of his plays. In games he's started, his 1st down% is extremely impressive - 44.7%. Quite possibly the best in the league. This certainly changes the discussion and gives us credibility to the idea that Ponder knows how to move the chains. Just... less often on 1st and 2nd down - his completion percentage on those downs is 39.5% and his completion percentage on 3rd down is an astonishing 70%, which drives his 1st %. Joe Webb's completion percentage on 1st downs is over 67%, and he gains 6.1 yards per carry on 1st down. I think the nature of the discussion stays the same - improvement on 1st and 2nd downs, especially versus his backup, but the numbers defining it are different.

Incidentally, I said I would say more about Aaron Rodgers here. Well, looking at these numbers really puts Aaron Rodgers into perspective – he has the highest completion percentage and highest yards per completion of any quarterback. He has the third lowest interception percentage, a better than average sack percentage and an average yards per sack. He leads the league in touchdown percentage. This means that in the normally clustered adjusted yards per adjusted attempt category, Aaron Rodgers leads the league by light years. He beats the second placed quarterback by 1.8 AYPAA in a year when beating just the average by that much would normally be impressive. He beats league average by 3.5 yards for a total of 9.26. To put that in context, Philip Rivers led the league last year with an AYPAA of 8.18, beating the league average by 1.52. In 2009, Philip Rivers was leading the league by beating the average by 1.9. It’s not that this is a career year or a HoF year for Aaron Rodgers. It is an historic year.

Next up is Wide Receivers. I won't get into Burton or Arceneaux, but I want to see if I can crack the code on Harvin, Aromashodu, Berrian, Camarillo, and Jenkins. Yes, I know Berrian's not on the team any more and I am as happy as you are about that.