Another great article on Russell... high QBR games

[Cartire]

I do completely agree that Football has way to many factors with 22 people on the field to really be accurately measured through SABR stats. However, they can still show a decent overall perpspective.

I do have to show you though how subjectiveness is not included in QBR as much as people think. You mention that the error on the books should be an error on QBR, but QBR only takes from the books so not the be subjective. You talk about how the fumble/touchdown was all fluke because of the positioning of the players at the time, but thats the only fair way to measure it. You have to just take the book results regardless of what it looks like, or else bias and subjectiveness becomes the more deciding factor.

That being said, i will reitterate that you are correct again. Football is has to many outside factors to incorporate SABR metrics accurately.

I'm probably being a little unfair on the bookkeeping point. I can't help myself though. QBR deserves it.

You misunderstand my point was on the Wilson fumble though. I am all for tracking fumbles and holding them against QBs, since some QBs are significantly more fumble prone and it's important to know that. What I don't think is smart is penalizing the QB for what happens after the fumble. Please explain how it is Wilson's fault that the ball just happened to bounce right to Wilkerson who was at the right place at the right time heading the right direction?

It makes about as much sense as penalizing David Akers because his blocked kick happened to bounce right at a sprinting Richard Sherman. But of course we'd never count something like that against a kicker, because doing so would be silly. I think it all points to QBR being a stat that tries to do too much with one stat. It would probably be a better stat if they broke it down into pieces instead. Make it four stats instead- maybe a stat for turnovers, a stat for clutchness, a stat for effeciency, and a stat for win probability.

I'm an abstract thinker, and I don't sense that you are, nor were the people that made QBR. What matters to non-abstract thinkers is logic and process. What matters to me is the end result. History is littered with failed inventions that were brilliant on paper but dysfunctional in practice. Though I think the process passer rating uses is arbitrary and flawed, I generally prefer it to QBR because passer rating more closely resembles the eyeball test of the two.

I can agree with the statement about the 6 points not being wilsons fault. But the metric is based on how well he helped a team win. And it measure every single play. So if the average play doesnt result in 6 points, then its not as detrimental to the win/loss. If it does, then it is. So him losing the ball and them scoring 6 made the play effect the win/loss more then just a fumble loss.

Is it fair, I guess its not, but that does becomes subjective. So to keep the stat pure (as can be), you have to account the fact that play was worse then a normal fumble, no-score.

Quick Edit: who the hell cares about SABR stats in relation to kickers ... lol ...just joshin you.

and btw, QBR includes clutch factors.

 

[RolandDeschain]

QBR also significantly overvalues running, and depending on how you look at it, over-penalizes turnovers or gives too many points for having none. Actually, Colin Kaepernick is the perfect example of the biggest flaw of QBR. I wanted to see how Kaepernick compared to Wilson, and if this list started at 3 games of 90+ QBR instead of 4, Kaepernick would have been on it, and his 3 for 11 earns him a 27.27% rating which would put him 4th on the list, immediately behind Aaron Rodgers and ahead of Tom Brady.

At this point, I became...Concerned. Kaepernick hadn't passed the eyeball test to that extent that I had seen, so I looked at the games he broke 90 QBR in, and found the problem. Here are the basic stats for his three 90+ QBR games of his career to date for games he was the starter in:

Vs. Chi, 97.5 QBR. 16 of 23 completed for 243 yards earning 69.6% completion rating and a long of 57 yards with 2 TDs, 0 ints. A nice day, but certainly not spectacular. He rushed 4 times for 10 yards and 0 TDs.

Vs. Green Bay, (divisional round of playoffs) 94.7 QBR. 17 of 31 completed for 263 yards earning a 54.8% completion rating, with a long of 45, 2 TDs, and 1 int. FAR from a spectacular day as a passer. I'd say 8 times out of 10, that earns you a loss. However, he ran 16 times for 181 yards and 2 TDs. This was the "Green Bay's defense decided to block for Kaepernick instead of against him" game, where they literally opened up holes the size of CANHawk's mom's lady bits for him to run through. Every Packers fan on the planet was embarrassed that day, before they even lost the game.

@ Atlanta, (NFCCG) 92.6 QBR. He was pretty efficient in this one, 16 of 21 for 233 yards earning a 76.2% completion rating with a long of 33 and 1 TD, 0 ints. Seriously though, you're called nearly perfect for a game for completing 16 passes and throwing for 1 TD? Another example of "9 times out of 10, you will lose a game with those passing stats". H rushed twice for 21 yards and 0 TDs in this one.

Anybody impressed with these "amazing" games that earned Kaepernick a 90+ QBR yet? Yeah, me neither. Don't get me wrong, his team won these three games, but we're discussing the QBR stat and it's obvious that it has some serious issues. It almost seems like some important honcho at ESPN has a hard-on for QBs that can run and sought to make last year's crop of rookie QBs look better than they really were, or something.

Now, let's look at Wilson's 5 90+ QBR games.

Vs. New England, 91.4 QBR. 16 of 27 for 293 yards, 59.3% completion, long of 51, 3 TDs and 0 ints. 5 rushes for 17 yards, 0 TDs. Now, the completion rating was only average, but 293 yards and 3 TDs with no ints is a pretty good day. Not spectacular. Probably not worth a 90+ QBR based on what ESPN claims a 90+ QBR performance should be, but pretty good. Arguably better than any of Kaepernick's three, or perhaps better than 2 and call the GB one a wash; but for strictly the passer stats, it's better than any of Kaepernick's 3 for sure. Kaepernick only had more than 2 passing TDs in one game the whole year, playoffs included, and that was @ NE.

@ Detroit, 93.7 QBR. 25 of 35 completed for 236 yards, 71.4% completion, 2 TDs, 1 int. 1 rush for 9 yards. Honestly, it seems like maybe ints isn't the issue with QBR, it seems to overvalue efficiency. The average pass was 6.74 yards, that is bad; but the completion rating was high. This is not a 90+ QBR passing game by ESPN's definition, in my opinion. A nice day, I'd say just shy of a "good" one, but certainly not stellar as the QBR would indicate.

@ Miami, 90.7 QBR. 21 of 27 for 224 yards, 77.8% completion, 2 TDs, 0 ints. 5 rushes for 38 yards and 0 TDs. Ok, completion percentage is ranked very highly in QBR, there's no doubt about that. It's the only thing that can account for a 90+ QBR when you have 2 TDs and 224 yards with no ints. A good day, but not great.

@ Buffalo, 99.3 QBR. Holy smokes. 14 of 23 for 205 yards with a 60.9% completion rating, 1 TD, 0 ints. 9 rushes for 92 yards and 3 TDs in the first half, though. Back to having more evidence that running (and completion %) are overvalued in QBR.

Vs. San Francisco, 97.6 QBR. 15 of 21 for 171 yards with a 71.4% completion and 4 TDs to 1 int. 4 passing TDs with only 171 yards passing on 15 completions is pretty damned good. 6 rushes for 29 yards, 0 TDs.

I'd say 4 of these games are better than all of Kaepernick's, except 1 of those 4 is not better than Kaepernick's Green Bay game; but looking at passing stats between the two, it's clear that Wilson earned more QBR points as a passer than Kaepernick did.

I'll be curious to see how this next season shakes out.

 

[kearly]

I can agree with the statement about the 6 points not being wilsons fault. But the metric is based on how well he helped a team win. And it measure every single play. So if the average play doesnt result in 6 points, then its not as detrimental to the win/loss. If it does, then it is. So him losing the ball and them scoring 6 made the play effect the win/loss more then just a fumble loss.

Is it fair, I guess its not, but that does becomes subjective. So to keep the stat pure (as can be), you have to account the fact that play was worse then a normal fumble, no-score.

Quick Edit: who the hell cares about SABR stats in relation to kickers ... lol ...just joshin you.

and btw, QBR includes clutch factors.

To wrap this up, I'm going to go back to a point that might have been brushed over. It matters not how smart a stat is, it matters how well that stat actually reflects reality.

If we had all the time in the world, motivation, training, and access to tape, the best way to evaluate is to simply watch the games with an educated eye and draw conclusions from that. Of course, that's a ton of work, and unless you are actually employed by an NFL team or otherwise have absolutely no life, that's not a viable option for the vast majority of us. Enter stats. Stats tell us in a single piece of information something incredibly complex with multiple parts. This is great because it saves a ton of time and is accessible to anyone willing to look the stats up.

Of course, during this information compression process you will invariably lose details and important pieces of information. This is why even the best stats can have some problems.

Language is the same idea. Ever have an idea in your head that you found incredibly difficult to put into words? That's because language takes little tangents of an idea instead of the whole thing all at once. If you saw an amazing painting, you would describe the colors, the subject, the style, but you'd still only be presenting the idea in peices and forcing the listener to use their imagination to compile those pieces into a new whole.

With passer rating, it does a really nice job of being a "quick and dirty" performance synopsis, but it only tells you some things while leaving certain details out. It's possible to have a 100 passer rating in a game but struggle in the clutch, for example. At the very least though, you know that the QB played very well on the whole, even if he had some imperfections.

I think QBR is an attempt to paint a more complete picture, but in the end it still suffers from the impossibility of describing a performance with data, while having (IMO) major issues with how it weighs its components.

Not the least of which is win probability, which I think very little of as an individual stat. Something you hear all the time from the brightest coaches in the league- "you pass to score, you run to win." This idea is backed up by the recent success of Seattle and SF, and also from the fact that 100 yard rushers are far more likely to win than 300 yard passers. But did you know that according to EPA (win probability), the average NFL rush attempt is worth negative win probability?

That might make some sense if you think in terms of just yardage per play, or the fact that teams that make big comebacks for huge WP gains typically do so with passing, and on the flipside a lot of running is done by teams who are already sitting at high WP numbers (holding big 4th quarter leads) and gain very little WP even from successful runs. But everyone, probably even statisticians, know that running the football in general has a very high correlating factor to winning, which you could practically write a book explaining the various reasons. So for that reason and many others, don't buy into WP as a stat for serious analysis. It should be for entertainment value only, like knowing your odds of winning at any given moment in a game, but not a tool for evaluating individual players.

Anyway, back to the original point. The point of a stat is to be a shortcut- an accurate portrayal of reality for those that didn't watch the games. If a QB goes out there and gives the performance of a lifetime (Wilson vs. Falcons), but QBR gives him a 60 (his score from that game), then QBR is wrong. Simple as that. Too many times last year I watched a QB give a performance for which QBR was wildly at odds with the eyeball test. And that makes it a garbage stat, even if it was made with good intentions by bright people. Maybe changes will be made in future seasons and it will be the best QB stat with time, but it's way off far too often for me to take it seriously as it is currently weighted.

 

[SE174]

So Seahawks.net summed up recently...

Article questioning RW = OMG SO FULL OF BULL CRAP!!!
Article praising RW = Best article ever!

One was backed up with hypotheticals and conjecture, the other on stats. :thcoffee:

 

[kearly]

a bunch of stuff

It's actually the inclusion of rush stats that makes me badly want to like QBR, but I just can't. Currently, DVOA and DYAR are the only stats that factor rushing into QB performance at all, which I feel is absolutely important to include and is one of the few good arguments in favor of QBR. The threat to run for 30 at any time without needing to commit a player as a safety valve is like having a 12th player on offense. It's soooooooo good for a QB to have that ability and I think within a generation it could very well become a requirement in the NFL.

I do 100000% agree though that QBR overvalues Kaepernick, though maybe not for the games you listed. He flat out embarrassed Green Bay and his rate stats were awesome in pretty much all the games you cited. That said, Kaepernick is the kind of QB that any stat would struggle with, because he's succeeding despite flaws that humans can detect but algorithms cannot. Passer rating and rate stats like Kaepernick a lot too, though QBR seems to like him the very most. That said though, until humans other than Pete Carroll actually stop Kaepernick with their defenses, it might not matter if QBR is wrong, because flaws and all, Kaepernick might still shred NFL defenses.

 

[RolandDeschain]

Football Outsiders comes the closest. Kaepernick was ranked the 13th best QB by them, Wilson the 8th. I'm just waiting for their 2013 guide to come out before I buy a year's subscription. Also, I lol'd at you (of all people) putting "a bunch of stuff" in my quote box. (I don't mean that in a denigrating way, it's just funny.)

 

[HawKnPeppa]

So Seahawks.net summed up recently...

Article questioning RW = OMG SO FULL OF BULL CRAP!!!
Article praising RW = Best article ever!

OK, so the author of the article that questioned Russell Wilson admitted that he didn't watch any of the games where Russell was a difference maker, and actually carried the team, yet he argues that you can't put the game in his hands. Did you actually read the article, and are you in up-to-speed with your own team, or do you prefer to oversimplify things? We all know which option requires less effort.

 

[RolandDeschain]

Lol, HawKnPeppa owned HighlandHawk there. Though, HighlandHawk's post is certainly valid for some people...I'm personally starting to get rather tired of "OMG, WRITER X IS ALWAYS HORRIBLE AND A TOTAL MORON, CAN YOU BELIEVE HE SAID XYZ?" when every single writer people say that about has other things praising the Seahawks in other ways.