Game Managers vs Gunslingers

[Super Nomario]

Interesting post over at FootballPerspective about whether analysts tend to overvalue interceptions:
 
Some gems from it:

 
The Game Managers list is populated by many of the worst modern passers, including notorious busts such as Gabbert, Russell, Harrington, and Boller, plus a host of rookies and washed up veterans. One can certainly go too far when it comes to sacrificing yardage as a way to avoid interceptions.
...
It appears that quarterbacks can get away with high interception rates as long as they move the ball efficiently. Why are Gunslingers so successful? And why are there so many Game Manager types in the modern game, when it doesn’t usually lead to winning?
...
... [T]ake a closer look at Aaron Rodgers. Despite owning the lowest interception rate in NFL history, he has a middling record in close games, and a downright terrible record when coming from behind. Why? He’s not taking the risks necessary to optimize his chances of winning. Even when trailing, which calls for a more aggressive strategy, Rodgers will usually take a sack rather than force a throw downfield. Avoiding interceptions keeps his stats looking pretty, but he has almost certainly left several wins on the table in the process.
...
Consider the unlikely Super Bowl runs over the past decade; every one of those teams had a Gunslinger type QB (Joe Flacco in 2012, Eli Manning in 2011 and 2007, Kurt Warner in 2008, Jake Delhomme in 2003). While all of them look terrible when things go wrong, their high risk styles give their teams a chance even when they’re overmatched.

 
538 has done a few pieces on the subject:
http://fivethirtyeight.com/features/skeptical-football-manning-vs-messi-and-andrew-luck-experiment/
http://fivethirtyeight.com/features/skeptical-football-the-aaron-rodgers-enigma/
 
To be clear: it's not that interceptions are good, it's that if you're not throwing any interceptions, you're probably being overly cautious with the ball. On the other hand, Brady (one of the best ever at avoiding INTs) has won three Super Bowls, and Favre (despite being the best QB of his generation) only won one. I do think in terms of averageish QBs, you're much better off with a gunslinger - it's easier for Flacco or Eli to get hot and not throw picks for three games than it is for Alex Smith to get hot and make plays for three games, even though over the long haul they're probably fairly similar in terms of value.
 
With respect to Rodgers, I think some of the issue is that Green Bay isn't willing to just have him throw down after down even when it makes sense (either based on the opponent or based on the game situation). He threw just 26 times in a playoff loss to San Francisco last year. He's never thrown more than 48 times in a game; Brady's done that four times just this year, including last week. I'm not sure whether that reflects on Rodgers at all or if it's just McCarthy.

 

[Toe Nash]

How many times did Rodgers have the ball in a good position because of defense or special teams vs Manning or anyone else? That's a fairly large part of the equation that is left out when talking about his not being able to come back.

 

[Silverdude2167]

Looking at the analysis, I would take the QB's they call 'gunslingers' over the "game managers" any day of the week. Those are distinctly different classes of QB's (the players not the terms) and not looking at any of the top QB's in the league. The argument that Rodgers should take more risks is not supported by fact.
 
Further that is 5 out of 24 QB's and only 3 winners that were "gunslingers" if we go back to 2003 as they do, so why should you want a gunslinger qb?
 
Flacco may have won a Superbowl as a "gunslinger", but a strong case can be made that he lost the Pats game last week with his second interception. That was the worst possible play he could have made at that moment in the game.

 

[BigJimEd]

And I'm not sure I'd put Flacco in the gun slinger category. He threw a ton of int in 2013 (and the team ended 8-8) but throughout his career that hasn't necessarily been the case.
In 2012, he had one of the lowest int% in the league as well as total amount of interceptions.

 

[behindthepen]

I thought you were going to be posting today's 538 post:  http://fivethirtyeight.com/features/the-10-types-of-nfl-quarterbacks/
 
It uses distribution of QBR's by game to find similar quarterbacks, instead of looking at averages.  
 
THey had this to say about post-2006 Favre (whom they categorize as a "Roller Coaster":
 

 
One of the most interesting groups involves a player who is no longer in the league. Favre’s games after 2006 landed in the same cohort as post-2006 Michael Vick and Vince Young, two quarterbacks known for making plays with their feet and not much else. We can explain this by breaking down Favre against Dalton. Although their average QBR is nearly identical, the vast majority of Dalton’s games fall between a QBR of 25 and 75. Favre, the ultimate freelancer, has a big bump around 15 QBR and another between 80 and 85. Dalton won’t win a team the game, but he probably won’t lose it. Favre, however, is likely to do either.
 
That difference results from playing style, according to John Westenhaver, president of Football Evaluations and a long-time quarterback talent evaluator. An average quarterback makes about half of his throws using nontraditional mechanics because he’s forced out of the pocket or rushed, but the former Green Bay Packers star made many more than that, often to his detriment. “Favre, to me, put that to the extreme. Although he’s passed for a gazillion yards, I think he leads the league in interceptions.7Sometimes you have to make a decision: Am I going to throw from this alternative platform, or is it best to select some other alternative, which may be to take the sack, throw the ball out of bounds, run the ball?”
 

 

[Morgan's Magic Snowplow]

On one level, I completely agree that the "Game Manager" type of quarterback is very flawed and unlikely to bring much success to an NFL team.  But I also think this guy has simply set up the statistical comparison between game managers and gunslingers in a deceptive way.  By selecting for QB seasons with greater than 110 NY/A+ in defining the Gunslingers, you're selecting for some great performances in terms of both accuracy and yardage pickup by definition.  Very few QBs are capable of doing that. Tom Brady, for example, has only cleared that bar in about half his seasons.  Add in the interception criteria and what you're basically picking up are QBs who had great seasons throwing the ball but also were on the wrong side of variance in terms of getting balls intercepted.  Conversely, by selecting for seasons for the Game Managers with less than 90 NY/A+ - which is hard to do unless you're really inaccurate and rarely throwing downfield - you're basically just selecting for very bad quarterbacking and then taking the subset of bad seasons in which the QB got fairly lucky with picks. The ultimate problem here is that you're picking up accuracy, not just downfield throwing, with NY/A+ and accuracy is a huge part of quarterbacking.  Mainly what that comparison tells me is that its better to be a good QB and get unlucky to throw more picks than usual than to be a shitty QB who gets fairly lucky in throwing less picks than might be expected.  Right.  But that doesn't tell me much about the tradeoff between risk taking and interceptions, conditional on being a fairly good, accurate QB.
 

 
I see a different, more encouraging corollary: Throwing interceptions is a byproduct of aggressive, optimal quarterbacking. The interceptions themselves are not good, but the willingness to risk throwing them is.

 
I also think that this point, which is the ultimate claim, is basically wrong.  Obviously seeking rewards as a QB entails accepting risk.  That's such a simplistic point that it doesn't really need to be made.  But the real question is whether or not that tradeoff needs to be linear and I would suggest that it does not.  The best QBs - optimal quarterbacking in the author's terms - are ones that identify matchups and game situations and then have the ball placement such that they can regularly make throws that are high reward and still relatively low risk.  That's why guys like Peyton (except his rookie year), Brady, Montana, Marino, Rodgers, and Young never show up on that gunslinger list.

 

[vintage'67]

Looking at this statement " Rodgers will usually take a sack rather than force a throw downfield. Avoiding interceptions keeps his stats looking pretty,"   I think Rogers takes too many sacks may be a fair criticism, but I think the author leaves out a major point. A QB's options are not always a high risk throw (with the resulting rise in interception rate); some of those he needs to throw away.  If he threw it away more, he could lower the sacks without a (significant) increase in interception rate.  Whether he is too conservative when trailing is a separate, interesting question. 

 

[wade boggs chicken dinner]

here's the post:  http://www.footballperspective.com/guest-post-are-interceptions-overrated/
 
From the comment section:
 
"The main thing that this post shows is that NY/A is a better measure of QB performance than INT%. Which is unsurprising. The high NY/A, lots of INTs group presumably also had higher ANY/A, QBR, DVOA, EPA, WPA, etc. than the group with low NY/A & few INTs."
 
This seems correct especially in this particular period of football.  I would be interested in knowing whether QBs who have a NY/A throw more damaging interceptions:  i.e., pick 6s into the flat, as opposed to deeply thrown interceptions that are more like punts.

 

[Kliq]

On one level, I completely agree that the "Game Manager" type of quarterback is very flawed and unlikely to bring much success to an NFL team.  But I also think this guy has simply set up the statistical comparison between game managers and gunslingers in a deceptive way.  By selecting for QB seasons with greater than 110 NY/A+ in defining the Gunslingers, you're selecting for some great performances in terms of both accuracy and yardage pickup by definition.  Very few QBs are capable of doing that. Tom Brady, for example, has only cleared that bar in about half his seasons.  Add in the interception criteria and what you're basically picking up are QBs who had great seasons throwing the ball but also were on the wrong side of variance in terms of getting balls intercepted.  Conversely, by selecting for seasons for the Game Managers with less than 90 NY/A+ - which is hard to do unless you're really inaccurate and rarely throwing downfield - you're basically just selecting for very bad quarterbacking and then taking the subset of bad seasons in which the QB got fairly lucky with picks. The ultimate problem here is that you're picking up accuracy, not just downfield throwing, with NY/A+ and accuracy is a huge part of quarterbacking.  Mainly what that comparison tells me is that its better to be a good QB and get unlucky to throw more picks than usual than to be a shitty QB who gets fairly lucky in throwing less picks than might be expected.  Right.  But that doesn't tell me much about the tradeoff between risk taking and interceptions, conditional on being a fairly good, accurate QB.
 

 
This is where I am at too. I'm not huge into statistics, but it seems like the author just took a look at the numbers but never really factored in playing style, which is what you normally think of when it comes to the terms "gunslinger" and "game manager." I watched JaMarcus Russell in College, he was most certainly not a game manager. He couldn't be considered a gunslinger in the pros by these metrics because he sucked. Same can be said for a guy like Sam Bradford, who led the NCAA in AY/A his sophomore year with 11.1. Obviously guys change their playing styles when they get out of their college system, but it is hard to ever consider a guy like JaMarcus a "game manager."

 

[ivanvamp]

The amazing thing about Brady is that, after his first few years, he could no longer be fairly considered a "game manager", given the number of times he threw every year.  When I think of "gunslinger" I think of a QB that takes a lot of risks.  Brady is in some ways the ultimate combination.  You can let him stand back there and throw 45+ times without really worrying that he's going to turn it over very much.  
 
He has thrown the ball 40+ times in his career 53 times during the regular season and 11 times in the playoffs.  Of those 64 times (that's 4 full seasons' worth of games), he's had:
 
- 0 INT 22 times
- 1 INT 27 times
- 2 INT 10 times
- 3 INT 3 times
- 4 INT 2 times
 
That's 64 interceptions in those 64 games.
 
Contrast this with Bret Favre.  Over his career he threw 40+ times 78 times during the regular season and 5 times in the playoffs.  Of those 83 times (more than 5 full seasons' worth of games), he's had:
 
- 0 INT 15 times
- 1 INT 25 times
- 2 INT 25 times
- 3 INT 15 times
- 4 INT 2 times 
- 6 INT 1 time
 
That's a total of 134 interceptions in just 83 games.  In 51.8% of his games throwing 40+ times, he had 2 or more interceptions.  In a whopping 21.7% of those games he had 3 or more interceptions.  
 
Brady, meanwhile, had 2 or more interceptions in games throwing 40+ times in just 23.4% of his games.  And in just 7.8% of those games he had 3 or more interceptions.
 
I know these numbers don't take into account the kinds of throws each guy made.  But when I think of "game manager" I typically think of throwing a limited number of times, not just the types of throws you make.  When I think of "gunslinger" I think of a guy who tries to fit the ball in small windows, whether that throw is 40 yards downfield or to a guy 7 yards downfield who is double-covered.
 
Brady throws the ball a TON, but when he does, he just doesn't tend to turn it over that much, compared to "gunslingers" like Favre.

 

[TheRooster]

Great post, brilliant stats that prove something many of us suspected.  Thank  you.

 

[Rustjive]

Brady is in some ways the ultimate combination.

It's a little funny to me that you bring up Brady here when the original post actually had the ultimate combination - Aaron Rodgers. This is the QB with the highest Y/A average (which accounts for 'type of throws') in the SB era and by far the highest Adjusted Y/A average in history.

I think the point that SN makes is correct - Rodgers simply doesn't pass that much. He's never been in the top 5 passing attempts in any year, in fact, he's only been in the top 10 in passing attempts once since 2010.

If you look at Rodgers' stats in close situations (down a score, < 5 minutes to play) since his career began he's actually been the best QB in the League. Whether GB needs be utilize him even more or be better as a team is up for debate, but his overall performance is tops at the moment (we'll see about Luck in the future): http://www.pro-football-reference.com/play-index/play_finder.cgi?request=1&match=summary_all&search=&player_id=&year_min=2008&year_max=2014&team_id=&opp_id=&game_type=&playoff_round=&game_num_min=0&game_num_max=99&week_num_min=0&week_num_max=99&quarter=4&quarter=5&tr_gtlt=lt&minutes=5&seconds=00&down=0&down=1&down=2&down=3&down=4&yds_to_go_min=&yds_to_go_max=&yg_gtlt=gt&yards=&is_first_down=-1&field_pos_min_field=team&field_pos_min=&field_pos_max_field=team&field_pos_max=&end_field_pos_min_field=team&end_field_pos_min=&end_field_pos_max_field=team&end_field_pos_max=&type=PASS&type=RUSH&is_turnover=-1&turnover_type=interception&turnover_type=fumble&is_scoring=-1&score_type=touchdown&score_type=field_goal&score_type=safety&is_sack=-1&include_kneels=-1&no_play=0&game_day_of_week=&game_location=&game_result=&margin_min=-7&margin_max=0&order_by=yards&rush_direction=LE&rush_direction=LT&rush_direction=LG&rush_direction=M&rush_direction=RG&rush_direction=RT&rush_direction=RE&pass_location=SL&pass_location=SM&pass_location=SR&pass_location=DL&pass_location=DM&pass_location=DR#quarterback::2

 

[Super Nomario]

On one level, I completely agree that the "Game Manager" type of quarterback is very flawed and unlikely to bring much success to an NFL team.  But I also think this guy has simply set up the statistical comparison between game managers and gunslingers in a deceptive way.  By selecting for QB seasons with greater than 110 NY/A+ in defining the Gunslingers, you're selecting for some great performances in terms of both accuracy and yardage pickup by definition.  Very few QBs are capable of doing that. Tom Brady, for example, has only cleared that bar in about half his seasons.  Add in the interception criteria and what you're basically picking up are QBs who had great seasons throwing the ball but also were on the wrong side of variance in terms of getting balls intercepted.  Conversely, by selecting for seasons for the Game Managers with less than 90 NY/A+ - which is hard to do unless you're really inaccurate and rarely throwing downfield - you're basically just selecting for very bad quarterbacking and then taking the subset of bad seasons in which the QB got fairly lucky with picks. The ultimate problem here is that you're picking up accuracy, not just downfield throwing, with NY/A+ and accuracy is a huge part of quarterbacking.  Mainly what that comparison tells me is that its better to be a good QB and get unlucky to throw more picks than usual than to be a shitty QB who gets fairly lucky in throwing less picks than might be expected.  Right.  But that doesn't tell me much about the tradeoff between risk taking and interceptions, conditional on being a fairly good, accurate QB.

I think all of this is right - because interceptions are a low-probability event, throwing few interceptions is not necessarily a marker of skill (nor is throwing many interceptions necessarily a marker of a flaw), while something like Y/A is indicative of skill. The interesting part is that even in the years where these low Y/A QBs lucked into low INT rates, they were still shitty - it's not like they had one fluky season where they were effective because defenders dropped INTs.
 

 
I also think that this point, which is the ultimate claim, is basically wrong.  Obviously seeking rewards as a QB entails accepting risk.  That's such a simplistic point that it doesn't really need to be made.  But the real question is whether or not that tradeoff needs to be linear and I would suggest that it does not.  The best QBs - optimal quarterbacking in the author's terms - are ones that identify matchups and game situations and then have the ball placement such that they can regularly make throws that are high reward and still relatively low risk.  That's why guys like Peyton (except his rookie year), Brady, Montana, Marino, Rodgers, and Young never show up on that gunslinger list.

The other point that he makes (and the 538 guy makes) is that some of the risk / reward is situation-dependent. If you're up 10, you shouldn't be trying to sling balls into tiny windows. If you're down 10, it's a different story. The guys you list are all effective enough that they're rarely down 10 - but when they are, we'd expect a higher INT rate.
 

It's a little funny to me that you bring up Brady here when the original post actually had the ultimate combination - Aaron Rodgers. This is the QB with the highest Y/A average (which accounts for 'type of throws') in the SB era and by far the highest Adjusted Y/A average in history.

I think the point that SN makes is correct - Rodgers simply doesn't pass that much. He's never been in the top 5 passing attempts in any year, in fact, he's only been in the top 10 in passing attempts once since 2010.

If you look at Rodgers' stats in close situations (down a score, < 5 minutes to play) since his career began he's actually been the best QB in the League. Whether GB needs be utilize him even more or be better as a team is up for debate, but his overall performance is tops at the moment (we'll see about Luck in the future)

This 538 article points out that Rodgers has great numbers losing by 4-8 points, but has never come back from a 9+ point deficit, which seems crazy. It might be a small sample fluke, I don't know.
 
There is something somewhat mystifying about Green Bay's performance over Rodgers' career. Rodgers has started five games the Packers have lost in four of his seven seasons; Brady only has three such seasons, and Manning just four despite much longer careers. The Packers have "only" led the league in scoring twice while Rodgers has been QB - obviously leading the league in scoring twice in seven seasons is awesome, but I'd think they'd do it every year given Rodgers' efficiency numbers. In 2012 he completed 67% of his passes for 7.8 YPA and a 39 TD / 8 INT ratio - and the Packers finished just 5th in points and 13th in yards, while losing five games. I don't get it.
 
I'm guessing this is more due to preference on McCarthy's part to have a balanced attack than any deficiency on Rodgers' part. I do wonder if it inflates Rodgers' efficiency numbers somewhat, because he doesn't have those games like Brady has last week where they ask him to throw 50 times and basically be the running game with short passes.

 

[ethangl]

This 538 article points out that Rodgers ... has never come back from a 9+ point deficit, which seems crazy. It might be a small sample fluke, I don't know.

 
Well, that isn't true.
 
2009 W15, GB scores 22 points in the 4Q to come back from a 10 point deficit. Pitt wins on last play of the game.
2009 WC, GB scores 21 points in the 4Q during a 21 point come back. Arizona wins in OT.
2014 W2, GB rallies down 18 against the Jets.
 
Humorously, Matt Flynn had 4 double-digit comebacks last year. He didn't throw any 4th quarter picks, so that guy probably thinks he could have been a little more aggressive.

 

[Super Nomario]

 
Well, that isn't true.
 
2009 W15, GB scores 22 points in the 4Q to come back from a 10 point deficit. Pitt wins on last play of the game.
2009 WC, GB scores 21 points in the 4Q during a 21 point come back. Arizona wins in OT.
2014 W2, GB rallies down 18 against the Jets.

You're right; the actual statistic was Rodgers had never won a game where he trailed by 9+ in the second half. The first two didn't end up as Ws and the game against the Jets was just 5 points at halftime.

 

[ethangl]

It is small sample size. Rodgers generally does not get "good" opportunities to lead multi-score comebacks.
 
Since 2009, the Pats and Packers are 32nd and 31st, respectively, in number of 2nd half drives started down by 9 or more points. They are #1 (35%) and #2 (32%), respectively, in TD percentage on these drives, though. If the Packers had scored one just more TD in 5 seasons they'd be #2 at 34%. 
 
Coincidentally 32% is also their TD rate when down 1 to 8 points, which is #1 in the league.
 
In the 4th Quarter, their TD rate is significantly higher than the rest of the league when trailing -- 37%. Next closest is 33% (Dallas).
 
I don't really see what Rodgers is supposed to do, besides lead touchdown drives, in these situations -- and he has proven to be very good at that.
 
This 538 article seems like a classic case of coming up with a narrative and then Vanning the numbers until something kinda makes sense. I mean, maybe this "defense" thing kind of has an impact on your ability to win close games. I dunno, it's just a theory of mine.

 

[soxfan121]

It is small sample size. 

 
Yup. Numbers just do not tell you enough because the context has so much variation.