2016 NFL MVP Race

[amarshal2]

I'm really confused as to why baseline matters for a single season award.

If next season some completely middle of the road QB (say Andy Dalton) suddenly goes for 48 TDs and the Bengals go 14-2, while Brady/Rodgers go 35/10 or something, is Dalton moved down your list because his baseline isn't as high? That doesn't make sense to me.

My argument is essentially that it's such a team game it's too difficult to really know who was the best based on stats in a single year.

The best, non-homer example I have for this is that I think Brady 2016 > Brady 2007. Brady was great in 2007 but Moss, Welker, Stallworth, Gardner Faulk, and the o-line were a ridiculous set of weapons. If we put Tom Brady 2007 stat line up against Tom Brady 2016 stat line (edit: pro rated for 16 games) voters would give the award to 2007 every single time. And I think they'd be wrong.

So where does this leave us? Everyone would need to turn themselves into film nuts and watch 16x16 games which is unrealistic. I half seriously suggested an approach where rather than look at a single year stats we look at the overwhelming evidence over the weight of their careers combined with their performance in a single season and figure out who was probably most valuable. It's a flawed suggestion that the public wouldn't take seriously but I think it would lead to more "right" outcomes than the current approach.

 

[coremiller]

There's definitely some exaggerated style on my behalf. I get that Ryan is going to win the award. I get that most statistical cases support him. I get that nobody would ever take my approach seriously and that it would result in Brady or Rodgers winning the award every year. That's way too boring for the public who likes the "best story" award and likes arguing about the nuances of stupid shit.

But whether he should win or not, there are a few points I've made I actually believe:
-These stats are directional and too much is made of them for these awards. You don't hear BB quoting DYAR because he understands how flawed it is and how it doesn't replace watching a lot of film (which admittedly, I'm not). I really like stats in general but have never been persuaded by any of the football ones enough to pay attention after 2 minutes. They always look bizarre when ranked ordinally.
-Brady looks as good as he's ever looked. That's not enough to win him the award, I get that, but I have a really hard time buying that Matt Ryan was actually better in isolation. We'll never know.
-I really don't care about the 4 misses games. I understand the math, I understand it makes me look like a homer, but those games are illegitimate in my eyes. So, fuck the NFL, I'm Just prorating the last 4 games in my head.

As I said, Matt Ryan will win. I won't care very much. If I had a vote I'd vote for Brady because of how well he played.

FWIW, while you don't hear BB quoting DYAR, I think Football Outsiders' work is taken seriously by many people around the league. Schatz is kind of mum about it, but every now and then he mentions that teams have had FO do consulting work. And I would be shocked if a number of teams haven't developed their own proprietary analytics based on an similar-type EPA/WPA model.

Here's an example: http://www.footballperspective.com/interview-with-aaron-schatz/#more-4653

 

[RetractableRoof]

Not when he missed 1/4 of the season, no it can't.



Why are we considering Brady missed 4 games? This is the entire argument - no-one here is arguing that if you were to pick any player to play one playoff game, or if we're going by who is more likely to be better next year, or if Brady had played a full season, Brady wouldn't have won easily. But in reality, he missed 1/4 of the season - and even if we don't 'punish' him (ala 'steroid users can't win any award'), he still has to be treated as any injured player that misses 1/4 of the season when you're considering an award for most value over the entire season.

The rest of your post is interesting, but it still does seem like a lot of stretching to ultimately claim that, at best, Brady is roughly in the ballpark of Ryan

Fair question. Earlier I read a claim that Ryan had had a historic season (as determined by flawed and incomplete stats - though they may be the best we have available). That he was leaps and bounds ahead of everyone around him this year. Ad nauseum. Since the flawed stat being used doesn't separate his supporting cast from his own performance (as stated by the creators) I don't think using the same stat to illustrate the enormous advantage Ryan started each play with is much stretching. It's a crude attempt to see if the supporting cast really was significantly better. By showing the cast really was SIGNIFICANTLY better in the context of THAT statistic I am more comfortable making my own assessment that the two QBs were closer in performance than those screaming 'numberzzz!11' want me to believe. Or condescendingly screaming that 'flawed statistics have to be the basis of any intelligent discussion' bullshit.

After my effort, my poorly stated conclusion was that Ryans offensive/production performance was probably a small amount better than Brady's and he did a quarter season more of it. I'm not sure that's an unreasonable view.

However, when I used those flawed statistics to confirm my instincts, it gets questioned as a stretch. Makes it hard to have a discussion when you're told to shut up except for flawed statistics and then judged as a stretch for using them (poorly?).

ETA: If Super Nomario or others want or are able to correct my math (or stat use) and show that I didn't really show what I was trying to show I'm willing to revisit my conclusion.

 

[joe dokes]

FWIW, while you don't hear BB quoting DYAR, I think Football Outsiders' work is taken seriously by many people around the league. Schatz is kind of mum about it, but every now and then he mentions that teams have had FO do consulting work. And I would be shocked if a number of teams haven't developed their own proprietary analytics based on an similar-type EPA/WPA model.

Here's an example:

That's interesting compared with, say, baseballl, where they play every day and there's no all-22 (its getting there with the tracking stuff), so teams are "stuck" with relying analytical data on players they dont see every day. With a week between games, and film galore, NFL coaches and personnel folks can rely on their eyes more. Schatz referred explicitly to things like 4th down and penalty-accepting models. But for actual play of individuals, I wonder if coaches are ever really surprised to learn at the end of the season that left guard x was in the bottom 10%.

 

[Devizier]

However, when I used those flawed statistics to confirm my instincts, it gets questioned as a stretch.

That's because it's motivated reasoning.

Put simply, Y/A and its derivatives, passer rating, and even DYAR were all devised independently of this year's MVP argument. Whatever motivation there was behind developing these metrics, and whatever flaws they have, they are still objective, and that's important.

 

[RetractableRoof]

That's because it's motivated reasoning.

Put simply, Y/A and its derivatives, passer rating, and even DYAR were all devised independently of this year's MVP argument. Whatever motivation there was behind developing these metrics, and whatever flaws they have, they are still objective, and that's important.

Agreed, it's important. [However biased I might have been, I really was trying to keep an open mind and see if I was wrong in my belief that his supporting cast was heads and shoulders better. Being not as smartly as all of you folks, I know it was crude and maybe inaccurate - but I don't know how to show what I wanted other than to do what I did. I'm not holding it up as peer reviewed work, only that it does SEEM to illustrate what my (and others) instinct thought. Show me how to do the maths better.] However when the creators of DYAR have stated it doesn't separate a QB performance from his teammates, so it CAN"T be used as the ONLY method of assessing the performance - and doing so seems um... wrong? This isn't the "highest total TEAM offensive performance" conversation right?

I've re-read what I started to write and removed some useless snark aimed at others. It's pointless to continue the snark. Genuine thanks for a straight answer/response.

 

[Stitch01]

Im surprised there is controversy about this at this point, Ryan should win. He checks the conventional and statistical analysis boxes.

Sure, Id take Brady and Rodgers if you made me pick "which QB will play better in a game tomorrow" and I think Brady wins the award if he plays the first four games like he played the rest of the season, but most valuable player in 2016? Its pretty clearly Ryan by pretty much every metric (yes advanced stats aren't perfect and need to be taken in the context of team performance).

 

[Van Everyman]

Why is it surprising? Matt Ryan has been an average to slightly above quarterback for multiple seasons on a mid-market team that hasn't had any postseason success. Brady has been the greatest quarterback of all time over the course of 16 years and is coming off one of his best seasons ever. On top of that, he was forced to sit four games for reasons that equate to pure partisanship.

Yes, that is the definition of bias. But it is rooted at least as much in our familiarity as fans with his value over the course of his career – which I would argue may tell us more about Brady's value this year than even the most accurate advanced stats. Unlike Ryan, we have seen a lot of tape that demonstrates the consistent value Brady adds – with different OLs and OL coordinators, vastly different levels of receiving talent and so forth.

I also think familiarity leads us to place a higher value on decisionmaking than we would otherwise. SIRI tells me that "Tom Brady didn't have a great game against the Broncos. He completed 16 of 32 passes for 188 yards. And he had a quarterback rating of 68.2." And I'm not sure that DVOA would tell us any more than SIRI – even after adjusting for defense and the fact that it was a road game. But I'm pretty confident after last year's AFCCG and having watched almost every game the guy is played over the last 16 years that it was actually one of the best games he's played as a New England Patriot – better than the Titans game in 2009 when he threw for 6 TDs and had a rating of 152.8.

And I think that has to count for something in this race. How much, I can't tell you. But in a race this close, given all we know about the centrality of a quarterback's performance to his team's success, it has to count for more than nothing – which I feel is what a lot of posters here relying solely on stats are essentially arguing.

 

[Conigliaro's Potential]

And I think that has to count for something in this race. How much, I can't tell you. But in a race this close, given all we know about the centrality of a quarterback's performance to his team's success, it has to count for more than nothing – which I feel is what a lot of posters here relying solely on stats are essentially arguing.

Would you be making this same argument on behalf of Aaron Rodgers or Ben Roethlisberger? I feel like pro-Brady partisanship is driving a lot of the arguments against Ryan. (I'm a Falcons fan, to be fair, but I think I'm alone on that particular corner here.)

This very much feels like the sort of conversation people in forums like this were having about baseball stats 15 or 20 years ago. I've enjoyed that element of reminiscing, actually - feels like old times. And it's one of the best sabermetrically minded conversations and arguments about football I can remember seeing. I do feel like some of the anti-football stats people aren't that different from the anti-baseball stats people make fun of all the time, though; perhaps we're forgetting that advanced baseball stats aren't widely understood or accepted by everyone either?

 

[coremiller]

1) The race isn't actually that close, so we shouldn't need to be invoking wholly made-up "decision-making" tiebreakers.

2) Brady's game against Denver was worth 48 DYAR, a lot better than his box score stats would suggest, mostly because Denver's pass defense was awesome this year and so Brady gets a huge opponent adjustment. The opponent adjustment was worth alone was worth 79 DYAR: Brady was -31 YAR without it. I actually think this is a case where DYAR is quite useful, because of the opponent adjustment.

I want to say more about this decision-making business. It seems to resolve mostly around two facts: Brady's extreme TD/INT ratio, and a couple of mistakes Ryan made this year in key moments. I'll dispense with the latter first. Matt Ryan dropped back to pass 571 times this year. That we should ignore 99% of those attempts in favor of two or three plays for a season-long award is total nonsense. It's like denying a baseball player the MVP because he struck out in the 10th inning of a game in July. We should be looking at the whole body of work.

Re TD/INT ratio, this is a more analytically interesting issue. But first, to frame the debate we need to remember that the margins are fine here. Brady had a INT% or .5, while Ryan was 1.3. Over the course of a season, that works out to something like 4 additional interceptions (Ryan actually had 5 more, but threw more passes/played in more games), which is not a big amount.

Second, you can't analyze INT% on its own without doing a risk/reward analysis. A QB could have an extremely low INT% if they never take risks and alwayas check down or throw the ball away, but that wouldn't be very valuable. If anyone with a PFF subscription wants to post their Average Depth of Target data, that would hep inform this discussion. But I'll note that Ryan's Yards/Completion was 13.3, while Brady's was 12.2, which suggests that Ryan was throwing downfield more often and taking more risks. And Ryan was throwing downfield extremely well, since he had a higher completion % than Brady despite the higher Y/C (normally Comp % and Y/C are roughly inversely correlated), and a higher TD%. So one plausible hypothesis supported by this data is that Ryan ended up with a slightly higher INT% and worse TD/INT ratio because he was taking more risks, but those risks were justified because he also hit a lot more big plays.

This is where Expected Points Added would really be useful. But without it, ANY/A is as close as we'll get. The value of ANY/A is that it gives TDs and INTs their proper EPA weight on average and then rolls all of those factors into one easy-to-use stat.

 

[Stitch01]

Why is it surprising? Matt Ryan has been an average to slightly above quarterback for multiple seasons on a mid-market team that hasn't had any postseason success. Brady has been the greatest quarterback of all time over the course of 16 years and is coming off one of his best seasons ever. On top of that, he was forced to sit four games for reasons that equate to pure partisanship.

OK, I don't disagree that much with any of this (it underrates Matt Ryan some, but doesn't matter), but not sure what it has to do with MVP voting other than the fact that Brady missed a quarter of the season.

Yes, that is the definition of bias. But it is rooted at least as much in our familiarity as fans with his value over the course of his career – which I would argue may tell us more about Brady's value this year than even the most accurate advanced stats. Unlike Ryan, we have seen a lot of tape that demonstrates the consistent value Brady adds – with different OLs and OL coordinators, vastly different levels of receiving talent and so forth.

Well, yeah, that is bias in the sense that we aren't watching every snap that Matt Ryan took this year in the same level of detail so I dont put much weight into the head-to-head comparison. Also not sure what Brady's performance in 2005 or w/e tells us about whether he deserves an MVP in 2016.

I also think familiarity leads us to place a higher value on decisionmaking than we would otherwise. SIRI tells me that "Tom Brady didn't have a great game against the Broncos. He completed 16 of 32 passes for 188 yards. And he had a quarterback rating of 68.2." And I'm not sure that DVOA would tell us any more than SIRI – even after adjusting for defense and the fact that it was a road game. But I'm pretty confident after last year's AFCCG and having watched almost every game the guy is played over the last 16 years that it was actually one of the best games he's played as a New England Patriot – better than the Titans game in 2009 when he threw for 6 TDs and had a rating of 152.8.

He played better than his raw stats say he did against Denver.

Ive watched every game Tom Brady has played for the last 16 years and for sure don't think that was one of Tom Brady's best games as a Patriot.

And I think that has to count for something in this race. How much, I can't tell you. But in a race this close, given all we know about the centrality of a quarterback's performance to his team's success, it has to count for more than nothing – which I feel is what a lot of posters here relying solely on stats are essentially arguing

Cant get much more central to the game as a quarterback than showing up to and playing in the game. Deflategate was the bullshit to end all bullshit, but I dont think you can just magically wish it away when voting for MVP. You could probably talk me into Brady being a bit better than Ryan on a per game basis, but 25% better? I dont see the argument. Ryan was really good this year.

EDIT: Cant find ADOT at PFF, think it must be under the $249/month premium stats. Annoying they changed how their subscriptions work.

FWIW PFF grades Brady as almost literally God this year, and that's still not really enough vs. how they grade Ryan to give Brady the MVP

 

[Conigliaro's Potential]

Second, you can't analyze INT% on its own without doing a risk/reward analysis. A QB could have an extremely low INT% if they never take risks and alwayas check down or throw the ball away, but that wouldn't be very valuable. If anyone with a PFF subscription wants to post their Average Depth of Target data, that would hep inform this discussion. But I'll note that Ryan's Yards/Completion was 13.3, while Brady's was 12.2, which suggests that Ryan was throwing downfield more often and taking more risks. And Ryan was throwing downfield extremely well, since he had a higher completion % than Brady despite the higher Y/C (normally Comp % and Y/C are roughly inversely correlated), and a higher TD%. So one plausible hypothesis supported by this data is that Ryan ended up with a slightly higher INT% and worse TD/INT ratio because he was taking more risks, but those risks were justified because he also hit a lot more big plays.

Heh...this information is in the second paragraph of an article I just read on the SBNation Falcons site:

http://www.thefalcoholic.com/2017/1/5/14171824/matt-ryan-for-mvp-part-2-addressing-the-criticisms-aaron-rodgers-tom-brady

Nice piece, actually - some good retorts to common media (and occasional SoSHer) misperceptions. My favorite stat in this one: Matt Ryan when not throwing to Julio Jones = 290/405 for 3535 yards, 32 TDs, 3 INTs, 121.4 passer rating.

 

[Dahabenzapple2]

Heh...this information is in the second paragraph of an article I just read on the SBNation Falcons site:

http://www.thefalcoholic.com/2017/1/5/14171824/matt-ryan-for-mvp-part-2-addressing-the-criticisms-aaron-rodgers-tom-brady

Nice piece, actually - some good retorts to common media (and occasional SoSHer) misperceptions. My favorite stat in this one: Matt Ryan when not throwing to Julio Jones = 290/405 for 3535 yards, 32 TDs, 3 INTs, 121.4 passer rating.

How about his stats when Julio Jones is not on the field?

 

[BusRaker]

Nice piece, actually - some good retorts to common media (and occasional SoSHer) misperceptions. My favorite stat in this one: Matt Ryan when not throwing to Julio Jones = 290/405 for 3535 yards, 32 TDs, 3 INTs, 121.4 passer rating.

Not sure if that is relevant if defenses are double and triple covering JJ and leaving the other options open. I'd be interested to see similar stats with Brady to other receivers when Gronk is on the field.

 

[Rudy's Curve]

Ryan had a higher ANY/A+ (adjusts for era) this year than Brady did...in 2007. Now it's by the smallest of margins and an adjustment for weather would likely tip the scales slightly the other way, but it's pretty amazing how many people want to deny a historically awesome season the MVP.

 

[Super Nomario]

My rough EPA assessment (using PFR.com's numbers for play-by-play) has Ryan at +234.91 and Brady at +181.64. That actually makes Brady the slightly better player on a per-game basis (+15.14 vs +14.68), but obviously there's a significant gap because of the missed time. Worth noting:

• Does not account for weather
• Does not account for opposing defenses
• Does not account for supporting casts
• Does not account for fumble luck - Brady recovered all five of his own fumbles; Ryan lost two. Stats like DVOA treat all fumbles like half a lost fumble
• Includes DPI but not other "no play" penalties
• Some other weirdness I didn't factor out, like post-play penalties (e.g., personal fouls). I'll get rid of some of this, but there's always just some weird stuff and it gets grey when to sort it out.
• This doesn't factor in clutchness; so Ryan's pick-two or Brady's missed connection to Gronk on 4th-and-1 just count as negative points without considering how key those points were at that stage of those games.
• PFR.com's EPA does exactly align with other measures of EPA I've seen. I know Brian Burke counts TDs as 6.3 points rather than 7 because having to kickoff to the other team is worth -0.7 points, for instance, but PFR just uses 7.

Brady did an absolutely amazing job avoiding negative plays. He only had five all season that were -2 or worse (and one of those was the Edelman taunt last week that wasn't Brady's fault). Ryan had 22 such negative plays. Ryan had a lot more positive plays (+2 or better) to compensate - 91 vs 65.

 

[RetractableRoof]

1) The race isn't actually that close, so we shouldn't need to be invoking wholly made-up "decision-making" tiebreakers.

2) Brady's game against Denver was worth 48 DYAR, a lot better than his box score stats would suggest, mostly because Denver's pass defense was awesome this year and so Brady gets a huge opponent adjustment. The opponent adjustment was worth alone was worth 79 DYAR: Brady was -31 YAR without it. I actually think this is a case where DYAR is quite useful, because of the opponent adjustment.

I want to say more about this decision-making business. It seems to resolve mostly around two facts: Brady's extreme TD/INT ratio, and a couple of mistakes Ryan made this year in key moments. I'll dispense with the latter first. Matt Ryan dropped back to pass 571 times this year. That we should ignore 99% of those attempts in favor of two or three plays for a season-long award is total nonsense. It's like denying a baseball player the MVP because he struck out in the 10th inning of a game in July. We should be looking at the whole body of work.

Re TD/INT ratio, this is a more analytically interesting issue. But first, to frame the debate we need to remember that the margins are fine here. Brady had a INT% or .5, while Ryan was 1.3. Over the course of a season, that works out to something like 4 additional interceptions (Ryan actually had 5 more, but threw more passes/played in more games), which is not a big amount.

Second, you can't analyze INT% on its own without doing a risk/reward analysis. A QB could have an extremely low INT% if they never take risks and alwayas check down or throw the ball away, but that wouldn't be very valuable. If anyone with a PFF subscription wants to post their Average Depth of Target data, that would hep inform this discussion. But I'll note that Ryan's Yards/Completion was 13.3, while Brady's was 12.2, which suggests that Ryan was throwing downfield more often and taking more risks. And Ryan was throwing downfield extremely well, since he had a higher completion % than Brady despite the higher Y/C (normally Comp % and Y/C are roughly inversely correlated), and a higher TD%. So one plausible hypothesis supported by this data is that Ryan ended up with a slightly higher INT% and worse TD/INT ratio because he was taking more risks, but those risks were justified because he also hit a lot more big plays.

This is where Expected Points Added would really be useful. But without it, ANY/A is as close as we'll get. The value of ANY/A is that it gives TDs and INTs their proper EPA weight on average and then rolls all of those factors into one easy-to-use stat.

1) Since you've decided it isn't that close, we shouldn't bother talking any more right? You are comfortable applying a flawed stat(s) captured in an environment of largely unmeasurable interactions and declared that your statistics prove the case made. Then be comfortable in your skin and walk away.

2) You are giving Ryan credit for being more aggressive "because more results". "The results were justified because he also hit a lot more big plays". His approach works better because the high risk high reward pays off I guess - except that his team had leads late in games. His high risk approach, led to turnovers which cost his team wins. His team experienced a loss because of turnovers... and since the name of the game is to win, the risk when already leading did NOT pay off. Perhaps such high risk/high reward needs to be tempered when leading the game late. More of a hybrid approach - except then throwing the ball away instead of being sacked or throwing the ball away instead of forcing it in and risking a turnover comes at the expense of a portion of the gaudy numbers. You can't praise the gaudy numbers and then say we should ignore the game losing turnover because it's only one play among many. The aggression/style of play that garnered the numbers are responsible for the losses in those two situations (yes, my opinion only). You want to credit him for the aggression, but absolve him of the downside of the aggression when it fails within the team goal: winning.

Here's the thing. I assume the biggest regular season goal for any team is likely to be home field advantage throughout the playoffs. Last year the Patriots were home on the couch during the SB at least partly because they ran into a buzz saw defense on the road in the playoffs. Had they been at home, their odds of winning go up. That shows how valuable the home field advantage is, the #1 goal of teams for the regular season. Judging any player on any team has to be in the context of the team goals. It doesn't matter to me if Jordan scores 63 points in a playoff game if his team loses - he didn't meet the team goal. Ryan had a lead in two games and lost those games because of turnovers. He squandered the games in my mind. His team could have been 13-3, potentially the #1 seed (I don't know about the tie breakers, and can't be bothered to look). If you ask him if he'd trade a few points of completion percentage or a few points of DVAR right now for the #1 seed I think we all know what his answer would be. In my mind I can't praise or reward ALL the aggression if it comes outside of or at the expense of the team goals - whether it was intentional or not.

I've already said that I view Ryan's offensive numbers as being a bit better than Brady's and for 1/4 longer. But I also discount Ryan for not 'managing' his aggressive style better in the context of the team goals in those two lost games. That may not be equitable in your eyes, but it's where I am.

 

[coremiller]

My rough EPA assessment (using PFR.com's numbers for play-by-play) has Ryan at +234.91 and Brady at +181.64. That actually makes Brady the slightly better player on a per-game basis (+15.14 vs +14.68), but obviously there's a significant gap because of the missed time. Worth noting:

• Does not account for weather
• Does not account for opposing defenses
• Does not account for supporting casts
• Does not account for fumble luck - Brady recovered all five of his own fumbles; Ryan lost two. Stats like DVOA treat all fumbles like half a lost fumble
• Includes DPI but not other "no play" penalties
• Some other weirdness I didn't factor out, like post-play penalties (e.g., personal fouls). I'll get rid of some of this, but there's always just some weird stuff and it gets grey when to sort it out.
• This doesn't factor in clutchness; so Ryan's pick-two or Brady's missed connection to Gronk on 4th-and-1 just count as negative points without considering how key those points were at that stage of those games.
• PFR.com's EPA does exactly align with other measures of EPA I've seen. I know Brian Burke counts TDs as 6.3 points rather than 7 because having to kickoff to the other team is worth -0.7 points, for instance, but PFR just uses 7.

Brady did an absolutely amazing job avoiding negative plays. He only had five all season that were -2 or worse (and one of those was the Edelman taunt last week that wasn't Brady's fault). Ryan had 22 such negative plays. Ryan had a lot more positive plays (+2 or better) to compensate - 91 vs 65.

Thanks a ton for doing this, it's very interesting. Did you sort out out kneeldowns?

 

[coremiller]

1) Since you've decided it isn't that close, we shouldn't bother talking any more right? You are comfortable applying a flawed stat(s) captured in an environment of largely unmeasurable interactions and declared that your statistics prove the case made. Then be comfortable in your skin and walk away.

2) You are giving Ryan credit for being more aggressive "because more results". "The results were justified because he also hit a lot more big plays". His approach works better because the high risk high reward pays off I guess - except that his team had leads late in games. His high risk approach, led to turnovers which cost his team wins. His team experienced a loss because of turnovers... and since the name of the game is to win, the risk when already leading did NOT pay off. Perhaps such high risk/high reward needs to be tempered when leading the game late. More of a hybrid approach - except then throwing the ball away instead of being sacked or throwing the ball away instead of forcing it in and risking a turnover comes at the expense of a portion of the gaudy numbers. You can't praise the gaudy numbers and then say we should ignore the game losing turnover because it's only one play among many. The aggression/style of play that garnered the numbers are responsible for the losses in those two situations (yes, my opinion only). You want to credit him for the aggression, but absolve him of the downside of the aggression when it fails within the team goal: winning.

Here's the thing. I assume the biggest regular season goal for any team is likely to be home field advantage throughout the playoffs. Last year the Patriots were home on the couch during the SB at least partly because they ran into a buzz saw defense on the road in the playoffs. Had they been at home, their odds of winning go up. That shows how valuable the home field advantage is, the #1 goal of teams for the regular season. Judging any player on any team has to be in the context of the team goals. It doesn't matter to me if Jordan scores 63 points in a playoff game if his team loses - he didn't meet the team goal. Ryan had a lead in two games and lost those games because of turnovers. He squandered the games in my mind. His team could have been 13-3, potentially the #1 seed (I don't know about the tie breakers, and can't be bothered to look). If you ask him if he'd trade a few points of completion percentage or a few points of DVAR right now for the #1 seed I think we all know what his answer would be. In my mind I can't praise or reward ALL the aggression if it comes outside of or at the expense of the team goals - whether it was intentional or not.

I've already said that I view Ryan's offensive numbers as being a bit better than Brady's and for 1/4 longer. But I also discount Ryan for not 'managing' his aggressive style better in the context of the team goals in those two lost games. That may not be equitable in your eyes, but it's where I am.

This long rant just boils down to QBWINZ!!!! The whole point of an ANY/A or EPA style analysis is not to entirely discount mistakes but to put the mistakes in a proper risk/reward context.

But thanks for the little lecture about how home-field advantage is valuable in the playoffs, I never knew that before.

 

[RetractableRoof]

This long rant just boils down to QBWINZ!!!! The whole point of an ANY/A or EPA style analysis is not to entirely discount mistakes but to put the mistakes in a proper risk/reward context.

But thanks for the little lecture about how home-field advantage is valuable in the playoffs, I never knew that before.

Actually, it boils down to QBs shouldn't lose games that are already won...

 

[Hendu for Kutch]

This long rant just boils down to QBWINZ!!!! The whole point of an ANY/A or EPA style analysis is not to entirely discount mistakes but to put the mistakes in a proper risk/reward context.

But thanks for the little lecture about how home-field advantage is valuable in the playoffs, I never knew that before.

What he wrote doesn't even remotely boil down to "QBWINZ!!!!". You're being an asshole. I say this with no dog in the fight. I'm also not the first person to say so in this thread, so please take it to heart.

 

[coremiller]

Actually, it boils down to QBs shouldn't lose games that are already won...

Which is the QBWINZ!!! fallacy. Teams win games, not players.

 

[tims4wins]

Which is the QBWINZ!!! fallacy. Teams win games, not players.

Team had game won, QB threw pick-2, team lost game.

 

[Rudy's Curve]

Team had game won, QB threw pick-2, team lost game.

There was 4:32 left when he threw that and KC had all three timeouts. In what universe is that game won?

 

[tims4wins]

There was 4:32 left when he threw that and KC had all three timeouts. In what universe is that game won?

True, but their win probability went from ~72% to ~33% in one play.

 

[Van Everyman]

How heavily does that win probability factor in the quarterbacks performance?

 

[rodderick]

Which is the QBWINZ!!! fallacy. Teams win games, not players.

Sure, teams win games, but I don't think it's laughable to give recognition to the player whose performance determines the outcome of games far more than any other.

Do you think "HC WINZ" is a fallacy as well?

 

[amarshal2]

Would you be making this same argument on behalf of Aaron Rodgers or Ben Roethlisberger? I feel like pro-Brady partisanship is driving a lot of the arguments against Ryan. (I'm a Falcons fan, to be fair, but I think I'm alone on that particular corner here.)

I literally did just that. I made the exact same argument and I said it should be Brady or Rodgers on this basis.

 

[RetractableRoof]

Wall of verbose text... sorry.

What makes an MVP anyway? Is it only on field performance? Hypothetically, if Ryan takes a pay cut so they can afford Jones - isn't that valuable? Takes a receiver aside and helps him break down film, is that part of being an MVP? A quiet word with someone who isn't getting it done in the weight room, or is partying too hard... is that part of leadership? Takes a rookie receiver in training camp and works on tightening up his routes, is that leadership? Challenges a receiver that dropped an easy ball, is that leadership? Is leadership part of being an MVP? If he spots a opponents defensive tendency on film and it gets worked into the game plan, is that part of being an MVP? None of this is measurable - so it doesn't exist, right? But I believe both QBs (all 3 if we count Rogers) do these types of things. Does doing it better count towards being an MVP if we COULD measure it?

What is most valuable about being a QB ON the field? Is being "Matty Ice" valuable because your team never sees you sweat? Is making a joke in the huddle the way Montana did valuable? Is demanding the best from your receivers the way Marino did valuable? Is slapping them on the butt and going right back to them after they've dropped one valuable? Is it intimidating an opposing defensive coach into not trying a blitz because you eat them up?

What about the pre-snap read? If you are the best in the game at it - isn't that valuable to your team? Doesn't it get you a hell of an advantage? Calling an audible because the matchups says you'll get an advantage by doing so. If you are really good at that, isn't it a large advantage? Calling the right time out in the right moment because you can't get to the right matchup and the play in front of you is critical?

What about pocket presence? Being able to slide in the pocket to prolong the play and get an extra moment for the less talented receiver to get open, or your 2nd or 3rd read to get open - because you don't have an all world receiver on the field for the last 8 games. Knowing when to throw the ball away to avoid the sack and preserve field position - completion percentage be damned. Being able to roll out and throw on the run the way a Wilson and Rogers do it. Isn't that valuable? Being able to run for the first down or run as a weapon the way Rogers did? Knowing that your receiver is so good that a jump ball is as good as a perfect throw, and letting it fly?

What about accuracy? Executing the game plan to a T? If your coach benches, buries, and cuts players for turning the ball over and you've gone a whole season only turning it over twice - isn't that valuable to the team? What if you've consistently thrown the ball low so that only your receivers can catch it - knowing that you'll get the first down if they do but no YAC. Is that valuable? Knowing that the ball isn't going to sail and be an easy interception. You've spent years working on your mechanics to be able to drive the ball this way. Isn't that valuable? What if instead you've thrown the ball within inches of two defenders fingertips into your all world receivers hands, leading him to huge YAC numbers? Which approach is more valuable? Ultra protective of the ball, or ultra aggressive? If both meet the game plan to a T, and both are perfectly accurate for their intended result aren't they equally valuable to the team? Is one more valuable because it goes for 22 yards versus 8 - even if the risk of the turnover is higher?

One approach to determining the best QB was this one: http://ftw.usatoday.com/2016/08/nfl-quarterback-rankings-2016#method by Steven Ruiz. How can we talk about the MVP without talking about who the best QB is? Or is being the best QB different from this years QB performance? Aren't the attributes of being the best QB often those that make you the most valuable as well?

Lots of questions... and I don't see statistics answering many of them soon either. Determining the MVP is probably going to remain subjective for some time - because even reviewing film, we don't know the intentions of the QB before the snap, during the play. What we perceive as a mistake make be a calculated decision (successful or not). What we may view as protecting the ball may conversely be a missed opportunity.

All these factor into an MVP season, and few are measurable. Ignoring them to focus solely on a couple of stats that are accepted by their creators as flawed seems to be limiting to discussion.

 

[Rudy's Curve]

True, but their win probability went from ~72% to ~33% in one play.

Should Ryan be penalized for the interception being returned for two points (which is pretty fluky), or rather just a missed conversion?

 

[amarshal2]

Ryan had a higher ANY/A+ (adjusts for era) this year than Brady did...in 2007. Now it's by the smallest of margins and an adjustment for weather would likely tip the scales slightly the other way, but it's pretty amazing how many people want to deny a historically awesome season the MVP.

I remember when Russel Wilson had a higher ANY/A than Brady in Super Bowl 49. That was the last time I looked at ANY/A.

Should Ryan be penalized for the interception being returned for two points (which is pretty fluky), or rather just a missed conversion?

Throwing interceptions isn't like BABIP. They have control over where the ball goes. Some QBs have more situational awareness than others. Brady is conservative in games where his defense is playing well and he's aggressive when he needs to be. He also rarely throws pick 6's or interceptions in the end zone.

 

[tims4wins]

Should Ryan be penalized for the interception being returned for two points (which is pretty fluky), or rather just a missed conversion?

There was nothing fluky about the pick or the return. Yes, 100 yard returns don't happen often, but on that specific route, a pick probably results in a TD a fair amount of the time. In the old days of not returning conversion attempts, then he was correct to throw the pass. In this new world, that was a really poor decision that cost his team 40% of win likelihood.

Edit: let's not forget that in addition to this pick-2, he also threw a pick-6 in the game

 

[djbayko]

This discussion is fascinating. Unfortunately, there is only one statistic which matters - votes!

According to this Reddit user who's tracking public announcements, the current count is:

• Ryan - 3
• Brady - 2
• Carr - 2

Out of 50. He's going by the 2014 voter list posted upthread. As we suspected, this is going to be a very splintered vote.

 

[coremiller]

Sure, teams win games, but I don't think it's laughable to give recognition to the player whose performance determines the outcome of games far more than any other.

Do you think "HC WINZ" is a fallacy as well?

Probably not, for two reasons:

1) It's the head coach's job to be responsible for every part of the team, and the HC has control over every part of the team. It's not the QB's fault if the defense sucks, but it is the head coach's.
2) For the most part, we don't have many other ways to evaluate HC performance. Things like team DVOA or SRS are helpful in this regard, and are good for identifying coaches/teams who are over/underperforming their record, especially in the short term, but in the long term they mostly end up in the same place, so there's not much cost to just using wins. OTOH, we have much more granular data for evaluating QBs based on the things they actually have control over, and we know that the correlation between QB performance and wins can be very loose (just look at Drew Brees the last few years). In a world in which we had no access to QB stats, QBWINZ would work as a reasonable rough heuristic of QB performance because there is a loose positive correlation between QB performance and team performance. But thankfully we don't live in that world.

The bigger fallacy for head coaches isn't WINZ but RINGZ. The most historically underrated HCs are the ones who won a ton of games with different teams/organizations/QBs but never won a Super Bowl, e.g. Marty Schottenheimer or Andy Reid. Especially when there was freakish bad luck involved, which was the case on several occasions for Schottenheimer.

 

[rodderick]

Probably not, for two reasons:

1) It's the head coach's job to be responsible for every part of the team, and the HC has control over every part of the team. It's not the QB's fault if the defense sucks, but it is the head coach's.
2) For the most part, we don't have many other ways to evaluate HC performance. Things like team DVOA or SRS are helpful in this regard, and are good for identifying coaches/teams who are over/underperforming their record, especially in the short term, but in the long term they mostly end up in the same place, so there's not much cost to just using wins. OTOH, we have much more granular data for evaluating QBs based on the things they actually have control over, and we know that the correlation between QB performance and wins can be very loose (just look at Drew Brees the last few years). In a world in which we had no access to QB stats, QBWINZ would work as a reasonable rough heuristic of QB performance because there is a loose positive correlation between QB performance and team performance. But thankfully we don't live in that world.

The bigger fallacy for head coaches isn't WINZ but RINGZ. The most historically underrated HCs are the ones who won a ton of games with different teams/organizations/QBs but never won a Super Bowl, e.g. Marty Schottenheimer or Andy Reid. Especially when there was freakish bad luck involved, which was the case on several occasions for Schottenheimer.

Thanks for the well thought out and non condescending answer. Your take is very reasonable.

 

[Stitch01]

Wall of verbose text... sorry.

What makes an MVP anyway? Is it only on field performance? Hypothetically, if Ryan takes a pay cut so they can afford Jones - isn't that valuable? Takes a receiver aside and helps him break down film, is that part of being an MVP? A quiet word with someone who isn't getting it done in the weight room, or is partying too hard... is that part of leadership? Takes a rookie receiver in training camp and works on tightening up his routes, is that leadership? Challenges a receiver that dropped an easy ball, is that leadership? Is leadership part of being an MVP? If he spots a opponents defensive tendency on film and it gets worked into the game plan, is that part of being an MVP? None of this is measurable - so it doesn't exist, right? But I believe both QBs (all 3 if we count Rogers) do these types of things. Does doing it better count towards being an MVP if we COULD measure it?

What is most valuable about being a QB ON the field? Is being "Matty Ice" valuable because your team never sees you sweat? Is making a joke in the huddle the way Montana did valuable? Is demanding the best from your receivers the way Marino did valuable? Is slapping them on the butt and going right back to them after they've dropped one valuable? Is it intimidating an opposing defensive coach into not trying a blitz because you eat them up?

What about the pre-snap read? If you are the best in the game at it - isn't that valuable to your team? Doesn't it get you a hell of an advantage? Calling an audible because the matchups says you'll get an advantage by doing so. If you are really good at that, isn't it a large advantage? Calling the right time out in the right moment because you can't get to the right matchup and the play in front of you is critical?

What about pocket presence? Being able to slide in the pocket to prolong the play and get an extra moment for the less talented receiver to get open, or your 2nd or 3rd read to get open - because you don't have an all world receiver on the field for the last 8 games. Knowing when to throw the ball away to avoid the sack and preserve field position - completion percentage be damned. Being able to roll out and throw on the run the way a Wilson and Rogers do it. Isn't that valuable? Being able to run for the first down or run as a weapon the way Rogers did? Knowing that your receiver is so good that a jump ball is as good as a perfect throw, and letting it fly?

What about accuracy? Executing the game plan to a T? If your coach benches, buries, and cuts players for turning the ball over and you've gone a whole season only turning it over twice - isn't that valuable to the team? What if you've consistently thrown the ball low so that only your receivers can catch it - knowing that you'll get the first down if they do but no YAC. Is that valuable? Knowing that the ball isn't going to sail and be an easy interception. You've spent years working on your mechanics to be able to drive the ball this way. Isn't that valuable? What if instead you've thrown the ball within inches of two defenders fingertips into your all world receivers hands, leading him to huge YAC numbers? Which approach is more valuable? Ultra protective of the ball, or ultra aggressive? If both meet the game plan to a T, and both are perfectly accurate for their intended result aren't they equally valuable to the team? Is one more valuable because it goes for 22 yards versus 8 - even if the risk of the turnover is higher?

One approach to determining the best QB was this one: http://ftw.usatoday.com/2016/08/nfl-quarterback-rankings-2016#method by Steven Ruiz. How can we talk about the MVP without talking about who the best QB is? Or is being the best QB different from this years QB performance? Aren't the attributes of being the best QB often those that make you the most valuable as well?

Lots of questions... and I don't see statistics answering many of them soon either. Determining the MVP is probably going to remain subjective for some time - because even reviewing film, we don't know the intentions of the QB before the snap, during the play. What we perceive as a mistake make be a calculated decision (successful or not). What we may view as protecting the ball may conversely be a missed opportunity.

All these factor into an MVP season, and few are measurable. Ignoring them to focus solely on a couple of stats that are accepted by their creators as flawed seems to be limiting to discussion.

This is a eclectic mix of

--Things that will show up in statistics
--Things that you, me, and the MVP voters don't really have an informed opinion about
--An argument to vote for the QB of the most effective team offense

 

[RetractableRoof]

This is a eclectic mix of

--Things that will show up in statistics
--Things that you, me, and the MVP voters don't really have an informed opinion about
--An argument to vote for the QB of the most effective team offense

Then I already goofed up, because I intended no arguments in there... just thoughts. I guess I need to step away from the keyboard

 

[coremiller]

All these factor into an MVP season, and few are measurable. Ignoring them to focus solely on a couple of stats that are accepted by their creators as flawed seems to be limiting to discussion.

Motivated reasoning and appeals to subjective impressions and feelings limit discussion a lot more than statistical analysis does.

 

[RetractableRoof]

Most all discussion in a sports forum has an element of motivated reasoning in it. Not ideal, but it's there.

Appeals to subjective impressions... or perhaps trying to look past a model that is only capturing a fraction of the information.

Statistical analysis is valuable, pretending that this is a domain with anything more than partially complete view of the subject is dishonest. And belittling others who are attempting to look beyond a partially complete view has it's own name...

 

[Super Nomario]

Thanks a ton for doing this, it's very interesting. Did you sort out out kneeldowns?

Yeah, and spikes (but not tackles by Brandon Spikes). Also, plays where they completed a pass and then the receiver fumbled.

 

[tims4wins]

Phil Perry ‏@PhilAPerry 5m5 minutes ago

Phil Perry Retweeted AP NFL

Matt Ryan over Tom Brady for All-Pro honors. Would not be surprised if the results were the same for MVP.

 

[bradmahn]

If I had a vote, I would have voted Brady 1st Team All-Pro and Ryan MVP. Ryan's four games of play makes his contribution more valuable than Brady's simply by having played at a similarly high level for many more snaps. That said, as others have pointed out, I think Brady gets a boost by the fact his offensive production was with less talent around him and he did a better job avoiding mistakes.

Brady's historical INT-avoiding season is a big deal. I've seen it in the past but looking into ANY/A confirmed, INT/ATT is most negatively-correlated to wins. The highest positively-correlated statistic with wins is ANY/A and while Ryan leads by .22 yards in that stat, I'm convinced that has a lot to do with the talent around him, namely his two RBs, T. Coleman (12th in DVOA) and (D. Freeman, 14th). Blount, FWIW, was 18th.

If you take the 26% at 10.4 Y/A figure Ryan put up on play-action passing and compare it with Brady's (assumed) 19% at 9.5 Y/A number, it's a difference of 31 plays if Brady simply had attempted the same percentage of play-action passes as Ryan. If Brady merely produces at the same rate of 9.5 Y/A on those 31 plays, that's an additional 295 yards. Brady's resultant ANY/A would be 9.38. Before massaging, it was 8.81 vs 9.03 for Ryan.

This gets to the point other Patriots fans have brought up and explored with DVOA for their respective pass catchers: Ryan had a stronger supporting cast than Brady, both at WR and RB. Brady's edge at TE was significant before Gronk went out for the year and remained with Bennett afterward but I don't think it makes up for having two superior backs and maybe the best WR in the game not named Antonio Brown and is a big reason for the edge I give Brady in the All-Pro voting.

 

[tims4wins]

One thing on Ryan: apparently 4 of his 7 picks were in the red zone. And the 7 picks don't include the pick-2.

 

[HowBoutDemSox]

Collinsworth on the SEA/DET broadcast said he voted for Brady.

 

[djbayko]

Collinsworth on the SEA/DET broadcast said he voted for Brady.

• Ryan - 3
• Brady - 3
• Carr - 2

Known public votes out of 50 (I'm not actively searching so there may be more).

 

[djbayko]

Edit: I can't read.

 

[Mugsy's Walk-Off Bunt]

Looking forward to TB12 getting the real MVP Award tomorrow night.

 

[Ed Hillel]

Wow, I thought it would have been at least a 13-15 vote margin. That was close.

 

[Oppo]

Highest votes/games played! All about rate stats or something.

 

[mauf]

Hey, you guys ruined the TV show for me! Isn't there a "no spoilers" thread?