Jump to content
Washington Football Team Logo
Extremeskins

PFR: Winning vs. Stats


KDawg

Recommended Posts

http://www.pro-football-reference.com/blog/?p=8876

My favorite predictive stat for future success is net yards per attempt, i.e., yards per attempt with sack yards subtracted from the numerator and sacks added to the denominator. My favorite retrodictive or explanatory statistic is Adjusted Net yards per Attempt, which formed the foundation for how I graded quarterbacks in my Greatest QB of All-Time series. Adjusted Net Yards per Attempt starts with the basic NY/A formula, but adjusts the numerator with touchdown bonuses and interception penalties. But because those metrics are very random events -- important in determining winners, but not useful in determining future winners -- I prefer NY/A to ANY/A when judging a quarterback's true ability.

Why is that important? Because while Ryan performed well in ANY/A, that was largely because of his strong touchdown and interception rates. Coupled with his legitimately strong performances in clutch situations, it's easy to see why the Falcons won 13 games last year. But touchdowns, interceptions, and performance in high-leverage situations are factors that don't lend themselves well to repeatability. They tend to determine games, but they don't reflect skill or ability particularly well. And, in fact, while Ryan performed really well in certain areas, he was below the league average in net yards per attempt. Ryan ranked 25th out of 32 quarterbacks in NY/A in 2010, in a cluster behind Chad Henne, Matt Hasselbeck and ahead of Kerry Collins and Shaun Hill.

If you actually read the article, and I highly recommend it, it focuses on Matt Ryan. However, I think it brings up some interesting conversation pieces.

Here's another excerpt:

Not surprisingly, wins and net yards per attempt are strongly correlated. How rare is it for a quarterback to have the sort of success Ryan's team had with Ryan's individual statistics? I looked at all quarterback seasons from 1970 to 2009 which met the following three criteria:

•The quarterback won at least 10 games as a starter

•The quarterback won at least 70% of his games

•The quarterback had at least 10 times as many passing yards as rushing yards

138 quarterbacks met those three criteria, or roughly four players per season. In 2009, there were five such players -- Philip Rivers, Drew Brees, Peyton Manning, Brett Favre and Donovan McNabb: all were well above league average in NY/A. Only three did so in 2010: Tom Brady, Matt Ryan and Joe Flacco. Brady ranked 3rd in the league in NY/A with a 7.2 average. While Flacco only averaged 0.3 more NY/A than Ryan, Flacco finished 13th overall while Ryan was at the bottom of a crowded pack at 25th.

Of the 138 quarterbacks from 1970 to 2009, only 11 ranked below league average in NY/A. The full list, below, sorted according to the ratio of their NY/A average to the league average NY/A rate:

So, as the article suggests, it seems as if NY/A seems like a good indicator of performance.

When only 11 out of 138 quarterbacks on the list are below the league average in NY/A, that obviously means 127/138 are at or above that average. I'd say that's a really good indicator.

One of the arguments we've been in the last few weeks is winning vs. stats due to all of the Vince Young rumors.

I'm not going to lie to you and say I understand what everything in the article means, particularly the math :ols: Particularly right here:

the CC between winning percentage in year N and winning percentage in year N+1 was 0.09. The CC between NY/A ratio in Year N and winning percentage in year N+1 was 0.20.

The above may as well be written in Swahili. :ols:

Is NY/A a good predictive indicator and is ANY/A a good explanatory stat in your opinion?

This same author used the ANY/A stat to create his Greatest QB of All Time Article... Found here:

http://www.pro-football-reference.com/blog/?p=3378

His top five are Manning, Marino, Young, Montana, Tarkenton.

We have two 'Skins in the top 15 on that list :). I don't know if it's 100% accurate, but looking at the names near the top I'd say it's a pretty good indicator of QB performance.

Take a look at both articles and let's talk some football!

---------- Post added May-24th-2011 at 07:58 AM ----------

Let's put this in a bit of perspective using some 'Skins players/rumored people we're interested in:

Vince Young:

Career NY/A: 100

Career AY/A: 96

Rex Grossman:

Career NY/A: 93

Career ANY/A: 92

Donovan McNabb:

Career NY/A: 100

Career ANY/A: 106

Jason Campbell:

Career NY/A: 95

Career ANY/A: 98

Using these four, and the same formula used above it would put the QBs in this order:

McNabb

Campbell

Young

Grossman

I'd say that's fairly close to being accurate. Obviously, though, the formula doesn't take into account system differences.

EDIT: Let's also keep in mind that this is an indicator of a teams' passing game. There is no way to separate a QBs statistical performance from his teammates. They are all reliant on each other. This is purely an indicator of their actual performance and not their skillset.

Link to comment
Share on other sites

Interesting, the article mentions the impact of Roddy White (and why the Falcons FO was so desperate to add another great WR in Julio Jones).

2) Roddy White in any other situation. White's WPA is 2.60, more than half of Ryan's WPA. When not throwing to White, Ryan isn't the same guy. He's good, but not anywhere close to what he is with White in the game.

Forgot to mention that this surprises me when considering he has Tony Gonzalez on his team.

Link to comment
Share on other sites

One of the arguments we've been in the last few weeks is winning vs. stats due to all of the Vince Young rumors.

Is NY/A a good predictive indicator and is ANY/A a good explanatory stat in your opinion?

I'm initially going to say no, however, if we had other minimum guidelines to the search results, then yes.

Vince Young had a high NY/A and ANY/A last season despite incredibly few # of attempts per game as well as a low completion %. I think for a low number of attempts, when a scrambling QB is forcing the defense to go out of position, he should have a higher completion %.

EDIT: Let's also keep in mind that this is an indicator of a teams' passing game. There is no way to separate a QBs statistical performance from his teammates. They are all reliant on each other. This is purely an indicator of their actual performance and not their skillset.

Which is why we need to factor in VY's weapons. Kenny Britt and Nate Washington had huge seasons for the Titans in comparison to what they contributed in 2009. Britt especially had a breakout year. I think we also need to add Chris Johnson to the list of weapons the Titans had. For guys with at least 40 receptions, Britt and Washington were both in the Top 10 in Yards Per Catch.

I think VY's weapons greatly inflated his NY/A numbers.

Link to comment
Share on other sites

KDawg: Is NY/A a good predictive indicator and is ANY/A a good explanatory stat in your opinion?
The flaw in the author's argument is that the NY/A is not a QB stat simply because he labels it so. It's a stat which measures a team's passing game. It measures play design, play calling, and the execution by 11 players.

The author proves that the quality of the passing game is a big factor in winning NFL games. That's hardly a surprise.

In the same vein, the "quarterback rating" is a poor way to grade QBs, but it can serve better as a quick-and-dirty way to rank the offensive passing game and the defense's ability to stop the passing game than the total yardage stat. The QBR should be relabeled.

Link to comment
Share on other sites

EDIT: Let's also keep in mind that this is an indicator of a teams' passing game. There is no way to separate a QBs statistical performance from his teammates. They are all reliant on each other. This is purely an indicator of their actual performance and not their skillset.

The flaw in the author's argument is that the NY/A is not a QB stat simply because he labels it so. It's a stat which measures a team's passing game. It measures play design, play calling, and the execution by 11 players.

He actually says that in the article. Not sure which, but he points that out. But it is an indicator of the QBs performance. But, it's his performance in conjunction with the rest of his team.

Link to comment
Share on other sites

KDawg: This is purely an indicator of their actual performance and not their skillset.

My position is that it's not a very good indicator of actual performance. If the QB's impact was 90% of the stat, it would be usable. But in my opinion, that statistic is 50 to 60% garbage (not QB's individual performance).

Link to comment
Share on other sites

My position is that it's not a very good indicator of actual performance. If the QB's impact was 90% of the stat, it would be usable. But in my opinion, that statistic is 50 to 60% garbage (not QB's individual performance).

Regardless of the cirumstance, the QB has to do what is asked of him. If his statistics are good, he is doing what is asked on a consistent basis. And because, as we both know, it's impossible to isolate quarterback performance from the rest of the team, I think its one of the best statistical ways of being able to tell a capable QB from a not so capable QB aside from perception. And perception is a dangerous road to head down.

Link to comment
Share on other sites

Regardless of the cirumstance, the QB has to do what is asked of him. If his statistics are good, he is doing what is asked on a consistent basis. And because, as we both know, it's impossible to isolate quarterback performance from the rest of the team, I think its one of the best statistical ways of being able to tell a capable QB from a not so capable QB aside from perception. And perception is a dangerous road to head down.
We disagree on this.

You and I could look at Tom Brady's performance over a period of time and agree pretty much on his tangibles, and we could make some obvious deductions about his intangibles. I think we could agree that he's a grade A pocket passer lucky enough to have performed with the best supporting system in the league. But, Brady's reputation goes well beyond his abilities because of misunderstandings of statistics.

Link to comment
Share on other sites

But, Brady's reputation goes well beyond his abilities because of misunderstandings of statistics.

This is perception.

And for the record, I agree with you. However, there is no statistical way to prove that. There is statistical proof that he has been extremely consistent when in his system. Perception is an inexact science. You could pull Brady off the Pats and put him on the 'Skins and see an awful QB. Or maybe one similar to what's in NE now. We don't know and can't know until it happens, and even then it may not be accurate because of his age.

That's the problem with any player... It's tough to gauge their quality due to time in systems and reliance on the rest of the team. It doesn't matter the position, either.

Stats are the only concrete thing we have.

The rest is based on guessing, unless you're in the organization, and even then it's difficult to get a full read on his total abilities. What we do know, though, is that Tom Brady is a hell of a quarterback with the New England Patriots in the Belichick system.

Link to comment
Share on other sites

I agree with both of you.

I think a quarterbacks tangibles form the bulk of what makes a good QB.

But, at the NFL level quarterbacks (and players in general) aren't separated by that wide of a margin when it comes to physical ability.

But, the differences that exsist are important.

In regard to Brady I think one thing that gets overlooked is that Tom Brady throws a great ball.

He has very consistent, efficient (not to mention a much better arm then he gets credit for) and to my eye a very aesthetically pleasing throwing motion and all these go part in parcel with his above average ball placement.

The other factor used to judge quarterbacks is much more subjective because football at its core is a team game, and the OL, receivers, scheme all play a role in the QBs success/production.

But, when a QB shows far above average efficiency/production its imo a sign of his individual ability within that system.

I think an good example is the production Brady had within the system in 2007.

Link to comment
Share on other sites

I agree with both of you.

I think a quarterbacks tangibles form the bulk of what makes a good QB.

But, at the NFL level quarterbacks (and players in general) aren't separated by that wide of a margin when it comes to physical ability.

But, the differences that exsist are important.

In regard to Brady I think one thing that gets overlooked is that Tom Brady throws a great ball.

He has very consistent, efficient (not to mention a much better arm then he gets credit for) and to my eye a very aesthetically pleasing throwing motion and all these go part in parcel with his above average ball placement.

The other factor used to judge quarterbacks is much more subjective because football at its core is a team game, and the OL, receivers, scheme all play a role in the QBs success/production.

But, when a QB shows far above average efficiency/production its imo a sign of his individual ability within that system.

I think an good example is the production Brady had within the system in 2007.

I don't disagree.

However, even QBs with outstanding mechanics can fail in another system. That's why perception is difficult to work with. If Brady went to a team that didn't suit his abilities, and then didn't adjust to them, he wouldn't be near as good as he is now. He may still be good... or he may be awful due to the circumstances (not a fit, not comfortable, miserable, doesn't want to be there, etc.)

There's a ton that goes into it. But for the most part, I'd agree.

But it was argued that Ryan Leaf had great physical ability.

It didn't turn out that way.

Link to comment
Share on other sites

...Stats are the only concrete thing we have...
In your eyes stats are concrete. In mine, they are deceiving most football fans and many of the coaches.

Joe Gibbs showed that he was mightily impressed by the turnover stat and its correlation to winning. he recited it like a mantra even in Gibbs One. It was clear to me from what he didn't say, that he didn't understand the stat. I think his misunderstanding led to his preference for a game manager QB.

The rest is based on guessing...
The best anyone can do is make educated guesses. But, as I see it, that's better than working off misinformation (statistics that don't measure what they're supposed to measure).

---------- Post added May-25th-2011 at 10:52 AM ----------

I agree with both of you.

I think a quarterbacks tangibles form the bulk of what makes a good QB.

But, at the NFL level quarterbacks (and players in general) aren't separated by that wide of a margin when it comes to physical ability.

But, the differences that exsist are important.

In regard to Brady I think one thing that gets overlooked is that Tom Brady throws a great ball.

He has very consistent, efficient (not to mention a much better arm then he gets credit for) and to my eye a very aesthetically pleasing throwing motion and all these go part in parcel with his above average ball placement.

The other factor used to judge quarterbacks is much more subjective because football at its core is a team game, and the OL, receivers, scheme all play a role in the QBs success/production.

But, when a QB shows far above average efficiency/production its imo a sign of his individual ability within that system.

I think an good example is the production Brady had within the system in 2007.

I think Tom Brady's arm is among the best in the league, but the biggest factor in his success is the time he gets to throw. When consistently pressured by the Jets a couple of years ago, his QBR was 50. He was throwing off his back foot and missing targets badly even when the rush didn't come. He looked like a bum that day.
Link to comment
Share on other sites

This is perception.

And for the record, I agree with you. However, there is no statistical way to prove that. There is statistical proof that he has been extremely consistent when in his system. Perception is an inexact science. You could pull Brady off the Pats and put him on the 'Skins and see an awful QB. Or maybe one similar to what's in NE now. We don't know and can't know until it happens, and even then it may not be accurate because of his age.

That's the problem with any player... It's tough to gauge their quality due to time in systems and reliance on the rest of the team. It doesn't matter the position, either.

Stats are the only concrete thing we have.

The rest is based on guessing, unless you're in the organization, and even then it's difficult to get a full read on his total abilities. What we do know, though, is that Tom Brady is a hell of a quarterback with the New England Patriots in the Belichick system.

I think there are several recent examples of QB's leaving one system and going to another showing precisely that a QB putting up great numbers on one team puts up entirely diffent results on another. In the last 4 years we have Cutler with Broncos/Bears, Orton with Bears/Broncos, Cassel with Patriots/Chiefs and of course, the perennial arument source on this board with Campbell with Skins/Raiders (or McNabb with Eagles/Skins). Altough I will tend to discount the effect of any QB in the first year of a new system, these 4 QB's offer a unique opportunity to show, over the next few years, whether any QB stat is meaningful. It's the same player with different systems and different supporitng casts, yet each is still the number one QB on his new team. If the stat is a true indicator of QB worth, and not team worth, you would expect the stats to accurately reflect the change in team, since the QB himself did not change. There may not be enough data for a guy in is first year, but it should, over two or three seasons, reflect whether the stat is meaningful.
Link to comment
Share on other sites

I think there are several recent examples of QB's leaving one system and going to another showing precisely that a QB putting up great numbers on one team puts up entirely diffent results on another. In the last 4 years we have Cutler with Broncos/Bears, Orton with Bears/Broncos, Cassel with Patriots/Chiefs and of course, the perennial arument source on this board with Campbell with Skins/Raiders (or McNabb with Eagles/Skins). Altough I will tend to discount the effect of any QB in the first year of a new system, these 4 QB's offer a unique opportunity to show, over the next few years, whether any QB stat is meaningful. It's the same player with different systems and different supporitng casts, yet each is still the number one QB on his new team. If the stat is a true indicator of QB worth, and not team worth, you would expect the stats to accurately reflect the change in team, since the QB himself did not change. There may not be enough data for a guy in is first year, but it should, over two or three seasons, reflect whether the stat is meaningful.

I think all of those players remained fairly constant in terms of production when they switched teams with the exception of Cutler. The W/L column varied greatly due to the talent/lack of talent that was present, but the individual results that were put up were pretty alarmingly consistent. The thing that it shows is that QB's that are on poor offensive teams typically are undervalued (Campbell/Orton) and can produce at a higher level when given more to work with. On the flip side the players that came from an offensive Juggernaut (Cutler/McNabb) had a very difficult time adjusting to the weaker offensive teams. This suggests to me that a great deal of QB performance when it comes to winning and losing has to do with the supporting cast regardless of the individual numbers that the QB puts up.

Link to comment
Share on other sites

I think all of those players remained fairly constant in terms of production when they switched teams with the exception of Cutler....
Don't think so. Doing this from memory... but Cutler's QBR was the only one which remained constant. Orton's stats went up with the Patriots scheme run in Denver, Cassel's went way down moving from that scheme to KC..

The DVOA rankings for QBs which is a complex forumla which makes more sense than other rankings showed it isn't reliable as Cutler's ranking went way down and Orton's way up as a result of switching support systems. Cassel's ranking dropped way down too.

Link to comment
Share on other sites

I don't disagree.

However, even QBs with outstanding mechanics can fail in another system. That's why perception is difficult to work with. If Brady went to a team that didn't suit his abilities, and then didn't adjust to them, he wouldn't be near as good as he is now. He may still be good... or he may be awful due to the circumstances (not a fit, not comfortable, miserable, doesn't want to be there, etc.)

There's a ton that goes into it. But for the most part, I'd agree.

For me perception as nothing to do with my opinion of QB, and stats only has a some e.g. I think Campbell last year w/ the Skins was much better then his stats suggest, I think Favre last playoff run w/ the Pack was one of his best seasons, I think Cutler had his the best season of his career last year.

I think Brady has displayed a physical skillset to go along with efficiency that suggest that he would be a good QB in most systems.

My view of Brady wouldn't change if he played in the type of situation you suggest, although his stats would change.

But it was argued that Ryan Leaf had great physical ability.

It didn't turn out that way.

No doubt, but projecting college QBs into the NFL is another ball of wax.
Link to comment
Share on other sites

I think Tom Brady's arm is among the best in the league, but the biggest factor in his success is the time he gets to throw. When consistently pressured by the Jets a couple of years ago, his QBR was 50. He was throwing off his back foot and missing targets badly even when the rush didn't come. He looked like a bum that day.
Most QBs are effected by pressure and getting sacked/hit a little.

But, I agree with your point for as much as the sports media laud Manning and Brady for their footwork, both are more heavily dependent on good pass protection for their success then the other "elite" QBs like Brees and Rodgers.

---------- Post added May-25th-2011 at 09:14 PM ----------

I've heard this years ago. YPA is the best average to look at.

YPA is still a unit measure of the passing game.

Link to comment
Share on other sites

Archived

This topic is now archived and is closed to further replies.

  • Recently Browsing   0 members

    • No registered users viewing this page.
×
×
  • Create New...