Aaron Rodgers - Best 4th Quarter QB In GB In Last Decade ?

[Waldo]

I don't think QB rating is a very good evaluator. It is impacted too significantly by a couple things. If you really want to see it, run some hypotheticals with a QB who marches his team 80 yards through the air, but with the TD scored on a one yard run from the back, and another who has an offense that grinds it out mostly on the ground but he throws a 5 yard TD at the end.
9/15 for 80 yds 0 TDs = 74.3
3/5 for 15 yds and 1 TD = 104.1
Was the second performance really that much better? I don't think so. In both cases the offense scored, but in one the drive relied on the QB, in the other the passes were a change-up to the ball control drive.

One of the major components of QB rating is YPA. 15 throws in a single drive is A LOT. I believe the longest single drive the Pack had last year was a 16 play drive. 5.3 YPA is really not all that good (YPA tends to be constant and a good indicator of QB performance even if the offensive philosophy isn't). Keep the 80 yards, but change the x/y to reflect a completion % of 65%+ and a YPA of 7+ and the change to the QB rating should be quite significant for the positive.

I believe the stats came from here:

http://www.nfl.com/players/aaronrodgers/situationalstats?id=ROD339293

however the interpretation of the stat is incorrect, as it is performance in the 4th when the QB is ahead or behind by 7 or less.

[Patler]

I don't think QB rating is a very good evaluator. It is impacted too significantly by a couple things. If you really want to see it, run some hypotheticals with a QB who marches his team 80 yards through the air, but with the TD scored on a one yard run from the back, and another who has an offense that grinds it out mostly on the ground but he throws a 5 yard TD at the end.
9/15 for 80 yds 0 TDs = 74.3
3/5 for 15 yds and 1 TD = 104.1
Was the second performance really that much better? I don't think so. In both cases the offense scored, but in one the drive relied on the QB, in the other the passes were a change-up to the ball control drive.

One of the major components of QB rating is YPA. 15 throws in a single drive is A LOT. I believe the longest single drive the Pack had last year was a 16 play drive. 5.3 YPA is really not all that good (YPA tends to be constant and a good indicator of QB performance even if the offensive philosophy isn't). Keep the 80 yards, but change the x/y to reflect a completion % of 65%+ and a YPA of 7+ and the change to the QB rating should be quite significant for the positive.

Waldo, it was a hypothetical to demonstarte a point. Sure 15 attempts is a lot, and 5.3 YPA is low, but the other half of the hypothetical is even worse, 3.0 YPA, yet the QB rating is much higher.

What affects the calculation too significantly in my opinion, is a TD pass. Keeping all the other numbers the same, give the first guy a TD and his rating jumps from 74 to 97. (If you change the attempts and completions to 10/6 for 80 yards and 0 TDs, the rating increases to only 85). Take the TD away from the 2nd guy and his rating drops from 104 to 65.

When evaluating a QB's effectiveness in a comeback situation, it doesn't matter whether he engineers a drive with a winning TD pass or a drive with a winning TD run. My hypothetical should demonstrate that while each of the QB's was successful in the drive, one relied much more on the QB's passing ability, the other much more on the run blocking and runner. Using QB rating is just not an effective evaluation of the situation, in my opinion. That is what the hypothetical was designed to demonstrate.

[Patler]

I believe the stats came from here:

http://www.nfl.com/players/aaronrodgers/situationalstats?id=ROD339293

however the interpretation of the stat is incorrect, as it is performance in the 4th when the QB is ahead or behind by 7 or less.

You are probably correct about that. If he used the "4th Quarter within 7" numbers, which it appears he did, it would be ahead or behind. Using the QB rating to compare is even worse in that situation, in my opinion, because when you are ahead you aren't trying to do the things that lead to higher QB ratings. You are doing the things that tend to lead to lower QB ratings.

That page of the stats is interesting, but it never seems to have exactly what I am looking for. This is a good example. It has "4th quarter within 7" and it has "Ahead" "Ahead 1-8" "Behind" "Behind 1-8" etc., but it doesn't breakdown the 4th quarter by the ahead or behind categories. It would be real interesting if it did.

[Waldo]

I don't think QB rating is a very good evaluator. It is impacted too significantly by a couple things. If you really want to see it, run some hypotheticals with a QB who marches his team 80 yards through the air, but with the TD scored on a one yard run from the back, and another who has an offense that grinds it out mostly on the ground but he throws a 5 yard TD at the end.
9/15 for 80 yds 0 TDs = 74.3
3/5 for 15 yds and 1 TD = 104.1
Was the second performance really that much better? I don't think so. In both cases the offense scored, but in one the drive relied on the QB, in the other the passes were a change-up to the ball control drive.

One of the major components of QB rating is YPA. 15 throws in a single drive is A LOT. I believe the longest single drive the Pack had last year was a 16 play drive. 5.3 YPA is really not all that good (YPA tends to be constant and a good indicator of QB performance even if the offensive philosophy isn't). Keep the 80 yards, but change the x/y to reflect a completion % of 65%+ and a YPA of 7+ and the change to the QB rating should be quite significant for the positive.

Waldo, it was a hypothetical to demonstarte a point. Sure 15 attempts is a lot, and 5.3 YPA is low, but the other half of the hypothetical is even worse, 3.0 YPA, yet the QB rating is much higher.

What affects the calculation too significantly in my opinion, is a TD pass. Keeping all the other numbers the same, give the first guy a TD and his rating jumps from 74 to 97. (If you change the attempts and completions to 10/6 for 80 yards and 0 TDs, the rating increases to only 85). Take the TD away from the 2nd guy and his rating drops from 104 to 65.

When evaluating a QB's effectiveness in a comeback situation, it doesn't matter whether he engineers a drive with a winning TD pass or a drive with a winning TD run. My hypothetical should demonstrate that while each of the QB's was successful in the drive, one relied much more on the QB's passing ability, the other much more on the run blocking and runner. Using QB rating is just not an effective evaluation of the situation, in my opinion. That is what the hypothetical was designed to demonstrate.

I do think that it is at least decently fair in evaluating though, QB passes are tough throws.

6/10 80 yd, 0 TD, 85 rating
3/5 15 yd, 1 TD, 104 rating

The first guys 8.0 ypa is pretty good. But his completion % isn't that great, especially for later in a game where the defense might be playing the long ball and clock squeeze moreso than a coverage that would work for 4 quarters. At best he picked up 6 first downs by throwing it, but more likely something like 4. A QB's job gets much more difficult in the red zone, TD passes are the toughest ones of all.

The second guys ypa is terrible. However his completion % is the same as the first guy. He isn't driving the ball as far down the field, and the stats look like dumpoffs. 1-2 first downs is the most likely number. But he did complete a TD pass, which is a more difficult throw than any the first guy made.

One thing about rating, is isn't that great for evaluating performance at the micro level. Rating doesn't know whether that is one drive, or a QB that went 300/500 for 1,500 yards and 100 TD's, with 0 ints, over a season, as the sample size increases, rating gets better and better.

In fact, NFL.com doesn't even list the rating for QB's that fail to throw at least 14 passes/gm on average (224 over a 16 game season), as the stat was never really meant to be used to evaluate at the micro level, as at the micro level there is no relationship between TD:Int, something that must be established for the formula.

At the micro level, IMO the best measure is YPA, that tends to be the only performance indicator that runs true through almost any sample size, at least down to the drive level. With YPA, your sample data shows QB1 to be drastically superior to QB2. It overshoots a bit though IMO, as the TD cannot be ignored.

[pbmax]

But there are others who stopped by just to complain, not discuss anything. Those are the type of posts that need to end. Except for Patler. Patler can post about his flower garden and I would probably learn something.

Thanks! (But are you implying I dropped by just to complain??? )

Not at all. But since I was essentially complaining about non-contributing posts (not trolling, but not advancing the thread), I felt I needed to get you an exception for your critique of the admonishment.

[Patler]

I do think that it is at least decently fair in evaluating though, QB passes are tough throws.

6/10 80 yd, 0 TD, 85 rating
3/5 15 yd, 1 TD, 104 rating

The first guys 8.0 ypa is pretty good. But his completion % isn't that great, especially for later in a game where the defense might be playing the long ball and clock squeeze moreso than a coverage that would work for 4 quarters. At best he picked up 6 first downs by throwing it, but more likely something like 4. A QB's job gets much more difficult in the red zone, TD passes are the toughest ones of all.

The second guys ypa is terrible. However his completion % is the same as the first guy. He isn't driving the ball as far down the field, and the stats look like dumpoffs. 1-2 first downs is the most likely number. But he did complete a TD pass, which is a more difficult throw than any the first guy made.

One thing about rating, is isn't that great for evaluating performance at the micro level. Rating doesn't know whether that is one drive, or a QB that went 300/500 for 1,500 yards and 100 TD's, with 0 ints, over a season, as the sample size increases, rating gets better and better.

In fact, NFL.com doesn't even list the rating for QB's that fail to throw at least 14 passes/gm on average (224 over a 16 game season), as the stat was never really meant to be used to evaluate at the micro level, as at the micro level there is no relationship between TD:Int, something that must be established for the formula.

At the micro level, IMO the best measure is YPA, that tends to be the only performance indicator that runs true through almost any sample size, at least down to the drive level. With YPA, your sample data shows QB1 to be drastically superior to QB2. It overshoots a bit though IMO, as the TD cannot be ignored.

I agree with most of that, but I am not will to presume the TD pass was necessarily a more difficult throw from a mechanical performance aspect for the QB. In fact a short TD pass can be an easy toss mechanically for a QB of a predominantly running team when the defense is expecting the run. The pressure of a third and long or fourth and long can make that throw much more difficult than a first and goal when the QB looks for a specific situation and otherwise just throws it into the stands.

I know neither completion percentage was great. I used those to keep the percentage the same for each player and remove that as a significant factor. The intent was to emphasize the impact of throwing a TD pass or not throwing a TD pass.

I find QB rating to be an only mildly interesting stat for seasons or careers because it swings so widely on one play of the small sample. There have been many times when a QB plays quite poorly for 3.5 quarters, but with his team down by a couple TDs completes a bunch of passes for a bunch of yards and a meaningless TD as time expires to salvage a QB rating for the game that is OK when his actual performance during meaningful drives was quite poor.

[pbmax]

The QB rating is essentially completion percentage combined with adjustments similar to Adjusted Yards Per Attempts figures. This is then scaled and shifted to get the scale from 0 to 158.3.

At http://www.profootballreference.com , they provide individual adjusted Yards Per Attempt figures and calculate thusly:
Yards + (TDs * 10) - (INTs * 45)= Adjusted Yards Per Attempt (APYA).

The thinking is that yards are the basic measure of offensive progress and we need our stat to measure how efficiently the QB gets those yards. But not all yards are equal you say. A 3 yard Pass TD is worth much more, and contributes more to winning than almost any other 3 yard pass, even if it is for a first down. So any TD pass gets you a 10 yard bonus here in AYPA.

An interception yields possession and field position (if you hadn't been picked, at worst you punt, at best you keep picking up yards and first downs). AYPA penalizes you 45 yards.

The QB formula converts everything to completions. It starts with completions, adds a completion for every 20 yards, adds 4 completions for each TD and subtracts 5 completions for each INT. The adjusted number of completions is then divided by attempts, scaled then shifted.

[Waldo]

I've always liked the rating formula. I think once you understand it, and its limitations, you are fine.

One of the biggest things is to compare apples to apples. I don't think that a 25 att/gm QB can be compared to a 35 att/gm QB via rating. They are asked to shoulder an entirely different load. But it can compare QB's of similar workloads to one another.

And there is a significant amount of data jitter that smooths over time. Aaron was actually pretty consistent rating wise from one part of the season to the other, even if the individual games jumped around.

[pbmax]

Nearly all stats are affected by garbage time. Good defenses look worse, good offenses bury the needle, etc.

The biggest decision in taking apart a football game is deciding what the down by down objective is. What best measures progress toward winning? Touchdowns, possession, first downs, yards? None of them are perfect. And none match the finite and discrete pitch by pitch matchup in baseball between batter and pitcher.

But eliminating garbage time would make some things much more clear. You can buy access to the database at Football Outsiders which has all the GameBook stats broken down play by play. It would be much more flexible than the NFL.com stats or even foxsports stats page, which is the best conventional one I have found.

Only Harvey and Patler (now Chevelle) have found stats that aren't covered by the foxsports database. Usually the most difficult ones to find consider down by down or yards to go considerations.

[sharpe1027]

QB rating, alone, does not prove or disprove the amount of 4th quarter "comebackiness" a QB exhibits. I find the entire discussion kinda silly. What exactly is "comebackiness?"

Rodgers had good statistical numbers and the team had a bad W/L record. Many studies have shown/suggested that most of our decisions are based upon emotion and our gut feelings rather than logic. I think this is one of those cases, and we each have a feeling about Rodgers and will use logic to backup our feeling.

Stongest evidence of a lack of comebackiness:
No come back wins.

Strongest evidence of comebackiness:
Several go-ahead drives in the 4th quarter.

Verdict:

Last year the team's comebackiness sucked, and Rodgers was part of the team.

[sharpe1027]

And I would like to know if he has a flower garden, and if he does, how it's doing and what types of flowers he has in it. In the Romper Room, of course.

I beg your pardon, Patler never promised you a rose garden.

Christ, I'll have that damn song in my head for a week now.

[pbmax]

The original statistics quoted have nothing to do with combackiness. It says, 4th Quarter QB Ratings When the Teams are Within 7. I missed the Within part the first four times I read it (unless Chevelle has edited it). Which is a statistical attempt to mean the game is in doubt and still tightly contested.

And good statistical numbers and bad W/L record indicates not that the stats are deceiving, but that there may be other factors.

QB rating, alone, does not prove or disprove the amount of 4th quarter "comebackiness" a QB exhibits. I find the entire discussion kinda silly. What exactly is "comebackiness?"

Rodgers had good statistical numbers and the team had a bad W/L record. Many studies have shown/suggested that most of our decisions are based upon emotion and our gut feelings rather than logic. I think this is one of those cases, and we each have a feeling about Rodgers and will use logic to backup our feeling.

Stongest evidence of a lack of comebackiness:
No come back wins.

Strongest evidence of comebackiness:
Several go-ahead drives in the 4th quarter.

Verdict:

Last year the team's comebackiness sucked, and Rodgers was part of the team.

[Waldo]

Football is an odd sport. All of the so called rules or mantras that fans/media have come up with have limitations.

Problem is, we broke them all last season in some way or another.

Last season was a cruel hoax to rational thought. If we are going to be bad, dammit, let the indicators show that we are a bad team. If we are good, dammit, win games. When looking at the commonly accepted indicators, I could make a very, very convincing case that the 2008 Packers were the best "bad" team in the history of the NFL.

[sharpe1027]

The original statistics quoted have nothing to do with combackiness. It says, 4th Quarter QB Ratings When the Teams are Within 7. I missed the Within part the first four times I read it (unless Chevelle has edited it). Which is a statistical attempt to mean the game is in doubt and still tightly contested.

And good statistical numbers and bad W/L record indicates not that the stats are deceiving, but that there may be other factors.

Admittedly, I took the liberty of reading between the lines, but the Favre "nut huggers" comment suggested this was directed at the endless debate about Rodgers inability to close the deal in the 4th quarter. The posts following the original post provided further support for my interpretation, as others reached the same reasonable conclusion.

As for deceiving stats, I think you missed my point. The stats are not the problem, the lack of any logical/concrete goal is the issue. Proving or disproving a QB's comebackiness is not difficult because of the stats, but because it has no concrete meaning.

[swede]

If we are going to be bad, dammit, let the indicators show that we are a bad team. If we are good, dammit, win games. When looking at the commonly accepted indicators, I could make a very, very convincing case that the 2008 Packers were the best "bad" team in the history of the NFL.

A valid point. Losing a bunch games by 3 points or less contributed mightily to what you suggest.