Quarterback Rating

[Huckleberry]

NFL Passer Rating rankings:

Rk  NAME               RATING
1   D. Brees           101.20
2   C. Palmer           98.70
3   D. Huard            97.60
4   T. Romo             96.60
5   P. Rivers           96.30
6   P. Manning          95.90
7   D. McNabb           95.50
8   M. Bulger           90.30
9   D. Carr             87.50
10  M. Brunell          86.50
11  T. Brady            84.90
12  J. Losman           84.80
13  S. McNair           83.10
14  M. Hasselbeck       81.60
15  E. Manning          80.30
16  C. Pennington       79.60
17  J. Delhomme         79.40
18  B. Leftwich         79.00
19  B. Favre            78.00
20  J. Kitna            77.30
21  B. Roethlisberger   74.90
22  R. Grossman         74.50
23  A. Smith            74.00
24  M. Leinart          73.90
25  M. Vick             73.60
26  J. Harrington       72.90
27  C. Frye             72.00
28  B. Johnson          71.00
29  J. Plummer          70.50
30  B. Gradkowski       66.30
31  V. Young            65.70
32  A. Walter           55.00

I added runs as pass attempts and completions with appropriate yardage, rushing TDs to the total, sacks as incomplete passes with appropriate negative yardage, and fumbles lost to INTs to come up with a total turnover number. The resulting "QB Rating" rankings:

Rk  NAME               QB RAT
1   D. Brees            96.59
2   P. Manning          94.43
3   D. McNabb           91.73
4   T. Romo             91.64
5   C. Palmer           90.39
6   P. Rivers           89.80
7   D. Huard            88.35
8   S. McNair           80.81
9   M. Brunell          80.50
10  B. Leftwich         79.50
11  M. Bulger           78.68
12  T. Brady            76.87
13  M. Vick             75.93
14  D. Carr             75.55
15  E. Manning          74.12
16  V. Young            73.10
17  C. Pennington       72.03
18  B. Favre            71.89
19  M. Hasselbeck       71.59
20  J. Delhomme         70.55
21  J. Losman           70.43
22  M. Leinart          69.51
23  J. Harrington       69.18
24  R. Grossman         67.80
25  J. Plummer          67.38
26  A. Smith            66.50
27  B. Roethlisberger   66.46
28  J. Kitna            65.50
29  B. Johnson          64.51
30  C. Frye             62.39
31  B. Gradkowski       58.12
32  A. Walter           35.10

Effect of the new stat compared to basic rating:

Rk   NAME             Effect
1   V. Young            7.40
2   M. Vick             2.33
3   B. Leftwich         0.50
4   P. Manning         -1.47
5   S. McNair          -2.29
6   J. Plummer         -3.12
7   J. Harrington      -3.72
8   D. McNabb          -3.77
9   M. Leinart         -4.39
10  D. Brees           -4.61
11  T. Romo            -4.96
12  M. Brunell         -6.00
13  B. Favre           -6.11
14  E. Manning         -6.18
15  B. Johnson         -6.49
16  P. Rivers          -6.50
17  R. Grossman        -6.70
18  A. Smith           -7.50
19  C. Pennington      -7.57
20  T. Brady           -8.03
21  B. Gradkowski      -8.18
22  C. Palmer          -8.31
23  B. Roethlisberger  -8.44
24  J. Delhomme        -8.85
25  D. Huard           -9.25
26  C. Frye            -9.61
27  M. Hasselbeck     -10.01
28  M. Bulger         -11.62
29  J. Kitna          -11.80
30  D. Carr           -11.95
31  J. Losman         -14.37
32  A. Walter         -19.90

As you can see, running well isn't the only way to do well in the stat. It's certainly the biggest factor, but Peyton Manning does well relative to his peers because he gets rid of the ball quickly, doesn't take sacks, and doesn't fumble. Obviously running well is the only way to get a good positive number, but being a pocket passer doesn't mean you have to take a huge hit. The guys at the bottom of the list have fumbled frequently and, of course, taken a lot of sacks. Of course, this effect stat is skewed by their passing skill. A player with a 90 passer rating will have a more negative effect number for the same running stats as a passer with a 70 passer rating. So here's the Running Rating:

Rk   NAME            Run Rat
1   B. Leftwich        95.47
2   P. Manning         95.31
3   V. Young           95.05
4   M. Vick            81.22
5   D. McNabb          75.90
6   S. McNair          73.88
7   T. Romo            71.73
8   P. Rivers          66.50
9   J. Plummer         60.33
10  M. Leinart         53.94
11  J. Harrington      49.53
12  C. Palmer          48.90
13  A. Smith           39.17
14  M. Brunell         38.89
15  D. Brees           38.39
16  C. Frye            38.11
17  T. Brady           37.18
18  E. Manning         36.41
19  B. Roethlisberger  35.05
20  B. Johnson         34.28
21  D. Carr            33.89
22  C. Pennington      32.56
23  B. Favre           29.05
24  B. Gradkowski      28.91
25  D. Huard           27.92
26  R. Grossman        25.93
27  J. Losman          24.86
28  M. Hasselbeck      21.34
29  M. Bulger          18.40
30  J. Kitna            9.52
31  J. Delhomme         8.78
32  A. Walter           0.00

At first it seems odd that Leftwich and Manning in particular are at the top of the list. But you have to consider that the NFL rating formula does not give negative values. If you're negative, it's a zero for that category. So this doesn't reflect the real impact of running on total value. So I removed the zero constraint (giving most of the QBs negative components for yardage and some of them negative for turnovers, i.e., fumbles lost). I also removed the 2.375 maximum constraint as they were being calculated simultaneously. The only thing this affected was Vince Young's "running completion percentage" value as his ratio of runs to sacks is the highest:

Rk   NAME            Run Rat*
1   V. Young           98.11
2   B. Leftwich        82.11
3   M. Vick            81.22
4   P. Manning         73.70
5   D. McNabb          69.06
6   S. McNair          63.38
7   T. Romo            56.73
8   P. Rivers          49.08
9   J. Plummer         48.00
10  M. Leinart         29.40
11  J. Harrington      26.01
12  C. Palmer          23.00
13  A. Smith           22.78
14  C. Frye            21.94
15  D. Carr            20.58
16  T. Brady           18.40
17  D. Brees           18.01
18  B. Gradkowski      17.71
19  M. Brunell         15.28
20  B. Johnson         14.96
21  C. Pennington      14.67
22  E. Manning         11.11
23  B. Roethlisberger   8.00
24  J. Losman           3.17
25  D. Huard           -0.75
26  M. Hasselbeck      -1.18
27  R. Grossman        -4.51
28  B. Favre           -8.28
29  J. Kitna          -16.19
30  M. Bulger         -19.74
31  J. Delhomme       -26.30
32  A. Walter         -44.05

I just put this up for discussion after somebody on another board asked me to calculate this. Obviously they were a Vince Young fan and I, of course, am as well. But the one thing I noticed is that Andrew Walter has, to put it mildly, struggled this year. Sticking a young guy in behind that OL may not have been wise. I thought I remembered his being fairly mobile in college (a shade less than Plummer?) so I'd be interested in hearing thoughts from Raiders fans.

[dixieflatline]

This is some seriously good analysis Huck. The problem is that QB rating is a horrible metric. Adjusting to add the QB's rushing yards helps but even then the system has some big flaws. The "running completion percentage" though is very interesting. I would love to see how a QB's numbers change as he ages. Do they gradually decline or do QBs tend to fall off a cliff as he gets older?

[Easy Mac]

So you're saying Andrew Walter really is a bad QB?

[dawgfan]

Much like the standard QB rating formula, it's really more effective as a team measurement since the quality of one's offensive line plays a big part in your running effectiveness portion (especially the number of sacks taken).

Hasselbeck is a fairly mobile QB, but he's taking a lot more sacks this year due to a much weaker O-line in front of him.

[Huckleberry]

Originally Posted by dixieflatline

This is some seriously good analysis Huck. The problem is that QB rating is a horrible metric. Adjusting to add the QB's rushing yards helps but even then the system has some big flaws. The "running completion percentage" though is very interesting. I would love to see how a QB's numbers change as he ages. Do they gradually decline or do QBs tend to fall off a cliff as he gets older?

I agree that the NFL formula is a horrible metric. I believe completion percentage is overemphasized and yards per attempt is underemphasized. I do, though, agree that the higher negative impact of interceptions as compared to the college rating is appropriate. Each possession is more valuable in the NFL. Whether or not the degree of emphasis is right I don't know.

I think the mobile QBs that age the best are the ones that are able to still get out of sacks by being quick enough to get out of the pocket and throw it away or else being able to quickly identify dumpoff receivers. The numbers indicate that Steve McNair does a great job of this.

I'd like to see the NFL formula modified. I don't like the 0-2.375 constraint although that usually only applies in games, not over seasons. But I also mentioned above my real beef with it. David Carr has the lowest yards per completion number in the league but it doesn't hurt so he's still in the Top 10 in passer rating. There's no way he's Top 10 in helping his team win.

[Raiders Army]

It's funny but we were having a discussion at work this morning about QB ratings. They don't make any sense. College and pro formulas are different. You can't do them in your head. You can incomplete passes but still have a perfect rating?

[Huckleberry]

Originally Posted by dawgfan

Much like the standard QB rating formula, it's really more effective as a team measurement since the quality of one's offensive line plays a big part in your running effectiveness portion (especially the number of sacks taken).

Hasselbeck is a fairly mobile QB, but he's taking a lot more sacks this year due to a much weaker O-line in front of him.

I'm not sure how that differentiates it from any football statistic.

Would LaDainian Tomlinson have 29 touchdowns on the Raiders?

[dawgfan]

Originally Posted by Huckleberry

I'm not sure how that differentiates it from any football statistic.

It doesn't. Football statistics are much harder to parse out as indicative of individual achievement than baseball statistics - that's just the nature of the sport.

I'm just pointing out that you have to be a little careful when looking at these stats in putting too much emphasis on what they suggest about an individual player.

[dixieflatline]

Originally Posted by Huckleberry

I do, though, agree that the higher negative impact of interceptions as compared to the college rating is appropriate. Each possession is more valuable in the NFL. Whether or not the degree of emphasis is right I don't know.

I completely agree with everything you said except this part. Are there fewer possessions in the NFL per game? This seems unlikely especially with the new clock running rules college has put in (and no 2 minute warning). It is possible that you are right about this too but I hadn't thought about this and am curious now.

Pulling an individual's performance from the team's performance is very hard to do in football but I think people are getting closer. The team that can accurately do this first will get a huge boost.

[jbergey22]

Originally Posted by dixieflatline

I completely agree with everything you said except this part. Are there fewer possessions in the NFL per game? This seems unlikely especially with the new clock running rules college has put in (and no 2 minute warning). It is possible that you are right about this too but I hadn't thought about this and am curious now.

Pulling an individual's performance from the team's performance is very hard to do in football but I think people are getting closer. The team that can accurately do this first will get a huge boost.

Much fewer in the NFL...Clock runs on out of bounds plays, first downs unlike college.

[JeffW]

Huckleberry,

Check out:

http://www.footballoutsiders.com/stats/qb.php

If you haven't already. It's a comprehensive rating system for QBs that measures their value over a replacement level QB.

[EagleFan]

Originally Posted by dixieflatline

I completely agree with everything you said except this part. Are there fewer possessions in the NFL per game? This seems unlikely especially with the new clock running rules college has put in (and no 2 minute warning). It is possible that you are right about this too but I hadn't thought about this and am curious now.

Pulling an individual's performance from the team's performance is very hard to do in football but I think people are getting closer. The team that can accurately do this first will get a huge boost.

Yes. The clock does not stop as often as in college, plus the play clock for college is quicker.

[Huckleberry]

Originally Posted by JeffW

Huckleberry,

Check out:

http://www.footballoutsiders.com/stats/qb.php

If you haven't already. It's a comprehensive rating system for QBs that measures their value over a replacement level QB.

I like their thought process but they don't include positive rushing yards and do include all fumbles. Bad passes that a defender drops aren't counted as interceptions, so I don't see why fumbles that are retained by the offense should count against you.

I redid the passer rating formula based on emphasizing yards per attempt at the expense of completion percentage and slightly magnifying the negative effect of INTs. A possession lost in the NFL is a big deal. Results:

1   Huard            100.89
2   Brees            100.88
3   McNabb           100.15
4   Palmer            99.68
5   Rivers            97.28
6   P. Manning        96.72
7   Romo              95.88
8   Bulger            90.79
9   Brady             87.16
10  Brunell           86.41
11  Hasselbeck        86.02
12  Losman            84.87
13  Carr              84.03
14  E. Manning        83.94
15  McNair            82.79
16  Leftwich          81.10
17  Delhomme          81.05
18  Favre             80.46
19  Vick              79.80
20  Grossman          78.91
21  Pennington        77.81
22  Leinart           75.30
23  Smith             74.77
24  Kitna             74.58
25  Harrington        73.62
26  Roethlisberger    73.38
27  Plummer           72.59
28  Gradkowski        69.87
29  Young             69.68
30  Frye              69.08
31  Johnson           68.18
32  Walter            56.55

Plugging in the running (sacks, runs, fumbles) data, the QB rating looks like this:

1   McNabb           97.19
2   Brees            96.79
3   P. Manning       96.68
4   Palmer           92.95
5   Huard            92.74
6   Rivers           91.27
7   Romo             91.18
8   Leftwich         82.59
9   Bulger           81.18
10  Brunell          81.04
11  McNair           80.53
12  Brady            78.82
13  Vick             78.18
14  E. Manning       78.14
15  Hasselbeck       76.96
16  Young            75.86
17  Favre            74.29
18  Carr             73.20
19  Delhomme         72.68
20  Leinart          72.23
21  Losman           71.71
22  Grossman         71.59
23  Pennington       70.96
24  Harrington       70.21
25  Plummer          69.35
26  Smith            67.77
27  Roethlisberger   67.41
28  Kitna            65.10
29  Johnson          62.71
30  Frye             61.86
31  Gradkowski       60.15
32  Walter           38.67

[JeffW]

They have a second section for rushing.

Re: fumbles. There is a ton of statistical evidence for fumble recovery rates, etc, so it's reasonable to count fumbles as partially lost regardless of the actual outcome of the play. It's not as clear cut with interceptions and you can't separate near interceptions form incompletes without analyzing every passplay.

[Passacaglia]

A possession lost in the NFL is a big deal.

I just want to chime in to agree with this, and to bitch about how when I made fumbles and interceptions -6 points in my fantasy league, everyone cried to holy hell about it.

[stevew]

Originally Posted by Raiders Army

It's funny but we were having a discussion at work this morning about QB ratings. They don't make any sense. College and pro formulas are different. You can't do them in your head. You can incomplete passes but still have a perfect rating?

Pro rating is not horribly hard to guestimate in your head, once you know the ranges. It's just 4 values with a ceiling limit added together. At some point you go over the max for the rating, and that just means you get the max for that quarter.

[JeffW]

Originally Posted by Passacaglia

I just want to chime in to agree with this, and to bitch about how when I made fumbles and interceptions -6 points in my fantasy league, everyone cried to holy hell about it.

Yeah, well, fantasy football doesn't remotely mirror NFL football anyway.

[Passacaglia]

Originally Posted by JeffW

Yeah, well, fantasy football doesn't remotely mirror NFL football anyway.

Course not, but you can't blame me for trying...unless you're in my league, I guess.