Quarterback rating

[burmafrd]

Dr Z tore it apart in his column. And he is exactly right. A system designed in 1973 that HAS NOT been changed since. talk about stupid. Even the dumbest fan has figured out that 1973 and 2007 have absolutely no relation at except that NFL football is played .

http://sportsillustrated.cnn.com/2007/writers/dr_z/10/25/rating.system/index.html


I think I have a better system. Take percentage of completion passes X .4
YPA X 7. Add TDs and subtract INTs. And add a qualifier that the QB has to have at least averaged 20 attempts per game.
Right now that would put Brady at 119.7
Romo at 90.2
Garcia at 91.2
Peyton Manning at 89.9

Right now Brady is at 137.9, Manning is103.5, Garcia is 106.2 and Romo is95.6
My system rewards higher yds per attempt over completion percentage.

I would also throw out spikes, consider them non-attempts.
My system also would make it very difficult to get over 100. right now Brady is playing unbelievable- so he rates a score well over 100. But breaking 90 would also be very hard.

It would also penalize the dinkers with such an emphasis on YPA.
AND it would also penalize the INT kings.
I am also looking at maybe somehow adding a mark for getting first downs, and also completion % on third down and long that gets a first down.- more then 5 yds.
But that might make it a little too complicated.
What do you guys think?

 

[theogt]

I think I have a better system. Take percentage of completion passes X .4
YPA X 7. Add TDs and subtract INTs. And add a qualifier that the QB has to have at least averaged 20 attempts per game.

That's just a very simplified (also read as "dumbed down") version of the QB Rating formula.

 

[zrinkill]

Anyone have Troy's version?

 

[Tusan_Homichi]

Anyone have Troy's version?

Here is the Aikman Efficiency Ratings:

http://troyaikman.wordpress.com/category/aikman-efficiency-ratings/

But that's only for teams. I'm not sure where you'd find individual player rankings.

 

[theogt]

Quick summary of the article:

The QB rating is bad because:

1. Many people don't know how it works;
2. QBs are better today than in the past; and
3. A QB can have a high passer rating in a game, yet still lose the game.

 

[YosemiteSam]

Efficiency is what QBs should be rated on. As far as I'm concerned TDs should have nothing to do with it. Any completion could be a TD based on the location of the completion. QB John could be 5 for 50 with 5TDs a QB Bob could be 50 for 50 with zero TDs. INT's should count against them, but not if the ball bounced off the receiver first, but to add something like that to a rating makes it much more difficult to look back on unless it's all calculated at game time.

 

[kmd24]

Z's article reads like the ramblings of a bitter old man.

Here are his logical inconsistencies that had me chuckling:

• No one knows how the ratings are computed, yet QB's know how to game the system to pump up their rating for meeting contract incentives
• The rules have changed to make passing easier, and yet Z feels that the ratings should be changed so that the average rating matches that of the 70's.
• Acheiving a high rating and losing discredits the QB rating, yet nothing he seems to be proposing, such as recalibrating the rating to modern day averages, will reflect whether the game was won or lost.

There are indeed problems with the rating from a statistical sense, such as:

• completion percentage is confounded with the other three components of the rating (ypa, td %, and INT %)
• QB rating cannot differentiate between a great downfield passer and a passer that merely throws short passes to receivers that run well after the catch
• QB rating cannot effectively differentiate some single game performances
• most importantly, all categories are arbitrarily weighted the same regardless of their statistically determined impact on winning, losing, or even scoring.

However, making the QB rating more statistically correct, as I am advocating in the second list, would only make the statistic less accessible to the average fan, which seems to be a major sticking point for Z.

 

[kmd24]

Efficiency is what QBs should be rated on. As far as I'm concerned TDs should have nothing to do with it. Any completion could be a TD based on the location of the completion. QB John could be 5 for 50 with 5TDs a QB Bob could be 50 for 50 with zero TDs. INT's should count against them, but not if the ball bounced off the receiver first, but to add something like that to a rating makes it much more difficult to look back on unless it's all calculated at game time.

I disagree with your POV on TD passes. It's a lot harder to complete a TD pass than it is to complete a pass of equal length in virtually any other situation.

 

[YosemiteSam]

I disagree with your POV on TD passes. It's a lot harder to complete a TD pass than it is to complete a pass of equal length in virtually any other situation.

Yes, but a WR screen on your own 5 yard line that is ran 95 yards for a TD has nothing to do with a QB's skill. It's about WR's skill at running after the catch, yet a QB would get full credit for it.

 

[joseephuss]

The QB rating system is not perfect, but it doesn't have to be. No one uses it as the be all end all for judging QBs. Especially those in charge such as coaches and general managers. They use what they see on the field to go along with the QB rating. It is just a tool used to supplement other evaluation processes.

Most fans don't understand how it works. Why do they/I need to? All I really need to know is a high number is good and a low number is bad.

Football and the passing game has evolved since the QB rating was introduced. That makes it difficult to compare QBs of different eras, but not impossible. That is why it is just a tool because there should be a whole lot more that goes into comparing Johnny Unitas and Peyton Manning.

Unitas' career rating was 77.75 and Manning prior to this season was 94.38. Just looking at that and easy Manning was better, but there are many that had the great opportunity to watch Unitas play and what he brought to the NFL. They can make the argument as to who was better without even referencing QB rating.

They can also utilize it knowing just how much the QB ratings have gone up on average. In Unitas' era the NFL average rating was 66.13 and during Manning's tenure it was 79.34. Now you can make relative comparisons. Unitas' rating is 17.6% higher than the league average and Manning's is 19.0% higher than his contemporaries.

I think it is too difficult to go in and start subtracting interceptions because it went off a receivers hands. It starts becoming really subjective because you have to determine if the receiver truly had a good chance at catching the ball. All that kind of stuff gets averaged out in the end. There are plenty of passes that should have been intercepted that were dropped by defenders.

If anything I would like to see yards per completion and TDs per completion included into the equation. A guy may be having a lousy day at completing a high percentage of his passes, but if his completions go for big yards and TDs he is helping his team. I would gladly take 5 for 20 for 200 yards and 3 TDs over a 15 for 20 for 200 yards and 2 TDs. As completion percentage has gone up over the last few decades, yards per completion has gone down. Yards per attempt has remained fairly constant, but YPC has dropped approximately by 2.5 yards. It was more difficult to complete passes in previous eras, so each completion had to count and count big.

 

[dallasfaniac]

QB Ratings are laughable.

You could put Ryan Leaf on a team with Jerry Rice in his prime, Randy Moss when he's trying and Terrell Owens with a Sharpie in his sock and low and behold he could post a rating slightly over 100.

But then you put Favre out on the field with just 5 men blocking, Bledsoe and Testeverde running routes and he'd put up 90. Who's the better QB huh?

 

[theogt]

If anything I would like to see yards per completion and TDs per completion included into the equation. A guy may be having a lousy day at completing a high percentage of his passes, but if his completions go for big yards and TDs he is helping his team. I would gladly take 5 for 20 for 200 yards and 3 TDs over a 15 for 20 for 200 yards and 2 TDs. As completion percentage has gone up over the last few decades, yards per completion has gone down. Yards per attempt has remained fairly constant, but YPC has dropped approximately by 2.5 yards. It was more difficult to complete passes in previous eras, so each completion had to count and count big.

If you measured it by completion, the 5 for 20 guy would have a much, much higher rating than the 15 for 20 guy. That doesn't make much sense.

 

[InmanRoshi]

QB rating, like any state, is flawed, but flawed != useless. Any year you can look at the QB ratings and the good ones are usually at the top and the bad ones are usually at the bottom.

 

[joseephuss]

If you measured it by completion, the 5 for 20 guy would have a much, much higher rating than the 15 for 20 guy. That doesn't make much sense.

I didn't say just do it by completion, but to incorporate YPC into the equation along with the current criteria. That would help raise the rating of the guy who was 5 for 20 because his completions would have meant more and produced more.

In my example, the 5 for 20 guy produces a QB rating of 108.3 while the 15 for 20 guy produces a rating of 139.6. I am not saying the 5 for 20 guy should have the higher rating, but if you can somehow incorporate the YPC and TDs/C, then those ratings may get closer together.

 

[kmd24]

Yes, but a WR screen on your own 5 yard line that is ran 95 yards for a TD has nothing to do with a QB's skill. It's about WR's skill at running after the catch, yet a QB would get full credit for it.

Well, I made a similar point in an earlier post, but the QB rating is designed to apply to large numbers of passes, so the effect of such a play should be small, assuming that it isn't the norm for that particular QB.

Also, you should note that there are also negative effects - like drops, receivers falling down, tipped balls for INT's, hail mary INT's (much more frequent than hail mary TD's) - that affect the QB rating, and they have little to do with QB skill.

The assumption of the statistician is that these outliers will average out, though clearly there will often be a bias when comparing QB's due to variations in receiving corps.

 

[InmanRoshi]

I disagree with your POV on TD passes. It's a lot harder to complete a TD pass than it is to complete a pass of equal length in virtually any other situation.

I disagree with that. In no other situation does the defense go all out against the run than they do on the goalline. Plenty, if not the majority, of TD passes thrown inside the 5 yard line come off playaction boots where the defense bit on the run fake and left the TE sitting all alone in the back of the endzone.

 

[theogt]

I didn't say just do it by completion, but to incorporate YPC into the equation along with the current criteria. That would help raise the rating of the guy who was 5 for 20 because his completions would have meant more and produced more.

In my example, the 5 for 20 guy produces a QB rating of 108.3 while the 15 for 20 guy produces a rating of 139.6. I am not saying the 5 for 20 guy should have the higher rating, but if you can somehow incorporate the YPC and TDs/C, then those ratings may get closer together.

Good point.

 

[Doomsday101]

Game vs Buffalo really hurt Romo numbers and I doubt he will have another game like that anytime soon. I do agree the rating system they currently use is a bit flawed.

 

[AdamJT13]

I disagree with your POV on TD passes. It's a lot harder to complete a TD pass than it is to complete a pass of equal length in virtually any other situation.

But a 1-yard TD pass to a wide open tight end significantly raises your passer rating. In his prime, Troy Aikman threw barely more than one TD per game, while Emmitt Smith annually led the league in touchdowns because we mostly punched it into the end zone on the ground. Would Aikman have been a better passer if he had gotten four or five more short TD passes each season? No, but his rating would have been much higher.

Personally, I think the length of the passes -- how far downfield the ball is caught -- should be part of the equation, too. A screen pass that goes for 20 yards is not equal to a post pass that goes for 20.

 

[joseephuss]

But a 1-yard TD pass to a wide open tight end significantly raises your passer rating. In his prime, Troy Aikman threw barely more than one TD per game, while Emmitt Smith annually led the league in touchdowns because we mostly punched it into the end zone on the ground. Would Aikman have been a better passer if he had gotten four or five more short TD passes each season? No, but his rating would have been much higher.

Personally, I think the length of the passes -- not the length of completions -- should be part of the equation. A screen pass that goes for 20 yards is not equal to a post pass that goes for 20.

Yes, that would provide a more accurate representation. A little bit more difficult to incorporate, but can be done. There are systems out there that do rate QBs on more statistical criteria than the current accepted QB rating system. They are even more difficult to understand and are not referenced very often, but probably are better systems.