Are people looking at the wrong stats when assessing QBs - can we come up with something better?

[K.O. Johnny]

Now hear me out folks - I'm not saying I'm right, just want other people's take.

On one of the Cooley threads Zoony who according to his posting record has been around, had this to say:

"As an aside, I think "TDs" is the most over-rated stat in all of football. Why should a QB's rating shoot way up if he completes a pass in the paint rather than at the 3 yard line? What difference does it make in terms of how good a player the QB is? Same with receivers and RBs."

Are we looking at the wrong stats in evaluating QBs, because I agree with the above quote. It's up there with the win/loss ratio and I think it skews passer rating. It's like looking at average gain per carry of a back - and i'll wager off the top of my head that Riggo's is probably the lowest of HOF running backs in the SB era.

Is there a different set of figures we could come up with to grade QBs - AtlantaSkinsFan, I'm calling you in particular cos you like this kind of stuff.

For instance: I'd like to see a combination of say (yards per pass x comp%)divided by interception% with maybe 3rd down conversion coloring the final result.

But the way passer rating treats TD and ints is a slightly peculiar way http://en.wikipedia.org/wiki/Passer_rating#NFL

I know there's more to QB play than rating but unless you can watch game film and grade each play versus defense/ field position/score/time left, we gotta go with numbers

[Lombardi's_kid_brother]

Yards per attempt and TD/Int ratio are probably the two most important stats to look at. I don't know how you would combine them into some kind of OPS thing though.

---------- Post added January-5th-2011 at 12:46 PM ----------

To take this a step further, the top five in yards per attempt this year are Rivers, Rodgers, Roethlisberger, Vick and Brady. If you had to name the five best QBs this year, you would almost certainly name 4 of them and Roethlisberger is perpetually underrated.

The TD/Int Ratio for them.

Rivers: 30/13

Rodgers: 28/11

Roethlisberger: 17/5

Vick: 21/6

Brady: 36/4

So....a combination of the two would probably end up with Brady and Vick #1 and #2. And that would almost certainly be the proper way to rank the QBs this year.

It would also do what a stat should do: Prove what you already know and provide an insight that you do not know. You knew Vick and Brady had the two best years. You did not necessarily know how great Roethlisberger really is.

(Completion percentage should probably be factored in since it is noteworthy. But everyone who is worthwhile completes between 61 and 64 percent of their passes. I'm not sure you want to over-emphasize something that may be a lot of noise. At the same time, it would penalize Jason Campbell, which is always fun).

---------- Post added January-5th-2011 at 12:50 PM ----------

By the way, McNabb was 15th in Average per Attempt, but his TD/INT ratio was 14/15. So, yea...he stunk. And Drew Brees was shockingly mediocre this year - which if you had him in fantasy (like I did) you already knew.

[K.O. Johnny]

I just think the TD/INT stat can reflect more the offensive philosophy of the team than the true skill of the QB. Do they stretch it long ball and then dink it in the red zone and try and punch it in with a big nasty fullback - that would skew a qb's stats down, when he might have managed an great 70 yard drive with some critical 3rd downs

[darrelgreenie]

(I swear we have this thread/conversation every year).

DVOA isn't bad but it too has its flaws.

I agree and think YPA is useful but its a better measure of the entire passing game then it is of the QB alone.

YPA is effected by scheme deep passing WCO offense, Coryell type offense vs a short WCO or a short passing Erhardt+Perkins offense.iIts effected by quality of the receivers WR, TE and RB

YPA is also effected by the quality of the OL which is linked to the scheme b/c if you can't pass protect for 5 and 7 step drops you can't run a downfield passing game.

If we had to rely only on stat i would want a stat that to basically do the impossible:

o quantify the quality of pass protection-Sacks and Hurries

o factors in the quality of the passes attempted w/ different values assigned for the distance downfield

+ QB rating + DVOA

[Destructis]

Wins/Losses and offensive points scored.

[K.O. Johnny]

(I swear we have this thread/conversation every year).

Sorry - I'm just a rookie

I think you're right concerning quality of pass protection - very, very critical

[Lombardi's_kid_brother]

Yes, a short passing game versus a long passing game will impact YPA. But it will also make up for it in more receptions. Brady was not exactly heaving the ball downfield this year like he was in 2007. Yet, his YPA was fifth in the league and the second highest he ever put up (behind that insane 2007 campaign).

[K.O. Johnny]

Wins/Losses and offensive points scored.

The wins/losses is simplistic and takes absolutely no account of the team's defense.

[Lombardi's_kid_brother]

Sorry - I'm just a rookie

I think you're right concerning quality of pass protection - very, very critical

You need isolate pass protection and take it out of the equation as much as possible. You are tring to measure the individual abilities of the QB - not the team's passing game.

Roethlisberger plays behind a terrible O-line. Yet his numbers are great.

The Giants have a decent o-line. Yet Eli threw a million picks this year.

---------- Post added January-5th-2011 at 01:06 PM ----------

Wins/Losses and offensive points scored.

Ok. You are no longer allowed to discuss this, Joe Morgan.

[K.O. Johnny]

You need isolate pass protection and take it out of the equation as much as possible. You are tring to measure the individual abilities of the QB - not the team's passing game.

Roethlisberger plays behind a terrible O-line. Yet his numbers are great.

The Giants have a decent o-line. Yet Eli threw a million picks this year.

---------- Post added January-5th-2011 at 01:06 PM ----------

Okay - now we're getting somewhere

And you also need to take into account the running game

[Lombardi's_kid_brother]

Okay - now we're getting somewhere

And you also need to take into account the running game

Why?

[K.O. Johnny]

Why?

Say you have a lousy run game - pass rushers can come hard and the DBs can back off the LOS slightly. Okay, that's put in a very simplistic manner, but I remember the saying "those who live by the pass will die by the pass" - it may not seem as pertinent in today's pass happy NFL but I think you forget that at your peril

[Lombardi's_kid_brother]

You know, if you just added YPA to TD/Int Ratio, the rankings would be:

Brady: 16.9

Vick: 11.6

Roethlisberger: 11.6

Freeman: 11.5

Rivers: 11

Cassell: 10.8

Rodgers: 10.6

Flacco: 9.9

Schuab: 9.6

Ryan: 9.6

Orton: 9.5

Cutler: 9

Garrard: 9

[Destructis]

The wins/losses is simplistic and takes absolutely no account of the team's defense.

QB's are like Head Coaches. They are judged on wins/losses. Can you tell me Brady's TD/Int ratio? How about his yards per attempt? WITHOUT looking it up. Now can you tell me how many SB's he has won?

It's all about the wins. Nothing else matters.

[Oldfan]

...Is there a different set of figures we could come up with to grade QBs - AtlantaSkinsFan, I'm calling you in particular cos you like this kind of stuff...

What you're looking for doesn't exist. It's the Holy Grail for statisticians interested in Football. ASF's original projections for QBs was a variation of the Lewin Forecast which is still promoted on the Football Outsiders website. Not many statisticians take it seriously.

Here's the problem in a nutshell. Let's say that QB-A and QB-B both have equal talent potential. However, QB-A plays with support that allows him to perform at 90% of his potential while QB-B plays with support that allows him only to play to 50% of his potential. There's no way in hell you could possibly know that these two QBs have equal talent by grading their performances.

The only intelligent way to grade QBs is to do it as a scout might, by grading their tangibles. As to the intangibles, these are difficult to grade because they're intangible. Mostly, people hype or trash a QB on the intangibles because they can't be proven wrong.

[K.O. Johnny]

QB's are like Head Coaches. They are judged on wins/losses. Can you tell me Brady's TD/Int ratio? How about his yards per attempt? WITHOUT looking it up. Now can you tell me how many SB's he has won?

It's all about the wins. Nothing else matters.

Good job you weren't GM at San Fran when Tampa traded steve Young.

Hey, were you the GM at Tampa per chance?

[Lombardi's_kid_brother]

QB's are like Head Coaches. They are judged on wins/losses. Can you tell me Brady's TD/Int ratio? How about his yards per attempt? WITHOUT looking it up. Now can you tell me how many SB's he has won?

It's all about the wins. Nothing else matters.

I knew his TD/Int ratio.

Also, he won his first two Super Bowls because of his kicker. So, he can blow me.

[K.O. Johnny]

What you're looking for doesn't exist. It's the Holy Grail for statisticians interested in Football. ASF's original projections for QBs was a variation of the Lewin Forecast which is still promoted on the Football Outsiders website. Not many statisticians take it seriously.

Here's the problem in a nutshell. Let's say that QB-A and QB-B both have equal talent potential. However, QB-A plays with support that allows him to perform at 90% of his potential while QB-B plays with support that allows him only to play to 50% of his potential. There's no way in hell you could possibly know that these two QBs have equal talent by grading their performances.

The only intelligent way to grade QBs is to do it as a scout might, by grading their tangibles. As to the intangibles, these are difficult to grade because they're intangible. Mostly, people hype or trash a QB on the intangibles because they can't be proven wrong.

So all this give up the house for Luck - or anyone else for that matter is nonsense, as the whole thing is a crapshoot then? Both tangibles and intangables

[Lombardi's_kid_brother]

What you're looking for doesn't exist. It's the Holy Grail for statisticians interested in Football. ASF's original projections for QBs was a variation of the Lewin Forecast which is still promoted on the Football Outsiders website. Not many statisticians take it seriously.

Here's the problem in a nutshell. Let's say that QB-A and QB-B both have equal talent potential. However, QB-A plays with support that allows him to perform at 90% of his potential while QB-B plays with support that allows him only to play to 50% of his potential. There's no way in hell you could possibly know that these two QBs have equal talent by grading their performances.

The only intelligent way to grade QBs is to do it as a scout might, by grading their tangibles. As to the intangibles, these are difficult to grade because they're intangible. Mostly, people hype or trash a QB on the intangibles because they can't be proven wrong.

I think you can look at 3 or 4 numbers side by side and use something of a balancing test. I could probably quantify if I wanted to. I ignore intangibles.

I think a good pro quarterback in 2010 has to have a completion percentage over 60 percent. But I don't think completion percentage is that important once 60 percent is reached. So, I wouldn't build a model where it played a big factor except as a penalty for under 60 percent.

I know you still like Campbell, but I think it's interesting that his best year for completion percentage is the year where he was sacked the most. So, I don't think "pass rush" or "offensive line protection" or whatever necessarily matters that much. (Roethlisberger's best year for completion percentage was also the year he was sacked the most).

[The Robert Griffin Experience]

Then why do quarterbacks with seemingly high tangibles bust? Guys like Russell, Leaf, even Shuler? Also, what factors are "tangible versus non-tangible"? It's possible scouts overrated the tangible abilities of high-profile QB busts, or that they were missing something like accuracy or awareness, but if tangibles are all that matter, then what tangibles did the big QB busts lack that kept them from being successful?

[Lombardi's_kid_brother]

So all this give up the house for Luck - or anyone else for that matter is nonsense, as the whole thing is a crapshoot then? Both tangibles and intangables

Oldfan has a tendency to digest statistical models and then somehow use them to make pronouncements on things that cannot be proven. It's a weird posting style.

---------- Post added January-5th-2011 at 01:50 PM ----------

Then why do quarterbacks with seemingly high tangibles bust? Guys like Russell, Leaf, even Shuler? Also, what factors are "tangible versus non-tangible"? It's possible scouts overrated the tangible abilities of high-profile QB busts, or that they were missing something like accuracy or awareness, but if tangibles are all that matter, then what tangibles did the big QB busts lack that kept them from being successful?

If I was drafting a QB out of college, I would look at two numbers first and foremost. His college completion percentage and his IQ. The two biggest "intangibles" that kill college QBs are stupidity and laziness. And sometimes those go away after a while.

[darrelgreenie]

Sorry - I'm just a rookie

I think you're right concerning quality of pass protection - very, very critical

No apologizes needed.

If anything i'm at fault for not being more clear w/ my tone which can get lost on the internet.

Even if something similar has been discussed i'm all ears for a good discussion.

HTTR and welcome to the forum.

[Oldfan]

So all this give up the house for Luck - or anyone else for that matter is nonsense, as the whole thing is a crapshoot then? Both tangibles and intangables

No. I told you that grading QBs the way scouts do it is the only intelligent way. If the scout knows his stuff, he grades what he can see. However, he would not use stats like completion pct to grade accuracy because then he'd be grading the QB and about seven things that have nothing to do with the QB. He would grade accuracy by watching every throw on film. As for the intangibles, you'd want to eliminate players with the IQ of a grapefruit or was a convicted rapist, you know -- due diligence.

---------- Post added January-5th-2011 at 01:02 PM ----------

I think you can look at 3 or 4 numbers side by side and use something of a balancing test. I could probably quantify if I wanted to. I ignore intangibles.

I think a good pro quarterback in 2010 has to have a completion percentage over 60 percent. But I don't think completion percentage is that important once 60 percent is reached. So' date=' I wouldn't build a model where it played a big factor except as a penalty for under 60 percent.[/quote']A good youth coach could take a 12 year-old QB and give him a 60% completion rate just by calling passes he can complete 60% of the time.

I know you still like Campbell,

I projected Jason to fail when I first saw his mechanics. Al Saunders and Jim Zorn saved his career by reworking his mechanics. I gave him a C minus in 2009.

[Lombardi's_kid_brother]

Peyon Manning was 63 percent passer in college.

Ryan Leaf was a 56 percent passer in college.

Tom Brady was a 63 percent passer in college.

Akili Smith was a 55 percent passer in college.

Ben Roethlisberger was a 64 percent passer in college.

It's not an end-al be-all thing since Jamarcus Russell was something like a 67 percent passer (but I have no idea how to judge these spread option QBs).

But I would never draft a college QB who did not hit 60 percent consistently.

[Oldfan]

Oldfan has a tendency to digest statistical models and then somehow use them to make pronouncements on things that cannot be proven. It's a weird posting style.

Here's what I said. Why don't you try finding fault with it rather than making your sweeping generalizations. You know' date=' like real debate?

What you're looking for doesn't exist. It's the Holy Grail for statisticians interested in Football. ASF's original projections for QBs was a variation of the Lewin Forecast which is still promoted on the Football Outsiders website. Not many statisticians take it seriously.

Here's the problem in a nutshell. Let's say that QB-A and QB-B both have equal talent potential. However, QB-A plays with support that allows him to perform at 90% of his potential while QB-B plays with support that allows him only to play to 50% of his potential. There's no way in hell you could possibly know that these two QBs have equal talent by grading their performances.

The only intelligent way to grade QBs is to do it as a scout might, by grading their tangibles. As to the intangibles, these are difficult to grade because they're intangible. Mostly, people hype or trash a QB on the intangibles because they can't be proven wrong.