ESPN.com: Finding context in the Stafford debate (intriguing J. Campbell reference)

[Shilsu]

Exactly, you're never going to have a perfect formula, but this one does show a strong correlation with NFL success.

It doesn't show much correlation at all.

Come up with your own abritrary system of grading each player's impact in the NFL and then plot it against the score that ESPN Research gave them.

[Thinking Skins]

Exactly, you're never going to have a perfect formula, but this one does show a strong correlation with NFL success. Remember these are all 1st round guys, so all are thought to have the potential to be elite.

C'mon. Look at how small the sample size is. Its about 30 players. Thats not enough to say this is an efficient formula. It predicts that Leinart is gonna be the best QB ever. And has Manning somewhere like 4th or 5th.

It'd be more interesting to see how this formula looks over all QBs taken in the draft. I wonder if this same correlation holds up. Probably not.

[HBnotBlades]

It'd be more interesting to see how this formula looks over all QBs taken in the draft. I wonder if this same correlation holds up. Probably not.

I completely agree that it wouldn't hold up over the entire draft. The reason this works is that it's comparing all 1st rounders and most (not all) of those guys came from top tier programs in big time conferences. If you start including the latter rounds you get more guys who played in smaller conferences or maybe even D1AA and that significantly skews the results. I would like to see it over the last 20 years though, that would be interesting.

But can you really say there's no correlation? Of the 8 players that have made the pro-bowl, 6 are in that top group. When you consider that Vick only made it on his running ability, that makes 6 of 7. In addition, there are 4 players on that list that have played in super bowls and all 4 or them are in that top category. It's not perfect, but you can't completely discount it.

[Riggo-toni]

ESPN Research proves once again that they are worthless.

Really!! 3 of their top 4 are already busts (okay, so aybe the jury is still out on Leinart).

[Thinking Skins]

I completely agree that it wouldn't hold up over the entire draft. The reason this works is that it's comparing all 1st rounders and most (not all) of those guys came from top tier programs in big time conferences. If you start including the latter rounds you get more guys who played in smaller conferences or maybe even D1AA and that significantly skews the results. I would like to see it over the last 20 years though, that would be interesting.

Extend it to more than 31 people and see how it works. Look at QBs taken over the first 2 rounds, or the first 3 rounds, or all QBs taken from major schools, or even QBs taken over a longer period of time. Its a BS formula and I bet the results will not hold up over a larger sample size.

But can you really say there's no correlation? Of the 8 players that have made the pro-bowl, 6 are in that top group. When you consider that Vick only made it on his running ability, that makes 6 of 7. In addition, there are 4 players on that list that have played in super bowls and all 4 or them are in that top category. It's not perfect, but you can't completely discount it.

Even in this thing its saying that Leinart is the best QB drafted in this span, and Vick amongst the worse? 3 of the top 4 QBs on that list have been called busts (Leinart, Couch, Smith). 2 of the next 4 are either busts or in "show me something positions" (Campbell, Leftwich, and an argument can be made that Rodgers is in a similar position), then the next 4 has a bust (Carr), two decent QBs (Culpepper and Pennington), and an elite QB (Rothlesburger). Then there's Flacco, Manning and Mcnabb.

You call that correlation? In their strong likelihood for success, they're right on

- McNabb

- Manning

- Roethlesburger

- Manning

- Rodgers (maybe)

- Rivers

and wrong on

- Leinart

- Couch

- Smith

- Campbell

- Leftwitch

- Pennington

- Culpepper

- Carr

I mean, thats batting .500 at best.

[THE HAMMER'IN HOG]

If you look at the list it basically indicates that their research and findings tell us nothing about future success. Alex Smith rated higher than Peyton? Meaningless!

Another joke of a formula for individuals who can't actually look at a player and gauge wether or not he can play at the next level. Any one who would take this approach has no buisness as a GM. This test does not gauge circumstances in which these numbers are derived, and that is the biggest gauge for success in the NFL, the type of players you have faced, compared to the type of talent you play with.

[SKIN4WAHOOZ]

If I had it on paper, I would consider it to be good......... toilet paper material! :pooh::pooh:

[Stadium-Armory]

I wouldn't say their analysis is worthless. Or foolproof. But its a good attempt at creating an analytical basis for predicting future sucess. Is it 100% spot on? Of course not, but it gives decision makers another tangible data point to use when making a decision. And just a gut feeling when looking at this chart tells me that its 'close enough' to be worth consideration.

[Stadium-Armory]

Another joke of a formula for individuals who can't actually look at a player and gauge wether or not he can play at the next level. Any one who would take this approach has no buisness as a GM. This test does not gauge circumstances in which these numbers are derived' date=' and that is the biggest gauge for success in the NFL, the type of players you have faced, compared to the type of talent you play with.[/quote']

I think this sort of thing is meant to be used in conjunction with traditional analysis as you describe. Which, as history shows, is hard. Not very many people can spot talent consistently using the 'eyeball' method. Maybe a handful of guys are good at it? 5 maybe? We certainly don't have one here..

[HBnotBlades]

I mean, that's batting .500 at best.

Exactly. So out of all 31 1st round QB's choosen since 1997, we have 8 total that because pro bowlers which gives you a 25.8% chance of of making a correct selection. Once again, remember these are all 1st rounders and every single one of them, from Peyton Manning to Akili Smith, was thought to be the future of the franchise. If you decided you would only select a 1st round QB if their score was higher than 20 on this metric, you would literally DOUBLE your chances of finding a successful QB with a 1st round pick. That's significant.

I also think you've been slightly missing what the formula shows, it's better at weeding out busts that correctly identifying future stars. Look at the bust group:

Akili Smith

Cade McNown

Joey Harrington

Michael Vick

Ryan Leaf

Jim Drunkenmiller

Kyle Boller

It's batting 1.000 on bust identification. That doesn't mean that every single bust is correctly identified, only that if one is identified, it's actually a bust. Thereby eliminating a whole chunk of terrible draft picks.

[ZRagone]

So anyone done the workup on Colt yet?

[HBnotBlades]

So anyone done the workup on Colt yet?

COLTS_SCORE= 0.5*(32) + 2.5*(70.39-60) + 5*3.11

COLTS_SCORE = 57.5250

However, there would be a red flag on him because he played in a gimmicky offense much like Alex Smith.

[Thinking Skins]

Exactly. So out of all 31 1st round QB's choosen since 1997, we have 8 total that because pro bowlers which gives you a 25.8% chance of of making a correct selection. Once again, remember these are all 1st rounders and every single one of them, from Peyton Manning to Akili Smith, was thought to be the future of the franchise. If you decided you would only select a 1st round QB if their score was higher than 20 on this metric, you would literally DOUBLE your chances of finding a successful QB with a 1st round pick. That's significant.

I also think you've been slightly missing what the formula shows, it's better at weeding out busts that correctly identifying future stars. Look at the bust group:

Akili Smith

Cade McNown

Joey Harrington

Michael Vick

Ryan Leaf

Jim Drunkenmiller

Kyle Boller

It's batting 1.000 on bust identification. That doesn't mean that every single bust is correctly identified, only that if one is identified, it's actually a bust. Thereby eliminating a whole chunk of terrible draft picks.

Mike Vick was not a bust. He was the Falcons Franchise. Many of these other guys were predicted to be really good or bad because of their small schools, their types of offenses, or their mechanics. And I trust these things much more than a simple formula.

You're saying that it predicts busts well, but half of its studs are QBs who were busts. If you're saying that we should ignore the stud projections and only look for this formula saying that a QB will be a bust, then that alone says the formula isn't as strong as they predict.

What does the middle category even mean? Hit or Miss? Does that mean that the formula can't predict this group well either? There are quite a few busts in that group as well.

In truth, any formula that predicts busts well is going to do well because its gonna be a good indicator because most first round QBs aren't elite QBs. Most of the time they wind up being bad to decent QBs who are in competition for a backup spot (Harrington, Ramsey, Boller, Grossman, Leinart, Smith, Leftwich, Carr, etc).

I mean 18 out of the 31 QBs on the list are decent to busts. Thats almost a 60 percent ratio. Is it hard to find a formula that matches that? I don't think so.

[Taylor 36]

COLTS_SCORE= 0.5*(32) + 2.5*(70.39-60) + 5*3.11

COLTS_SCORE = 57.5250

You beat me to it. This puts him only behind Leinart and second on that list. But, that should be expected from a guy who broke almost every NCAA record in three years (32 games).

[HBnotBlades]

Mike Vick was not a bust. He was the Falcons Franchise. Many of these other guys were predicted to be really good or bad because of their small schools, their types of offenses, or their mechanics. And I trust these things much more than a simple formula.

I thought about writing this, but I didn't think anyone would really argue about Vick. We can really just throw him out, he wasn't drafted for his ability to throw the ball, and the success he had at the NFL level wasn't because of his elite passing skills. Because this formula doesn't consider running with the ball, he should be ignored.

Also, I'm not arguing that this formula is better at judging a players success than in depth scouting. It's obviously not. I'm arguing that it's not complete and total BS like many are claiming, and shouldn't be completely ignored or written off as totally inaccurate.

You're saying that it predicts busts well, but half of its studs are QBs who were busts. If you're saying that we should ignore the stud projections and only look for this formula saying that a QB will be a bust, then that alone says the formula isn't as strong as they predict.

OK, so let's say that point of the formula is to identify busts. That would mean that incorrectly identifying a stud as a bust would be a false-positive and incorrectly identifying a bust as a stud would be a false-negative. That sounds confusing, I know. On second though, I'll just post a link explaining this significance in statistics in case you want to read about it.

http://en.wikipedia.org/wiki/Type_I_and_Type_II_errors

The point is that they are two different types of errors, and in this case there are very few (none) false positive errors (studs that were labeled as busts), but quite a few false negative errors (busts that are labeled as studs). Now, if both of those types of errors were very low, we would have stumbled across the holy grail QB scouting. We'd have a perfect predictor, teams would only draft great QB's, the masses would celebrate, farts would smell like flowers, and all would be good in the hood.

Unfortunately, that model doesn't exist and never will. There are too many variables to label who will be great and it depends greatly on which teams draft them, thus there will always be false negatives. On the other hand, this formula does a good job at having very few false-positives, ie. studs who were incorrectly labeled as busts. By the way, I'm calling anyone score at 0 or below a bust (the bottom category). Now this is important for two reasons.

First, it identifies highly likely future busts in the draft so you don't draft them. Duh.

Second, (this ones important) by eliminating players that are known to be busts from the total pool, you increase the percentage of the remaining pool that is made up of future pro-bowlers. Thereby increasing the chances of being successful in the draft.

Think of it this way, pretend you were on deal or no deal and there were 20 cases. Three with lots of money, and 17 with ham sandwiches. If you made a guess and the case you wanted without eliminating any, you would have a 15% chance of getting money. Now if you could figure out 10 cases that were very highly likely to contain ham sandwiches (maybe they smell like ham), and you eliminated those, you would be left with a 30% chance of getting money (3 out of 10), doubling your chances of success. Hooray!

I'm just trying to make the point that this formula shows some promise at identifying busts and that identifying them, even if it's only a few, can be very useful. If we take that list for 30 first round QB's (that aren't Vick) and call the 8 pro-bowlers successes, then we have a 26.6% success rate. Now if we eliminate the 6 that are highly likely to be busts, we end up with a 33.33% success rate. The draft success probability just improved without actually know who was going to be good.

PS. You can include Vick as a false-negative and the success rate still improves from 25% to 29%.

PPS. That post ended up being entirely too long.

[[[ghost]]]

Those Tulane QBs sure are money.

[HawaiianTime]

Here it is a little bit more readable.

Scores of First-Round Quarterbacks, 1997-2008  

Group I: Strong likelihood of success  
Player                School           Draft year    Score 
Matt Leinart          USC              2006          64.04 
Colt Brennan         Hawaii           2008           57.53
Philip Rivers         NC State         2004          48.44 
Tim Couch             Kentucky         1999          47.64 
Alex Smith            Utah             2005          44.88 
Aaron Rodgers         California       2005          40.58 
Peyton Manning        Tennessee        1998          39.47 
Jason Campbell        Auburn           2005          38.75 
Byron Leftwich        Marshall         2003          36.39 
Ben Roethlisberger    Miami (Ohio)     2004          33.85 
Chad Pennington       Marshall         2000          33.53 
Daunte Culpepper      Central Florida  1999          30.00 
David Carr            Fresno State     2002          23.97 
Joe Flacco            Delaware         2008          23.92 
Eli Manning           Ole Miss         2004          23.14 
Donovan McNabb        Syracuse         1999          21.62 

Group II: Hit-or-Miss  
Player                School           Draft year    Score 
Brady Quinn           Notre Dame       2007          18.93 
JaMarcus Russell      LSU              2007          18.64 
Rex Grossman          Florida          2003          18.39 
Vince Young           Texas            2006          18.21 
Carson Palmer         USC              2003          16.35 
Matt Ryan             Boston College   2008          9.14 
Patrick Ramsey        Tulane           2002          9.06 
J.P. Losman           Tulane           2004          7.86 
Jay Cutler            Vanderbilt       2006          2.39 

Group III: Busts  
Player                School           Draft year    Score 
Akili Smith           Oregon           1999          0.00 
Cade McNown           UCLA             1999          -6.41 
Joey Harrington       Oregon           2002          -6.85 
Michael Vick          Virginia Tech    2001          -11.32 
Ryan Leaf             Washington St.   1998          -16.92 
Jim Druckenmiller     Virginia Tech    1997          -20.25 
Kyle Boller           California       2003          -50.67 

Waiting for the Colt lovers to put in his stats from college and have him be at least 2x higher than anyone else...

COLTS_SCORE= 0.5*(32) + 2.5*(70.39-60) + 5*3.11

COLTS_SCORE = 57.5250

However, there would be a red flag on him because he played in a gimmicky offense much like Alex Smith.

You beat me to it. This puts him only behind Leinart and second on that list. But, that should be expected from a guy who broke almost every NCAA record in three years (32 games).

So I wonder what kind of excuses will be made to not have an open QB competition.

[Late Night Roach]

So I wonder what kind of excuses will be made to not have an open QB competition.

I have them memorized. Here ya go, take your pick:

a) Too many coaches & offenses to deal with; deserves more time to develop.

O'line sucks.

c) Receivers can't catch.

d) Receivers can't get open.

e) All of the above.

[Going Commando]

C'mon. Look at how small the sample size is. Its about 30 players. Thats not enough to say this is an efficient formula. It predicts that Leinart is gonna be the best QB ever. And has Manning somewhere like 4th or 5th.

It'd be more interesting to see how this formula looks over all QBs taken in the draft. I wonder if this same correlation holds up. Probably not.

You are getting hung up on the individual numbers and not the tier system. There is too much variation between the college offenses and their competition and so on, so the numbers mean very little in and of themselves. All that matters is that your number is high enough to fall into the top "safe" pick tier. It doesn't mean that Matt Leinart is going to be better than Peyton Manning. It just meant that, given his collegiate production, he was regarded as a fairly safe pick to make it as a successful QB in the pros.

Second, I don't see the purpose in using the formula over all QBs in the draft. Maybe I'd include second rounders because you still expect them to start one day. But Nobody drafts a QB after the second round and thinks that they are getting their guy to be a future franchise QB. Teams take the low risk projects around that range hoping to be pleasantly surprised, not guys they are banking on becoming the man right away.

The evaluation has to be different because the expectations are different.

[Going Commando]

I have them memorized. Here ya go, take your pick:

a) Too many coaches & offenses to deal with; deserves more time to develop.

O'line sucks.

c) Receivers can't catch.

d) Receivers can't get open.

e) All of the above.

I know this is off topic, but I just wanted to put my two cents in.

Here's one you didn't mention: Teams don't typically begin their year by manufacturing a quarterback controversy and then go on to have good seasons. A quarterback needs to know that his coach is behind him, or else he is going to play nervously and generally like crap. It matters a ton to these QBs when they know they have the confidence of the organization. There is a reason teams don't open the door for this kind of controversy much any more. Even teams that have first round QBs riding the bench, like the Chargers and Browns of the last couple years, didn't open up a QB competition in training camp. There simply isn't enough time to get a proper evaluation at the position done. You have to go with your man, get him the first team reps, and hope for the best. You make the change if the wheels come off during the year when you realize that you made the wrong choice.

Besides, Campbell was a middle of the road starter last year and who offers the strong promise of improvement next year. Is there really even a point to opening up the can of worms that a QB competition will bring?

If nothing else, Campbell provides at least what Collins does as a caretaker QB who doesn't commit a lot of turnovers, and more than Brennan who would be a first year starter who would probably do just the opposite. When the running game was working, this was all we needed.

[BKSkinsFan]

So intense study and research indicates that the likelyhood of drafting a franchise quarterback is 50/50. Gee thanks for all the hard work on that ESPN, I would have never guessed.