Theory on predicting NFL QB success

[Leonidas]

Something very interesting that Gil Brandt devised and has been tweaked by others. Brandt initially came up with the idea that a QB starting 40 or more college games is much more likely to be a successful NFL QB than those with less experience. David Lewin of FootballOutsiders.com took it a step further and said QBs with 35 or more starts who completed 57% or more of their passes were more likely to succeed. Granted, there are exceptions listed and no surefire way to predict this, but the probabilities would appear to way in Lewin's favor.

http://www.ohio.com/mld/ohio/sports/...o/17156243.htm

Former Cowboys player-personnel guru Gil Brandt came up with a statistic for judging quarterbacks, and he determined that first-rounders who have at least 40 collegiate starts tend to perform much better in the NFL than those who had less experience. Recent four-year, 40-start quarterbacks include Manning, Chad Pennington, Drew Brees, Carson Palmer and McNabb.

• Since 1997, seven quarterbacks who had fewer than 30 college starts were picked in the top 10: Akili Smith, Tim Couch, Ryan Leaf, Joey Harrington, Michael Vick, David Carr and Alex Smith -- according to an excellent article by David Lewin on FootballOutsiders.com.

• Yes, there are exceptions: Kyle Boller had 40 starts but flopped with the Baltimore Ravens. A tip-off should have been that he failed to complete 50 percent of his passes in several seasons. Tom Brady barely started for two years at Michigan, and he has won a few games with the New England Patriots. Cade McNown had more than 40 starts at UCLA and not much of a pro career. Ben Roethlisberger had 38 starts, and his experience showed when he had to start as a rookie for the Pittsburgh Steelers.

• Lewin took the 40-start theory and added that the closer a quarterback is to completing 60 percent of his passes, the more likely he'll be successful -- assuming he had close to 40 starts. Lewin wrote: ``Here's a list of the QBs drafted in the first two rounds over the last 10 years with at least 35 starts and completing 57 percent of their passes: Peyton Manning, Donovan McNabb, Daunte Culpepper, Chad Pennington, Drew Brees, Carson Palmer, Byron Leftwich, Eli Manning, Philip Rivers, Ben Roethlisberger, Jason Campbell, Matt Leinart and Jay Cutler... Are there any bad players on that list?''

• There are exceptions. Brady had 29 starts; Vince Young had 32 for Texas. Nonetheless, the theory seems to be right more often than wrong. For the record, top pick JaMarcus Russell had 29 starts and completed 62 percent of his passes.

• That brings us to Quinn, who started 46 games at Notre Dame, completing 58 percent of his passes. In the past two seasons, he was 62 percent with 64 touchdowns, with 14 interceptions. Given the major schedule he played and how he started for four years, Quinn should have a good chance to become a viable pro quarterback.

And for the record, here's how QBs in this year's draft compare. (note, the total number of starts may not be exact as that info was not always readily available, so I made my best WAG with the info I could find)

Oakland 1st round Jamarcus Russell 29 starts (assuming the column above is correct), 62% completion

Cleveland 1st Round Brady Quinn 46 career starts (if you assume the column above is correct) 58% completion

Detroit 2nd Round Drew Stanton 29 starts, 64.2% had a really down year his senior season (injured I presume) compared to a brilliant junior year (God the MSU webpage was a pain in the butt to find his total stats on as I had to access three different PDF files to put it all together because his individual bio didn’t have stats updated after his sophomore season)

Miami Second Round John Beck 43 games (not sure how many were starts, but it’s likely between 35-40 as he played 35 games as the starter in his last three seasons and had 8 games coming off the bench or sometimes starting as a freshman) 62.4% completion

Phillie 2nd Round Kevin Kolb 50 career starts (assuming he started all 14 games last year as he started every game his first three years, yes Houston had another ass painful page to deal with) 60% completion

Buffalo third round Trent Edwards 31 career starts out of 35 games, 56.3% although was over 60% his last two years

Dallas 4th round pick Isaiah Stanback 37 total games, I’m going to guess at most 20 were QB starts, 51.4% completion

Cinci 5th round Jeff Rowe 27 starts, 59.6% completion

Baltimore 5th round Troy Smith 27 career starts (31 games total as QB, 44 total as a player) 63.4% completion

Washington 6th round Jordan Palmer 46 total games, somewhere between 36 and 46 were starts, 60% completion

I’m not even going to bother looking up the guy from Coastal Carolina Minnesota drafted in the 7th

Russell, Stanton, Rowe and Smith all had great completion percentages but came up a bit short on overall experience, which struck me as odd as it seemed to me Smith and Stanton were there forever. I guess Smith lost a lot of starts with his competition to Zwick and his lost two starts for the suspension so he could have been much closer to 40. Edwards doesn’t meet the criteria anywhere. I suspect he is this year’s version of Kyle Boller, a guy who combined really well but didn’t have the numbers on the field to back up the talent. Quinn, Kolb, Beck, and Palmer hit all the criteria dead on, but other than Quinn against questionable competition. And why Dallas took Stanback is way the hell beyond me unless he is going to be a slash kind of guy.

If you take competition into consideration and go along with this theory, then Brady Quinn is your guy, I'd say followed next by Beck who I'd rate played better competition than Palmer and Kolb. Take your chances with Russell, Stanton, and Smith as they played well against tough competition but with fewer starts than the model allows. Personally I'd take either of them over Palmer or Kolb and would keep Beck in the mix.

[Logan]

You could've saved yourself a lot of typing and just read this thread.

[Leonidas]

Originally Posted by Logan

You could've saved yourself a lot of typing and just read this thread.

No one there did the research on this year's draftees. So not a wasted effort. And before the flaming begins, I neither support nor disagree with the theory. I just was curious how the guys coming out this year pan out according to this model. And I really, really have to think Stanback was drafted to return kicks and catch passes.

[JAG]

Originally Posted by Leonidas

And I really, really have to think Stanback was drafted to return kicks and catch passes.

Correct. Dallas has already stated that they're moving him to receiver / returner.

[Pumpy Tudors]

I'm no scout, but I wonder what makes somebody look at a college quarterback and think, "Wow, he'd make a great receiver!" I mean, obviously, it happens (Antwaan Randle El, Matt Jones, comedy Marcus Vick option), but it's strange to me. How come more running backs and defensive backs aren't being converted to receiver?

And why am I hijacking this thread? Sorry, guys.

[Abe Sargent]

Please note that the theory is for first round QBs, not for them over the course of the entire draft.

[Klinglerware]

I have to say that I am suspicious of the methodology. Number of starts seems like a reasonable metric--the more live game experience, the better, obviously.

But that completion percentage seems suspect since it is very difficult to compare passers with this stat without knowing anything about the context on how the completions were produced, especially in college where systems and competition levels vary wildly.

[cthomer5000]

I think there's about a million issues with limited sample size, back-testing, etc. More likely an interesting coincident than a "tool."

[Pumpy Tudors]

Does anyone know what Tampa Bay's record is when all of the following were true?

1) The game time temperature is below 50 degrees
2) It's the day immediately after an Ole Miss road win
3) A bizarre field goal attempt from a high school game is shown on SportsCenter within the previous 7 days
4) Terry Bradshaw is full of shit

Can someone look that up for me? Thanks.

[rkmsuf]

The secret of their success it that they're livin 25 hours a day.

[st.cronin]

Originally Posted by Pumpy Tudors

Does anyone know what Tampa Bay's record is when all of the following were true?

1) The game time temperature is below 50 degrees
2) It's the day immediately after an Ole Miss road win
3) A bizarre field goal attempt from a high school game is shown on SportsCenter within the previous 7 days
4) Terry Bradshaw is full of shit

Can someone look that up for me? Thanks.

576-0-2

[Passacaglia]

Originally Posted by Pumpy Tudors

I'm no scout, but I wonder what makes somebody look at a college quarterback and think, "Wow, he'd make a great receiver!" I mean, obviously, it happens (Antwaan Randle El, Matt Jones, comedy Marcus Vick option), but it's strange to me. How come more running backs and defensive backs aren't being converted to receiver?

And why am I hijacking this thread? Sorry, guys.

At least in the case of Randle El, he played a lot of WR in college. My only guess as to the thinking involved was that they knew he was better suited as a WR, but they figured they had to have their best guy at QB, and kept changing their minds about it. Then again, this was Cam Cameron, so he might have just been swtiching his position around because Ted Ginn, Sr. told him to.

[Maple Leafs]

Originally Posted by Pumpy Tudors

Does anyone know what Tampa Bay's record is when all of the following were true?

Weird stat: the Bucaneers have never allowed a point in a game in which they've returned a kickoff for a touchdown.

[DanGarion]

So how about all those guys that play 40 games and have 60% and never get drafted? I mean this is all fine and dandy, but if it's only looking at the 1st round, it's kind of funky...

[VPI97]

Football guy 1: "Hey, I think I found a way to see if a quarterback is going to be any good in the pros. Anyone who starts over 40 games in college is the guy to pick. If they have less than 30 starts, they're gonna suck. Just look at Manning vs. Leaf!!!"

Football guy 2: "What about Brady? He didn't have a lot of starts at Michigan"

Football guy 1: "Uhhhh...well, it only works for first rounders. Yeah."

Football guy 2: "Wasn't Patrick Ramsey a four year starter at Tulane?"

Football guy 1: "I'm only talking about top 15 picks here. Where was Ramsey taken?"

Football guy 2: "Thirty-second"

Football guy 1: "See. Not the same."

Football guy 2: "What about Cade McNown? He was selected twelfth."

Football guy 1: "I'm only talking about top 10 picks here. Can't you listen?"

Football guy 2: "Well, what about Troy Aikman? He only played for a couple years in college and he was the first pick in the draft. ...and look at Rick Mirer. He played for Notre Dame for a long time and then blew in the pros."

Football guy 1: "Yeah, but they're old. They don't count."

Football guy 2: "So what you're saying is that your method is a foolproof way to predict the success a college quarterback will have in the NFL...but only if they're selected in the first ten picks...and your supporting data only values what's happened in the last ten years?"

Football guy 1: "Yeah, I'm a genius"

Football guy 2: "Brilliant."

[Abe Sargent]

I think there might be some validity behind all of the bad statistics, but I'd need a real statistician to work it out. I think it might go something like this:

"At the top of the draft, the QBs that are taken have a statistical correleation with each other due to the fact that they are approximately equal in skill level in order to be selected in the same area. Even though individual skills may differentiate, the overall QB package is consistent. Therefore, we can look at statistics that may have an implication for how said top of the draft QBs correlate with their later success. The hypothesis being test is that when you are drafting QBs of approxmiate skill level (which can only be done at the top of the draft, not in later rounds), the QB with more starting experience is more likely to be the more successful one in the NFL."

But again, I'd need real datat, not that provided by "football writer and statistical amateur."

[MrBigglesworth]

Originally Posted by Klinglerware

I have to say that I am suspicious of the methodology. Number of starts seems like a reasonable metric--the more live game experience, the better, obviously.

I would say that a player with more than 40 college starts is someone that was highly regarded out of high school and played well enough in college to keep the starting job for four years, and so has a long history of success. So I don't know if college starts causes ability as much as college starts correlates with ability.

[sabotai]

Originally Posted by Pumpy Tudors

I'm no scout, but I wonder what makes somebody look at a college quarterback and think, "Wow, he'd make a great receiver!" I mean, obviously, it happens (Antwaan Randle El, Matt Jones, comedy Marcus Vick option), but it's strange to me. How come more running backs and defensive backs aren't being converted to receiver?

And why am I hijacking this thread? Sorry, guys.

My first guess would be because QBs already know all of the routes precisely. Or at least the ones they ran in college. They know exactly where the WR is supposed to be and when. How well they personally run the routes is a different issue, but they know all of the routes as well as anyone. RB don't really know the routes at all (they have their own, different set of routes to learn and execute), and DBs do to an extent, but not as well the QB.

And since they QBed, you know they will be sensative to how well the QB expects the route to be ran (while lifelong WRs might not be as sensative to it).

Secondly, if they run an offense that requires the QB to run a lot (a lot of options, a lot of QB draws, QB scrambling a lot by design), then the QB also has experience running with the ball, breaking tackles, avoiding pursuit, etc. Which is something the DBs (lifelong DBs anyway) don't have any experience doing.

[Klinglerware]

Originally Posted by MrBigglesworth

I would say that a player with more than 40 college starts is someone that was highly regarded out of high school and played well enough in college to keep the starting job for four years, and so has a long history of success. So I don't know if college starts causes ability as much as college starts correlates with ability.

That's a pretty reasonable argument.

[Vinatieri for Prez]

This just in: the higher your completion percentage and the more you play, the better QB are. Never would have thunk it.

It's absolutely lame to differentiate between a guy who played 38 games and 41 games, without taking into account about oh, I don't know, a hundred other variables.

[tanglewood]

The Football Outsiders original article where the guy chose 35 starts and 57% as a metric said that the model deliberatley 'cheats' somewhat by only allowing use on the first two rounds. Essentially it uses the NFL scouts and draft gurus to narrow down the huge field of college QBs into those who have obvious potential and then goes from there. That's why it doesn't get bogged down by small school QBs who start 4 years and put up 65% completetion rates in system offenses being found out at the NFL level, because scouts can see these are system QBs so they don't get picked high.

However the author did say that Colt Brennan of Hawaii will be an interesting test for the system. If he came out this year he was slated to be the third QB of the board and probably gone late-first/early-second. The system would favour him on completion percentage but have been overall quite down on him as he hadn't started the games. Lewin said basically that if he goes back to school, starts the whole season, puts up good numbers and the scouts still like him enough that he goes in the first two rounds, the system will rate him highly next year.

[ctmason]

Originally Posted by Leonidas

I’m not even going to bother looking up the guy from Coastal Carolina Minnesota drafted in the 7th.

You can bet your ass that in 25 years when I'm writing Tyler Thigpen's Hall of Fame induction speech I'm bringing this up...and calling you out!

[TazFTW]

Originally Posted by tanglewood

Lewin said basically that if he goes back to school

Check.

starts the whole season

Check barring injury.

puts up good numbers

Hawaii has two 1AA opponents on their 12 game schedule (freakin' Herman Frazier).

the scouts still like him enough that he goes in the first two rounds

Here's the key. I know June Jones added more plays with Colt taking snaps under center this Spring. When Colt was considering leaving early it was reported that the Dolphins were showing interest. With the number of Warriors ending up in Miami, it sure looks like Miami spent some time scouting Hawaii players.

the system will rate him highly next year.

Well duh.

[yabanci]

The full, much more detailed version of this article appeared in the 2006 Pro Football Prospectus.

So who is David Lewin? some statistical guru or some highly respected football analyst?

Here is his biography from Pro Football Prospectus:

"David Lewin is a sophomore student and quarterback at Macalester College in St. Paul, Minnesota. He has yet to accumulate enough starts or complete a high enough percentage of his passes to be considered worthy of a first-day draft pick, although he did toss a 70-yard touchdown against St. Olaf"

That explains why the theory sounds like something concocted by a 19 year-old kid at a tiny liberal arts college.