05-01-2007, 08:16 AM | #1 | |||
College Benchwarmer
Join Date: Jan 2003
Location: East Anglia
|
Theory on predicting NFL QB success
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 Quote:
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.
__________________
Molon labe |
|||
05-01-2007, 08:22 AM | #2 |
Head Coach
Join Date: Oct 2000
Location: NYC
|
You could've saved yourself a lot of typing and just read this thread.
|
05-01-2007, 08:26 AM | #3 | |
College Benchwarmer
Join Date: Jan 2003
Location: East Anglia
|
Quote:
__________________
Molon labe |
|
05-01-2007, 08:45 AM | #4 |
Grizzled Veteran
Join Date: Nov 2000
Location: St. Paul, MN
|
|
05-01-2007, 09:09 AM | #5 |
Bounty Hunter
Join Date: Oct 2000
Location: Pittsburgh, PA
|
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.
__________________
No, I am not Batman, and I will not repair your food processor. |
05-01-2007, 09:14 AM | #6 |
Hall Of Famer
Join Date: Dec 2001
Location: Catonsville, MD
|
Please note that the theory is for first round QBs, not for them over the course of the entire draft.
__________________
Check out my two current weekly Magic columns! https://www.coolstuffinc.com/a/?action=search&page=1&author[]=Abe%20Sargent |
05-01-2007, 09:19 AM | #7 |
College Starter
Join Date: Dec 2003
Location: The DMV
|
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. |
05-01-2007, 09:31 AM | #8 |
Strategy Moderator
Join Date: Nov 2001
Location: North Carolina
|
I think there's about a million issues with limited sample size, back-testing, etc. More likely an interesting coincident than a "tool."
|
05-01-2007, 09:36 AM | #9 |
Bounty Hunter
Join Date: Oct 2000
Location: Pittsburgh, PA
|
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.
__________________
No, I am not Batman, and I will not repair your food processor. Last edited by Pumpy Tudors : 05-01-2007 at 09:36 AM. |
05-01-2007, 09:40 AM | #10 |
Head Coach
Join Date: Dec 2001
|
The secret of their success it that they're livin 25 hours a day.
__________________
"Don't you have homes?" -- Judge Smales |
05-01-2007, 09:43 AM | #11 | |
General Manager
Join Date: Oct 2004
Location: New Mexico
|
Quote:
576-0-2 |
|
05-01-2007, 09:43 AM | #12 | |
Coordinator
Join Date: Oct 2000
Location: Big Ten Country
|
Quote:
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. |
|
05-01-2007, 09:46 AM | #13 |
Coordinator
Join Date: Jan 2002
|
Weird stat: the Bucaneers have never allowed a point in a game in which they've returned a kickoff for a touchdown.
__________________
Down Goes Brown: Toronto Maple Leafs Humor and Analysis |
05-01-2007, 10:20 AM | #14 |
Coordinator
Join Date: Nov 2003
Location: The Great Northwest
|
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...
|
05-01-2007, 10:56 AM | #15 |
Hokie, Hokie, Hokie, Hi
Join Date: Mar 2001
Location: Kennesaw, GA
|
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." |
05-01-2007, 11:24 AM | #16 |
Hall Of Famer
Join Date: Dec 2001
Location: Catonsville, MD
|
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."
__________________
Check out my two current weekly Magic columns! https://www.coolstuffinc.com/a/?action=search&page=1&author[]=Abe%20Sargent |
05-01-2007, 12:27 PM | #17 |
College Starter
Join Date: Dec 2004
Location: PA
|
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.
|
05-01-2007, 12:40 PM | #18 | |
General Manager
Join Date: Oct 2000
Location: The Satellite of Love
|
Quote:
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. |
|
05-01-2007, 01:49 PM | #19 | |
College Starter
Join Date: Dec 2003
Location: The DMV
|
Quote:
That's a pretty reasonable argument. |
|
05-01-2007, 03:20 PM | #20 |
College Starter
Join Date: Feb 2003
Location: Seattle
|
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. Last edited by Vinatieri for Prez : 05-01-2007 at 03:21 PM. |
05-01-2007, 05:49 PM | #21 |
College Benchwarmer
Join Date: Oct 2004
|
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. |
05-01-2007, 05:58 PM | #22 |
High School JV
Join Date: Oct 2000
Location: Mesa, AZ
|
|
05-01-2007, 06:56 PM | #23 | ||||
Pro Rookie
Join Date: Oct 2002
Location: Honolulu, HI
|
Check. Quote:
Check barring injury. Quote:
Hawaii has two 1AA opponents on their 12 game schedule (freakin' Herman Frazier). Quote:
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. Quote:
Well duh.
__________________
"Teams don't want to make the trip anymore," says Hawaii coach June Jones. "They come here, we kick their ass, they go home." Fire Ron Lee. |
||||
05-01-2007, 07:12 PM | #24 | |
College Benchwarmer
Join Date: Jun 2003
|
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: Quote:
That explains why the theory sounds like something concocted by a 19 year-old kid at a tiny liberal arts college. |
|
Currently Active Users Viewing This Thread: 1 (0 members and 1 guests) | |
Thread Tools | |
|
|