Is anything as insane as trying to predict whether a College QB can ultimately make it in the NFL?

Everyone has a theory – most of which surrounds a “visual” inspection from watching a player during games in college…”he has a big arm”, “he makes all the throws”, “he's got moxie”, “he’s a winner”, “love the mobility”. Is there anyone you know in any publication, or on TV, or even among the NFL Draft “experts”...that are really a “QB Whisperer” for college QBs?

Everybody has predicted some college-to-pro QB correctly along the way. Seemingly for every one correct “prediction”, there is an equal amount of  “wrong” prognostications. I was more in the Ryan Leaf camp, than the Peyton Manning camp…but that was more of a ridiculous youthful bias, because I just didn’t like Peyton Manning in college. I thought he was "cocky". Which I had no basis in any reality to know that, nor was that any reason to not like a particular QB. Blinded by my mind, with no real football analysis. I’ve gotten them right and wrong over the years as a fan…no great strategy, more just gut feeling.

Being anti-“Gut feeling” is the passion behind my research over the last few years and this website. I cannot trust my “gut”. My “gut feeling” is based on nothing concrete. Bill James and Billy Beane (among others) led the way of attempting to figure out a formula (of sorts) to project young talent to the higher levels (in baseball). My passion is to try to figure out the same for the NFL and Fantasy Football.

No “gut” and no formula can be 100% accurate. Like a gambler and like a stock investor, I just want to tip the odds in my favor. Whenever I hit a statistical theory that gets 70+% accuracy, I get excited. I am excited now – because I think I have stumbled upon and refined something with College QBs.

There are other QB "theories" out there. One is the "26-27-60 rule", which is:

1. 26 or greater scored on the Wonderlic

2. 27 games started in college

3. 60% or greater completion percentage

Bill Parcells has a popular formula or (semi) “mathematical” system/theory, for drafting a College QB. His theory is that a college QB m

1. Be a minimum 3 year starter

2. Has at least 23 wins in his career

3. Plays his senior season

4. Graduates college

With that you get Parcells draftee success Chad Pennington (and Philip Rivers, Drew Brees among others). You also get Parcells potential draft bust Chad Henne (or Matt Leinart and Cade McNown among others). In total, however, you get 50+% success rate with the Parcells theory. Again, no system can be perfect. Above 50% in this arena isn’t bad. The "26-27-60" rule I like a little better, because I believe in the pattern of Wonderlic level and QB future prowess (or should say I believe in low scores as a predictor of failure). Both the Parcells and "26-27-60" are nice building blocks of a process, but what if you had a system with a 70+% success rate? What if you could put a predictable number to a QB prospect, based on several telling variables?

That is my claim, on a current work in progress. However, it is not as simple as Parcells 4-category in/out theory. I will lay out a high-level look at my theory, and you can judge if this is the ravings of a madman.

 

The Theory:

I began with a simple thought process. Would the current NFL “elite” QBs have something statistically in common from college -- that set a benchmark to judge other QBs by? A benchmark that was hard for other QBs to achieve, and thus create a way of weeding out college QB prospects.

So it began, running numbers on Brady, Manning, Brees, Rodgers, Roethlisberger and Rivers. That was my baseline "elite" QB list, I’m sure someone will argue an addition or subtraction…but could we agree it’s a pretty stable foundation?

A good/great QB list, but could anything be found in common with them? 

 

The Ingredients:

I ran across a couple data points that the “elites” seemed to have in common, but it was hiding. Some serious trial and error needed to be done.

The core data that is the foundation for my mathematical system to judge College QBs is as follows (not going into exact detail for many reasons, seeking fortune by having something no else does is the main one...):

·         KEY MATCHUPS – the first thing I do is throw out ridiculous games (which typically leads to ridiculous results) or in other words, throw out games/stats against horrible teams on their schedule. Tim Tebow (or anyone) beating up on Charleston Southern or Florida Atlantic, why would we consider those statistics as a "tell" on a QB's ability? By contrast, Joe Flacco facing James Madison U…that is a more legit game, as Flacco played for Delaware -- more equal talent levels of teams. We look at only logical games (to not be too subjective on that, we typically just look at games against teams with winning records), which reduces the sample of a season to somewhere between 6-10 games for any QB. We only look at games played in their final year, not the entire body of work.

·         WEIGHTED MATCHUPS – playing big games on the road, playing top teams (and/or defenses) in the college ranks within their season, that’s worth looking at harder. I use strength of opponent for judging an individual games value for statistically rating a QB. A matchup against a Top-10 team is worth a lot more "weight" than their matchup against a 6-6 barely qualified for a Bowl Game matchup.

·         PERFORMANCE PER ATTEMPT – A 300-yard passing game in 30 pass attempts, is that not similar to a 200-yard passing game in 20 attempts?…or more to the point a 400-yard game in 40 attempts? It's all still 10 yards per pass attempt? For that matter 2 passing TDs in 20 Pass Attempts is the essentially the same performance as 3 passing TDs in 30 Pass Attempts. Colt Brennan and Graham Harrell should not have an advantage over Tom Brady because they throw 40+ times a game and compile bigger stats. We look at Passing yards, Passing TDs and INTs through the lens of the number of completions and attempts...not game compilations. Quality, not quantity.

·         PHYSICAL CHARACTERISTICS – Height, Hand-size, arm length are some items we look at. Speed plays no factor. Do any of the elite NFL QBs of today exude "big speed"? Mobility is not a characteristic of most/all elite QB (please no Michael Vick yet, we’ll get to that).

·         WONDERLIC/IQ – an unavoidable and a key data point we have access to on basic IQ and problem solving. There is a definite correlation to low Wonderlic scores and QB disappointment.

These (above) are all things that are "known" and quantifiable, and measurable back to the past. What I don't know...I don't know, and can't consider. A pending character issue/flaw, or a hidden injury -- we can't really quantify that.

 

The Results:

Devising a mathematical formula for all of this was complicated, and is still evolving as I re-study and re-test the research I have so far -- plus try to add to it. I have 59 QBs researched and input at the time of this writing. The classifications seem to be falling into 3 scoring ranges (keep in mind a 1.000 score is if a QB hit all the metric marks on the nose, but QBs can exceed 1.000 in any given sub-category and thus can be above 1.000 if they exceed enough of the categories.

With a numerical bar for each category set, the individual QB's results are then input, and the calculations take form. For simplicity, (crazy example) if every ultimately successful QB in the NFL averaged 3.0+ Passing TDs per game in college, and we set a bench mark for 100.0% grade in that category at 3.0 TDs -- then if a particular QB averaged 4.0 TDs, he would score a 1.250 in our system for that category. Take that same thought process times multiple categories with deductions for key "red flag" items and a little more complexity for the data/comparison levels of good/bad results within a category -- and hopefully that helps with some understanding of where we were going.

A 1.000+ QB is essentially performing well above the "elite" QB standard set, and has exhibited little or no red-flags. No one particular category (among the many we have) can rocket a QB to "great" level, no 1-2-3+ categories can be such an elaborate high score that it covers over other obvious statistical flaws. One red-flag really knocks a QB down from achieving near a 1.000 score, 2 red-flags are a near statistical killer from being in that .9000-1.000+ range.

The 2 tiers of scoring that have stood out so far (full player by players totals posted near end of the article):

1)      0.850-1.000+ = Seems to be the mark that puts you on the map of being a potential very good NFL starter, with the chance of being 'elite". All of the elite QBs have hit this level (Brady, Manning, Rodgers, Brees, etc). In total, there are 16 QBs who have been drafted already that scored 0.850+ so far in my mathematical system – of that there are 5 "question marks" QBs.

• Alex Smith = who has chance still, but I don’t like the way it looks at all

• Kevin Kolb = who I think will be an elite QB, but hasn't had a full chance to prove it. The mathematical system says elite possible too.

• Tony Pike = not enough info

• Byron Leftwich = not sure what to say here. Was a decent QB, got hurt.

      That leaves 12 of 16 QB's hitting this mark that have been good and/or great or are on their way (Bradford, Sanchez). 12 of 16 QBs predicted is a 75% mark, and if Kevin Kolb is as good/great as I think he is from this data and game tape study -- we could be at 13 of 16, or 81%...considering Leftwich as not worthy of being counted as "good/great. If Alex Smith and/or Tony Pike were to hit, this would be an amazing home run of a system.

 

2)      0.849 and lower  = The more likely to "bust" group, and the probability of the "bust" increases as the score goes lower, for the most part). There are good QBs in this range, but no elites. The best QB's sitting in this range are Tony Romo and Matt Ryan. Eli Manning, Donovan McNabb, and Jay Cutler are in this grouping as well. All good QB's, but are they elite? I see 7-8 decent QBs in the "bust" zone, of 43. An 81.3% chance of a "bust" if you fall into this scoring.

At this point I would want to say that a college QB scoring 0.850 or better has a 70+% chance of being a good NFL QB, and a potentially great QB (especially if above 0.950). I would also then say anyone under 0.850 has a 70+% chance of busting. Again, that leaves chances the system is wrong...but not many times. Stop and think for a second (assume I’m on to something) – can you imagine a mathematical system that would have alerted you to the superstar potential of Tom Brady 10 years ago in 2000 (the current # 3 QB on my list)? Or non-first round draft picks QB's like Drew Brees or Matt Schaub? Many could predict Peyton Manning, but who had Tom Brady or Joe Flacco ahead of time? With my mathematical system -- I would have. 

 

The Ratings:

Note:

·         “Adj” = adjusted results weighted for strength of opponent, and results against weak opponents thrown out.

·         "Equalized" Per game -- the QBs college stats in "key" games/tougher opponents based on an equal amount of passes per game for each QB (35 per game).

·         We're not putting all the data points on here, or the list would be a mile wide. Just clipped what I thought would be interesting on some of the per attempt, per completion tallies.

*In yellow are the elite/pioneer QBs that I started to build the foundation of this system upon to see if a system could be built.   

	P Score	QB	Yr	College	Comp Pct	Yds per Comp	Pass per TD	Pass Per INT	Equalized Yds Per game	Equalized TDs per game	Equalized INT per game
1	1.156	Bradford, Sam	2008	Oklahoma	65.5%	13.8	10.8	51.0	315.5	3.4	1.0
2	1.120	Roethlisberger, Ben	2003	Miami, Ohio	68.6%	13.7	17.1	58.0	333.0	2.0	0.7
3	1.109	Brady, Tom	1999	Michigan	63.8%	12.0	15.0	40.0	266.3	2.3	0.9
4	1.077	Manning, Peyton	1997	Tennessee	62.4%	11.7	15.6	44.4	252.3	2.3	0.9
5	1.022	Pennington, Chad	1999	Marshall	64.7%	13.6	17.0	37.4	282.1	1.9	1.0
6	1.015	Palmer, Carson	2002	USC	62.8%	12.6	15.8	45.7	265.2	1.9	0.6
7	1.007	Sanchez, Mark	2008	USC	66.0%	11.5	11.8	37.6	256.7	3.1	0.9
8	0.994	Rodgers, Aaron	2004	California	63.8%	11.0	18.6	37.3	253.1	1.5	0.6
9	0.993	Rivers, Philip	2003	NC State	67.4%	12.3	15.7	74.5	286.1	2.5	0.6
10	0.973	Leftwich, Byron	2002	Marshall	65.9%	12.1	17.6	44.0	276.7	2.1	0.8
11	0.947	Smith, Alex	2004	Utah	67.8%	13.2	11.3	90.0	307.1	3.0	0.4
12	0.917	Kolb, Kevin	2006	Houston	66.2%	13.8	15.1	272.0	310.5	2.3	0.1
13	0.904	Brees, Drew	2000	Purdue	60.8%	11.8	16.4	37.6	253.2	2.1	1.0
14	0.879	Pike, Tony	2009	Cincinnati	61.4%	11.7	13.9	44.6	218.0	2.4	0.9
15	0.873	Schaub, Matt	2003	Virginia	68.8%	9.9	23.6	47.1	238.3	1.5	0.7
16	0.859	Flacco, Joe	2007	Delaware	60.2%	12.5	22.4	179.5	264.2	1.6	0.2

**Less than 0.850, higher “bust” or never make it potential…  

	P Score	QB	Yr	College	Comp Pct	Yds per Comp	Pass per TD	Pass Per INT	Equalized Yds Per game	Equalized TDs per game	Equalized INT per game
17	0.822	Beck, John	2006	BYU	66.4%	12.6	16.3	61.0	287.3	2.2	0.5
18	0.811	Stafford, Matt	2008	Georgia	60.6%	14.3	14.4	34.3	283.9	1.9	1.4
19	0.782	Leaf, Ryan	1997	Wash State	52.8%	17.8	16.6	37.9	333.7	1.9	0.9
20	0.752	Ryan, Matt	2007	Boston College	58.8%	12.2	21.1	34.5	236.5	1.6	1.1
21	0.721	Couch, Tim	1998	Kentucky	66.6%	10.1	23.7	36.9	224.6	1.4	0.9
22	0.696	Carr, David	2001	Fresno State	62.7%	12.7	14.7	93.0	246.9	2.0	0.4
23	0.685	Manning, Eli	2003	Ole Miss	57.3%	12.2	19.6	31.9	239.9	1.7	1.1
24	0.649	Harrington, Joey	2001	Oregon	55.5%	11.5	20.0	50.0	222.8	1.7	0.7
25	0.645	Russell, JaMarcus	2006	LSU	62.9%	14.2	18.9	34.0	270.9	1.2	1.2
26	0.631	Campbell, Jason	2004	Auburn	72.1%	13.2	16.3	29.4	335.6	2.2	1.2
27	0.616	Brohm, Brian	2007	Louisville	63.2%	12.2	21.2	35.3	269.7	1.7	1.0
28	0.610	Leinart, Matt	2005	USC	60.8%	12.4	31.9	51.0	274.1	0.9	0.7
29	0.607	Freeman, Josh	2008	Kansas State	55.0%	12.7	21.8	30.0	260.0	1.6	1.4
30	0.596	McNabb, Donovan	1998	Syracuse	59.4%	12.7	14.1	49.3	290.7	2.7	0.5
31	0.592	Whitehurst, Charlie	2005	Clemson	65.2%	10.1	32.9	32.9	226.6	1.1	1.0
32	0.579	Quinn, Brady	2007	Notre Dame	58.8%	11.5	17.3	62.2	214.3	2.1	0.9
33	0.532	Booty, John David	2007	USC	62.1%	10.5	15.0	97.5	236.6	2.5	0.3
34	0.532	Young, Vince	2005	Texas	69.0%	14.0	11.8	35.5	310.6	2.3	0.9
35	0.489	Romo, Tony	2003	Eastern Illinois	62.3%	10.6	15.7	20.0	228.6	2.3	1.8
36	0.465	Olson, Drew	2005	UCLA	59.1%	14.7	11.0	51.3	256.1	2.5	0.4
37	0.461	Cutler, Jay	2005	Vanderbilt	57.1%	11.5	21.2	31.8	203.0	1.3	1.3
38	0.456	White, Jason	2004	Oklahoma	62.3%	11.7	10.9	25.9	233.4	2.4	2.0
39	0.454	Ainge, Erik	2007	Tennessee	60.8%	11.1	22.2	63.0	235.2	1.6	0.7
40	0.446	Walter, Andrew	2004	Arizona State	55.4%	11.8	19.3	57.8	237.6	1.7	1.0
41	0.382	Smith, Rusty	2009	Florida Atlantic	57.8%	12.8	18.5	37.0	243.0	1.5	1.3
42	0.377	Harrell, Graham	2008	Texas Tech	63.2%	12.6	14.3	65.8	282.4	2.2	0.4
43	0.357	Kafka, Mike	2009	Northwestern	63.0%	8.8	25.7	30.8	200.0	1.3	1.0
44	0.308	O'Connell, Kevin	2007	SD State	59.7%	10.6	52.5	52.5	222.3	0.6	0.8
45	0.271	Tebow, Tim	2009	Florida	67.8%	13.1	17.7	46.0	301.2	1.7	0.9
46	0.270	LeFevour, Dan	2009	Central Mich	65.3%	9.4	24.4	53.6	211.7	1.3	0.7
47	0.256	Greene, David	2004	Georgia	53.9%	13.0	15.7	51.0	245.8	2.1	0.9
48	0.237	Henne, Chad	2007	Michigan	58.6%	12.0	15.0	42.0	210.2	1.7	0.7
49	0.228	Smith, Troy	2006	Ohio State	63.9%	12.2	13.2	39.5	253.9	2.3	1.0
50	0.196	Brennan, Colt	2007	Hawaii	65.4%	12.0	18.6	20.5	321.3	1.9	2.5
51	0.192	Croyle, Brody	2005	Alabama	57.3%	13.2	21.4	171.0	233.9	1.4	0.2
52	0.145	Hall, Max	2009	BYU	65.5%	12.2	11.9	28.3	261.8	2.4	1.5
53	0.117	Skelton, John	2009	Fordham	61.0%	10.8	47.0	141.0	231.6	0.7	0.2
54	0.049	Orton, Kyle	2004	Purdue	52.8%	13.5	15.4	72.0	242.2	2.0	0.6
55	-0.074	McCoy, Colt	2009	Texas	66.5%	8.9	26.8	23.0	188.6	1.0	1.7
56	-0.310	Grossman, Rex	2002	Florida	56.1%	11.6	25.2	25.2	226.4	1.2	1.3
57	-0.364	Garrard, David	2001	East Carolina	54.7%	12.7	25.1	25.1	243.5	1.4	1.4
58	-0.436	Snead, Jevan	2009	Ole Miss	51.3%	13.1	46.8	15.6	209.2	0.5	2.9
59	-0.561	Anderson, Derek	2005	Oregon State	51.2%	12.9	20.1	29.3	223.0	1.7	1.3
60	-0.667	Robinson, Zac	2009	Oklahoma State	54.5%	9.8	30.4	21.3	176.4	1.1	2.4

 

The first reaction analysis:

Yes, Alex Smith in the upper group and David Garrard in the negatives are "misses". It’s everything in-between that starts to bring the results into focus…Bradford, Roethlisberger, Brady, P.Manning and Pennington aren’t a bad #1-5 to pop out of the system! Again, think of the haranguing NFL GM's had over whether to take those future elite QBs in their time period…or the consternation over selecting any college QB for that matter. My Top-5 (of the first 59 studied), would have 4 elite QBs...maybe 5 arguably.

Again, this Version 3.0 -- modifications are being made as more QB data is being analyzed and added.

Where is Michael Vick?

Vick had a bizarre final season in college (2000), he was on & off injured and only threw for 20+ passes in a game twice in 11 games played. Three games Vick had 11 or less passes, some of that due to in-game injury. In my system Vick has a 0.071 score, a likely bust for sure. Outside of 2009 in the NFL, Vick has been a "bust" as QB from a "stat" or passing QB level. His running ability and totals are amazing, but "elite" QBs are not "mobile" QBs...so this doesn't help his rating with us. Vick is such an anomaly and was injured his final college season, I have him "thrown out" of the list right now. If he was in, it would have said "bust"...and I think that is the right call, but is debatable. His final season wasn't really a fair representation of his ability due to injury, and that skews this for him.

What about the 2011 QBs?

I have tested 5 of them so far, however I am missing some key data on some (namely the Wonderlic scores). However, I can assume that data…and I will until I have it for sure. I will break down each incoming QB and add them to the master list one QB at a time. We will produce an article for each 2011 QB, breaking them down statistically – examining red flags and historically top flight numbers. Stay tuned for the first one on Missouri’s Blaine Gabbert.

Currently Sam Bradford is the highest rated QB we had, but I couldn't help running Andrew Luck through this system. Luck projects as the highest score I would have loaded into the system to date, if last season was his last and he was jumping to the NFL. More on Luck, and all the 2011 incoming QBs upcoming.

I am also adding other non-2011 QBs as I can throughout the off-season, as well as looking at new trends to add or modify. Updates will be posted all off-season.

 

 

 

 

Select a position from the tabs below to see stats and scouting information for that respective position.

• 
• 
• 
• 
• 
• 
• 
•