College stats don't lie
There are plenty of numbers bandied about at NFL draft time -- 40-yard dash times, bench press reps, vertical jump heights and even Wonderlic test results.
All are familiar to most NFL fans. And yet, we rarely hear experts mention a player's college statistics. Most fans assume college stats are not accurate predictors of NFL performance.
That's not always the case. My research of highly drafted quarterbacks since 1996 found that two college statistics adequately predict future NFL performance: games started and completion percentage. In fact, where a quarterback is selected in the draft has virtually no bearing on his NFL success. Games started and completion percentage are far better than the scouts at determining how good a player will be.
Over the past 12 years, teams have repeatedly drafted players who haven't shown the ability to consistently complete passes at the college level, and these players have consistently failed. For some reason, scouts expected players such as Kyle Boller (48 percent), Jim Druckenmiller (54 percent) and Ryan Leaf (54 percent) to suddenly figure out how to complete passes once they hit the NFL. Having a high completion percentage (60 percent or higher) is no guarantee of success, especially if it was done in a small number of games in a fluky system (Tim Couch being a strong example), but it is a prerequisite for it.
As to why games started should be an indicator of NFL success, there is a fairly obvious explanation -- good players start games. No one knows a player better than his coach, and if a coach decides he's good enough to start as a freshman, that's a good sign. Playing time also provides experience, which is crucial to the development of a young quarterback.
However, there is a more complex reason why games started is an important indicator. In general, NFL scouts do an excellent job of talent evaluation when they have enough information. The more film that exists of a player, the easier it is to find weaknesses. When scouts don't get sufficient information, they place too much weight on "measurables" and off-field workouts, and make mistakes like Couch, Leaf or Akili Smith.
Sometimes, when a player starts a lot of games, scouts have enough film to figure out that he is truly a "system quarterback," and not an NFL prospect. That's why Kliff Kingsbury and Chris Leak were not drafted high despite strong college numbers. Because of the assumption that scouts can do their job with the right information, these projections apply only to quarterbacks chosen in the first two rounds.
What does this analysis tell us about this year's crop of young quarterbacks? Let's look at the four passers likely to be chosen on the first day of the draft.
Matt Ryan (32 starts, 59.9 percent completion rate)
Ryan, likely to be the only quarterback selected in the top 10, and perhaps the whole of the first round, has great physical tools and looks the part. He stands tall in the pocket and delivers the ball with authority. However, his collegiate stats are average. The completion rate is a little less disappointing when we consider Ryan played in a downfield passing NFL-style offense for a coach who completely abandoned the running game and let Ryan throw an absurd 654 pass attempts (second-most in the nation). I'm sure facing defenses that knew Boston College was a pass-first team hurt Ryan's senior numbers. His senior completion percentage was 59 percent, but he completed 62 percent of passes the previous two years when the team was more balanced.
Nonetheless, I would be wary of guaranteeing $20 million to a guy who was not stellar in college. Statistically, the most similar recent college quarterbacks were Patrick Ramsey and Rex Grossman, who were a little worse, and Eli Manning, who was a little better. That's not a great group of comparable players, and taking Ryan in the top five is a significant risk.
Brian Brohm (33 starts, 65.8 percent completion rate)
A year ago, Brohm topped Mel Kiper's first 2008 draft board. Now it is a question whether he will even go in the first round. What changed over the last year? The answer is nothing. Brohm's weaknesses -- arm strength and mobility -- are the same as they were a year ago. So are his strengths: touch, decision-making, pocket presence and a strong work ethic. You can question his durability, but he had the same injuries in his past when he was a projected No. 1 overall pick a year ago. However, Louisville struggled this past season, finishing with a 6-6 record. So if we punished quarterback prospects because they played on bad teams, nobody in Denver would own a Jay Cutler replica jersey.
Statistically, Brohm has a profile that points to success. His 33 starts are less than stellar, and the injuries that kept him from starting more should be cause for concern, but his career completion percentage is the stuff NFL stars are made of. It is possible Brohm could fail -- maybe he'll get injured again, perhaps the completion percentage will prove to be the product of a gimmicky college offense -- but if I could pick him in the second half of the first round, I would be awfully excited about it.
Joe Flacco (26 starts, 63.4 percent completion rate)
The numbers that Flacco put up at Division I-AA Delaware are quite good. Completing 63.4 percent of passes is impressive, even against lesser competition. The problem is that Flacco ended up at Delaware in the first place, which is the same reason why he started only 26 games.
Flacco initially went to Pittsburgh. He redshirted his freshman year, and was unable to beat out Tyler Palko for the starting job after Dave Wannstedt arrived as head coach the following year. Palko went undrafted last year. If Flacco is as good as scouts believe he is, why didn't Wannstedt see that he was better than Palko? Wannstedt may not be the best coach in college football, but it is hard to believe he was that wrong.
Flacco is not a bad prospect, but seems like the kind of guy you take a flier on the second day, not someone who should be getting a multimillion-dollar signing bonus.
Chad Henne (47 starts, 59.7 percent completion rate)
Henne is the type of player scouts never miss on -- a four-year starter from an elite college program. His 59.7 percent completion rate isn't bad, but it isn't that good either. Henne never displayed the consistency of an elite quarterback while at Michigan, but he did show an NFL arm.
Henne's college stats are quite similar to those of another big-time recruit from a big-time school who had an inconsistent career -- Carson Palmer. Palmer started 45 games and completed 59.1 percent of passes, but finished on a positive note by winning the Rose Bowl and Heisman Trophy and consequently went No. 1 overall. Henne didn't finish as strong, and isn't quite as good a prospect, but isn't as far off as some might think. After a nice performance at the Senior Bowl, he seems to have an outside shot at the first round, and would be a good pick there. He is a virtual lock to be at least an above-average professional.
David Lewin is a former Division III college quarterback and co-author of the upcoming Pro Football Prospectus 2008.