In the April 19 issue of ESPN the Magazine, Peter Keating discusses a way for NFL teams to determine whether or not a first- or second-round QB will have a successful NFL career. (Insider subscription required.)

David Lewin, formerly an analyst for Football Outsiders and now with the NBA’s Cavaliers, has found that games started and NCAA completion percentage accurately predict NFL performance for QBs drafted in the first two rounds. To be more specific, the Day 1 QBs who go on to have the best pro careers complete at least 60% of their passes and start at least 37 games in college.

This theory has been around a while and when I was researching it last week, I stumbled across a similar article from 2007, so I’m not exactly sure how far back it dates.

Using college completion percentage makes a lot of sense. It seems intrinsically true that a QB that has trouble completing passes in college is also going to struggle with his accuracy in the NFL.

The number of starts is another strong stat. For a prospect to start 37 games, he basically needs to be a three-year starter. This indicates that he’s been around the block a few times, is reasonably durable and opponents have had a chance to game plan for him (and he’s still completed a high percentage of passes). It also seems reasonable to think that most good NFL quarterbacks wouldn’t have to be a backup for more than one season while in college.

Let’s take a look at the QBs selected in the first and second rounds of the ten drafts spanning from 1998-2007 to see if this theory still holds water.

First, let’s split the 37 first- and second-round QBs into four groups: a) those with a college completion percentage of 60% or greater and 37+ starts, b) a completion percentage of 60% or greater but 36 or fewer college starts, c) a completion percentage of less than 60% with 37+ starts, and d) a completion percentage of less than 60% and 36 or fewer college starts.

(Click on the graph to see a larger version.)

So, if the theory holds, QBs in the second quadrant (top right) would have the best chance to succeed in the NFL. There are a number of good QBs in that quadrant — Phillip Rivers, Donovan McNabb, Peyton Manning, Eli Manning, Ben Roethlisberger and Drew Brees. Chad Pennington and Daunte Culpepper were pretty good in their day as well, and Vince Young is still showing some promise. That’s nine out of 13 QBs that fit both pieces of the criteria.

Moreover, you don’t see a lot of good quarterbacks in the other three quadrants. Aaron Rodgers stands out, as does Carson Palmer and maybe Jay Cutler, but even he struggled mightily this season. That’s just three good QBs out of 25 prospects.

Which means there are a whole lot of busts. Akili Smith, Joey Harrington, Ryan Leaf, David Carr, Tim Couch, JaMarcus Russell, Rex Grossman, Patrick Ramsey, Marques Tuiasosopo, Quincy Carter, J.P. Losman, Brady Quinn…the list goes on and on. Kyle Boller is a great example of a terrible draft pick. The guy can’t complete more than 50% of his passes in college, but he’s able to throw the ball through the uprights while kneeling at midfield, so he ends up going in the first round.

Maybe it’s enough to eyeball these players and make a judgment, but let’s dig a little deeper and plot the QB rating and overall win percentage for those four groups. Here are the results:

(Again, click on the graph for a larger version.)

In terms of QB rating, five of the top six (and eight of the top 11) met both the 60% threshold in college completion percentage and the minimum number of starts. The only quarterbacks that didn’t meet both criteria but still won at least 50% of games and had a QB rating of 80+ are Aaron Rodgers and Carson Palmer.

So what does this mean for this year’s crop of signal callers? Well, looking back at the first graph, we see that Jimmy Clausen and Sam Bradford are both lacking the requisite number of starts. Bradford’s completion percentage is encouraging, but injuries during his senior season caused him to miss 10 starts. As long as this doesn’t indicate that he’s prone to injury, he should have a good NFL career. Jimmy Clausen is almost there in terms of starts and his completion percentage is pretty decent. Still he’s not in the top right quadrant, so the chips appear to be stacked against him.

Colt McCoy has an outstanding completion percentage, but Texas ran a ton of underneath stuff which boosted his numbers. Still, he’s accurate and he has plenty of experience, so the numbers say that he has a good chance to have a nice career. Tim Tebow is also in the top right quadrant, but with the changes to his throwing motion and with the way he runs the ball, he may not follow the traditional quarterback’s career path.

I asked our NFL guru, Anthony Stalter, if after looking at these numbers — would he change his approach to drafting a QB in the first two rounds? Here’s what he had to say:

It’s hard to argue with the numbers – they speak for themselves. When you look at the graph and see the names listed in the upper right quadrant of John’s graph, it’s hard not to buy into the theory that a high completion percentage and number of starts are good indicators of whether or not that player will succeed in the NFL.

It almost seems like it would be common sense for a team to be highly skeptical of player like Akili Smith, who started just 11 games in college. But yet, there he was, selected with the third overall pick in the 1999 draft because his athletic talents were off the charts. Same thing with Russell, who can reportedly throw a football 50 yards while sitting down, yet can’t figure out how to hit an open receiver 10 yards away while standing up.

Obviously there’s a lot more that goes into drafting a quarterback then whether or not he finished with a high completion percentage in college. As John noted, McCoy completed a ton of short-to-medium passes out of the shotgun that gave him inflated numbers. While he’s accurate, he doesn’t have a ton of arm strength and that could hurt him at the next level depending on what system he winds up in.

Which brings me to the point I always make around this time: matching players to the right scheme is key. McCoy probably wouldn’t excel in a pro style offense where he has to look off a safety and then use his arm strength to throw a 10-yard out route to the sidelines (the most difficult throw in the NFL). But based on his accuracy, his ability to take what the defense gives him and the fact that he gets the ball out of his hands quickly, he seemingly would be a perfect fit for the West Coast Offense. Plus, his football IQ is outstanding, which is something Manning (both of them), Tom Brady, Big Ben, Rivers and McNabb have as well.

But overall, this theory is sound and I’m sure it’s something intelligent GMs take a look at when it comes time to draft a quarterback in the first two rounds. Accuracy is huge, because it shows that the quarterback can read a defense, get the ball out of his hands quickly and find open targets. It should be more of a factor than how far a quarterback can throw from the seat of his pants, but unfortunately athleticism and arm strength still mesmerize many scouts and GMs around this time of year.

Photo from fOTOGLIF