Assessing Parcells' four criteria

The biggest question mark for any team going into a draft is how to measure intangibles at the quarterback position.

Personnel legend Don "Duke" Klosterman may have put this quandary best when he told former Giants general manager Ernie Accorsi, "Do not evaluate a quarterback the way you evaluate the other 21 positions. They're playing a different sport. With a quarterback, it's the things you can't put down on paper that make all the difference."

The problem this presents is simple: Those items that can't be put down on paper all generally lend themselves to emotional swings; a perfect example of this comes via the case of Tim Tebow. Tebow's significant pocket passer weaknesses have ESPN draft experts Mel Kiper (fourth round) and Todd McShay (third round) grading him as a less-than-stellar choice, but Tebow's seemingly off-the-charts intangibles have others -- such as Tony Dungy -- rating him as a first-round pick.

One solution to this quandary is to try to find a way to measure the impact of these seemingly unmeasurable characteristics. This is the path that Bill Parcells took when he put together the following four rules for drafting quarterbacks:

1. Be a three-year starter
2. Post at least 23 wins
3. Be a senior
4. Be a college graduate

The idea behind these guidelines is simple: Any quarterback who achieves each of these goals has proved that he can handle the rigors of frequent play, has a history of winning and will stick to a goal until he finishes it.

These sound like terrific mile markers to gauge intangibles, but does this system work? Do quarterbacks with these characteristics significantly outperform players who don't reach these goals?

To find the answer to these questions, I enlisted the help of the Stats & Information department at ESPN. I asked its researchers to compile the figures for all of the first-round picks in the BCS era (since the 1999 NFL draft) along with every other current starting quarterback who was drafted in that time in the four aforementioned categories.

The first study we did revolved around the three-year starter guideline. The initial quandary here was how to determine how many starts constitute three years. Every FBS (formerly Division I-A) team today plays a minimum of 12 games -- but that level has been in place only since 2006. In the years before that, the minimum was 11 games, so we ran the study with both 33 starts and 36 starts as the qualifying line.

At the 33-start level, quarterbacks were 637-542-1 (54 percent) in the NFL; at the 36-start level, they were 606-515-1, which is a very similar percentage mark of 54.1 percent.

College quarterbacks with fewer than 33 starts were 493-528-1, for a winning percentage of 48.3 percent; those with fewer than 36 starts were 524-555-1, for a winning percentage of 48.6 percent.

The difference here is an NFL win percentage of around 5.6 percent, or about nine-tenths of a win per season (16 games multiplied by 5.6 percent). Not quite a slam dunk for the high-volume college starts criterion, but certainly a notable positive indicator.

Next up is the 23 collegiate wins mark. The findings here were quite similar to the high volume of starts study. The quarterbacks who had a minimum of 23 wins posted an NFL record of 664-575-1, or a winning percentage of 53.6 percent. The quarterbacks who posted fewer than 23 wins had an NFL mark of 466-495-1, or a winning percentage of 48.5 percent. The favorable percentage differential was 5.1 percent, or nearly identical to the 33-36 starts criterion.

The third study looked at how senior quarterbacks fared compared with sophomores and juniors who declared for the draft. The seniors posted a professional record of 914-891-1, or a winning percentage of 50.6 percent. The second-year and third-year collegians did not post anywhere near as high of a volume of games as the seniors, but their 216-179-1 record equates to a 54.7 winning percentage, a mark that is 4.1 percent better than that of their counterparts.

The fourth criterion, graduating college, is a bit tougher to gauge because there are some instances where extenuating circumstances need to be taken into account. One example of this can be found in the case of Joe Flacco. Flacco had to put his final college classes on hold because of an odd confluence of rules. Here, we ended up giving any players in situations of this nature the benefit of the doubt and crediting them in the graded category.

The findings here were quite notable. Players who graduated posted a 965-855-1 NFL record, or a winning percentage of 53.0 percent. Quarterbacks who didn't graduate tallied an NFL mark of 165-215-1, or a winning percentage of 43.4 percent. That is a difference of 9.6 percent, or an average of around 1.5 extra wins per year. To put it into perspective, that mark is 75 percent greater than the average extra wins generated by the high-volume starts or collegiate wins criteria.

Each of these individual bars is high on its own, but the Parcells rules require that a player meet all four. That should mean only the best of the best make the final cut, but the numbers do not indicate that is the case. The combined record of the 16 quarterbacks who satisfied all four criteria (all of whom had at least 36 starts) was 494-432-1, or a 53.3 percent winning percentage. That is lower than the winning percentages generated by two of the individual criteria, so this combination of traits cannot make a claim as being the most notable success indicator.

That honor goes to the college graduate criterion. To put it another way, when talent evaluators are looking for a tiebreaker in making a choice of which quarterback to draft, they should turn to the classroom for guidance.

KC Joyner, aka the Football Scientist, is a regular contributor to ESPN Insider. He also can be found on Twitter (@kcjoynertfs) and at his Web site. Research was provided by John McTigue and Marty Callinan of the ESPN Stats & Information Group.