Pro Football Hall of Fame coach Bill Parcells had a list of criteria when he drafted a quarterback.
Among them: The quarterback had to complete 60 percent of his throws.
Parcells judged on productivity, not on arm strength or mechanics. He wanted a quarterback who started and won a lot of games, who completed a high percentage of passes and who took care of the ball. In short, he wanted a guy who had shown he could play, not shown he had potential.
Parcells' belief system takes on added meaning as the 2018 NFL draft approaches and the Cleveland Browns try to judge between Sam Darnold, Josh Allen, Baker Mayfield and Josh Rosen for the first overall pick.
All have strengths; all have concerns. The Browns believe they can find their franchise quarterback in this group, though they hope to let him sit and learn as a rookie.
One number jumps out for one of the quarterbacks under consideration: 56.
That is Allen's completion percentage at Wyoming. In two seasons as a starter, Allen completed 56.3 and 56 percent of his passes. That number has led to intense analysis, with many questioning whether it should be disqualifying in a league where 60 percent is the minimum and others saying Allen's overall skills override the number.
ESPN draft analyst Mel Kiper continues to project Allen to the Browns with the first pick. His retort on the completion percentage: "Stats are for losers."
"The kid won," Kiper said. "You say what was his record? When he was out there, they won football games. The stats, a lot of guys have stats and can't get their team over .500."
To Kiper, Allen's size, his ability to read and move and his top-level arm strength combine to overcome any issues. The Browns do not discount the arm strength, which matters in a cold and windy city like Cleveland.
"He has a cannon," Browns coach Hue Jackson said last week at the NFL meetings. "He can really throw the football. I think [Cleveland general manager] John [Dorsey] said it. He threw a ball that hit a guy in the sternum and you could hear it go 'boom.'
"I mean, he can really throw a football. And when he does it all right from a biomechanics standpoint and his body being in line, everything, he throws it as pretty as I've seen."
The other contention on Allen's behalf is that he wasn't surrounded by talent at Wyoming.
"With Josh, he is a tougher guy to evaluate, as I talk to people, because a lot of his tape is irrelevant," Allen's quarterback coach, Jordan Palmer, told 92.3 The Fan in Cleveland before the combine. "His stats are kind of irrelevant because it's hard to evaluate when you're an eighth-grader playing with sixth-graders."
It's a fair point, but it also ignores the fact that he played Wyoming's schedule, in the Mountain West Conference, which isn't like playing in the SEC.
There is strong thought that with coaching and work on fundamentals in an NFL offense with a set of checkdowns, the completion percentage can easily rise to 60.
But Jackson said last week -- and has said before -- that although a quarterback can improve fundamentally, he usually is what he is when it comes to completion percentage.
"I think you said it, you never feel like that that's something that you can just totally fix," Jackson said. "I think you can help improve it, but fix it, that's kind of hard to do. But at the same time, I watched him improve with some fundamentals that he needed to work on to give himself a chance to be more accurate."
That meant taking a shorter stride and making sure his mechanics were consistent. That was Palmer's focus as he tutored Allen over the past few months, and the Browns said Allen was better at his pro day than he was at the end of the season.
"With Josh, it was footwork and accuracy," Jackson said. "You could see definite improvement in him, but again, that wasn't a game he was playing in. That was a workout on his campus. We got to push that to what it would look like in a game, but there's no question he's improved."
Ten years ago, an analytics study on ESPN.com showed that the best predictors for future quarterback success were the number of games played and completion percentage. The study even went on to suggest that while a high completion percentage does not guarantee success, it is a prerequisite for it.
That was in 2008. The question the Browns must answer this month: Why was Allen below 60 percent, and can he be different when he gets to the pros?
The Browns dealt with a 52 percent completion rate from rookie quarterback DeShone Kizer in a winless 2017; they must evaluate whether Allen's talent can overcome the percentage, which in some ways seems to have been an accurate reflection of how Allen threw the ball.
ESPN Stats & Information research indicates that Allen threw more short passes than the other top quarterbacks in this draft. Thirty percent of his throws were five yards or shorter in the air, more than Rosen (29), Darnold (28) and Mayfield (18). In theory, those should be high-percentage passes, which should increase his overall percentage.
Allen also did not have a high percentage of passes dropped. Wyoming's 4.2 drop percentage was barely higher than the national average of 4.1 and less than UCLA's 6.3 for Rosen.
Pro Football Focus' analytics and stats show that Allen's 56 percent number had little to do with the talent around him. PFF reports that Allen completed just 13 of 42 passes thrown downfield (more than 20 yards from the line of scrimmage) and 1-of-13 when throwing deep to the right.
PFF eliminated drops, spikes, balls knocked down and throws affected by the pass rush, and Allen completed 65.7 percent, which ranked 86th in the country.
He was 51.4 percent against pressure (79th) and 71.8 percent from a "clean" pocket (72nd).
Football Outsiders is another analytics site that has developed its own metric, which is called QBase, and it has been accurate in predicting how a quarterback will transfer to the NFL. Allen receives the lowest grade by far of the top draftable quarterbacks, and the site calls his arm "a howitzer without a target."
What does history say about sub-60 percent passers and the transition to the NFL? There clearly are several ways to look at it. Brett Favre was 52.1 percent in college and he's in the Pro Football Hall of Fame. Matt Stafford has had an excellent career in Detroit, and he was 57.1 percent at Georgia.
Stafford is the quarterback most often compared to Allen, though in terms of physical ability and arm strength there also are comparisons to Ben Roethlisberger, a guy the Browns famously passed on in the draft.
But in the NFL timeline, for every Stafford there are two or three Kyle Bollers or J.P. Losmans, guys drafted high who did not hit 60 percent in college and who did not go on to great NFL success.
Since 2004, NFL teams have drafted 46 quarterbacks in any round who finished their college career below 60 percent. Fifteen of them never threw a pass in the NFL and 23 more never got to 60 percent as a pro. One who did threw six passes in his career.
In the past 10 years, 48 quarterbacks have been drafted in the first three rounds. Ten were below 60 percent in college. Eight improved their percentage by 2 percent, and five improved by at least 5 percent. Of those five, four went on to some level of success -- Matt Ryan, Tyrod Taylor, Stafford and Jay Cutler. Three of the four were taken in the first round. None has won a Super Bowl, though Ryan played in one. The four have combined to play in 16 playoff games (10 by Ryan) in 38 NFL seasons.
Among the rest, Christian Hackenberg has yet to play for the New York Jets; Jacoby Brissett was traded from the New England Patriots to the Indianapolis Colts and is 4-12 as a starter; and Jake Locker and Colin Kaepernick are out of football -- though Kaepernick did go to a Super Bowl with the San Francisco 49ers. Kaepernick was a 58.2 percent passer in college and 59.7 in the NFL.
Focus can be placed on Stafford, but the rest of the names -- Freeman, Mallett, Henne -- are by and large quarterbacks teams cannot build on.
ESPN Stats & Information data indicates that only four sub-60 percent quarterbacks drafted since 2004 were taken in the first round (Freeman, Ryan, Stafford and Cutler). Ryan has improved to 64.9 percent and Stafford and Cutler to 62 percent. But even Ryan and Stafford had qualifiers. Ryan barely missed 60 percent in college (59.9 percent), and Stafford improved his completion percentage in each of his three seasons, from 52.5 to 55.7 to 61.4 in his final year at Georgia. Freeman's career never went anywhere; he finished a five-year career 25-36 as a starter with a 57.6 completion percentage.
The only other quarterback to start regularly in 2017 who was sub-60 percent in college: Josh McCown.
Most quarterbacks who are sub-60 percent in college simply do not translate well to the NFL; those who do have special skills. Allen might be the guy with the special skills.
Jackson said the decision will not be made to find the best guy today, but the best guy in 2019 and 2020 and beyond -- in other words, the guy who can learn the most from watching for a year, then grow into the job. In some ways, that describes Allen perfectly.
All of this means the Browns have a lot to ponder as they go over the best player for the first overall pick.
"There's not any piece of information that we're not going to wade through, because I think it's too important," Jackson said. "This is a huge decision for our organization. This is a time for us to get it right, and I think we'll do that."