Cleveland Sports Insiders

Ernie Camacho's favorite blog…

The Ellis Quarterback Metric



McCarron excited to be at the top of the Ellis QB Metric (RONALD MARTINEZ/GETTY IMAGES)

Over the past two years I have talked about and worked on a formula taking the stats of college quarterbacks and trying to make a metric to judge which players would be the best pro quarterbacks. The thought started a year ago when I watched the Browns Brandon Weeden struggle as a rookie while Russell Wilson excelled in Seattle.

While I knew the answer and I had a bunch of numbers circulating in my head, the question was whether or not I could find which indictors are the most important, and how could I weight them to find a formula that would accurately show that measure.

I started by entering the data of every quarterback drafted in the top three rounds over the last decade. I then added every starting quarterback in the league, and over the last two years, I entered every prospect with a draft grade over fifth round or greater. I tried at least five or six models before settling on the one I have now. While it’s not perfect, it does seem to indicate a few things.

So how does it work?

Well I am not willing to give away the formula, as I’ve spent too many hours on it to just post it for free. I can give you some general insight into its complexities. I looked at several indicators, and weighed them by different amounts. I looked at 80 quarterbacks of varying success levels in the NFL and based the weighted numbers on this. Each year, as I add more players, I adjust the formula accordingly. I took the total data and the five indicators that I found and created this formula. The average adjusted score of all the players in the formula is a 100. Of the 80 players I have data on, only 34 managed to have a score of 100 or greater.

The highest score in my system is a 146.084, the lowest is a -43.7. The formula itself has two steps to it. If after the first step the quarterback has a negative value, then this is tallied as his final score. The second step is only completed if a player has achieved a positive score.

Here are my major limitations. First, I can’t use players who didn’t play Division I as of yet, as the stats are much harder to find and often incomplete. Second, the game has changed and this is shown by recent players putting up video game numbers. I don’t use any compiled data, just rate data so it helps to balance out these changes to a degree. Unfortunately, it doesn’t change the fact Peyton Manning basically played a different game the one being used now.

If my formula does not like a player, typically they do fail. The bottom ten players of the compiled data are outright busts, as most failed to last four seasons in the NFL. There are 80 quarterbacks in the system and the bottom fifteen are all horrible misses; though in fairness, 18 and 19 are Jay Cutler and Matt Ryan, who have worked out pretty well.

The formula also does a great job finding quarterbacks who will end up being playoff caliber quarterbacks. There are only three names in the top ten who one has to question. The first is Sam Bradford, who I don’t think we can judge just yet with any finality, as injuries have set back his career. The other two are Geno Smith, who had zero weapons with the Jets last year and a coach who threw him into the fire, and the big miss in former USC Heisman Trophy winner, Matt Leinhart. For those who are curious, the rest of the top ten was Alex Smith, Aaron Rodgers, Robert Griffin III, Andrew Luck, Russell Wilson, Cam Newton, and Chad Pennington.

I made the average score 100, and was able to balance out the numbers to get there: this is the current QB ratings in each tier.

Tier 1 Potential Franchise QB’s

  1. AJ McCarron Alabama 136.73
  2. Derek Carr Fresno State 127.46

Tier 2 Above Average QB’s

  1. Johnny Manziel Texas A&M 111.75
  2. Teddy Bridgewater Louisville 110.56
  3. Aaron Murray Georgia 109.01
  4. Blake Bortles UCF 108.1
  5. Tajh Boyd Clemson 105.46

Tier 3 Below Average QB’s

  1. Zach Mettenberger LSU 98.91
  2. Brett Smith Wyoming 97.88

Tier 4 Avoid at All costs QB’s

  1. Stephen Morris Miami 85.16
  2. Logan Thomas VA Tech Negative Score
  3. David Fales San Jose State Negative Score

So as you can see, this is a big year for the formula, a lot of names utilizing my metric are much higher on the list than on most national lists. Manziel and Bridgwater are all close to being on the franchise level, but nowhere do I see Derek Carr or A.J. McCarron that high. In three years, Smith, Carr, and McCarron will either make me a genius, or force me to make wholesale changes to my metric.

My one concern is that the changes in college football might make it impossible to compare players from even five years ago. If nothing else I feel confident that this system will at least remain a great way to avoid busts, going forward.


4 thoughts on “The Ellis Quarterback Metric

  1. Would be nice to see how QBs have fared in the past 5 drafts with your rankings. Or perhaps that is to come in a future posting?

  2. I have fixed the rankings now, I based the cut offs on percentages, had to make the top 10% to be franchise, then next tier and so on. Everything is math based

  3. Nice work jellis. How did you set the cutoff for each grouping? What was the criterion?

  4. Actually just noticed a glitch in my data Fales should be in the bottom grouping

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s