Behind the Numbers 10.12.07: The Quarterback Rating
The quarterback rating system is widely used in the NFL today. Anytime somebody wants to praise or discredit a quarterback, you will hear about their rating as if it were an accurate statistic. This sounds great when you’re bashing a quarterback and they have a horrendous rating, such as Ryan Leaf and his 50.0 rating. It also sounds legit when you’re praising Peyton Manning and mention that his career rating is a 94.8. Unfortunately this rating system isn’t bullet proof and, if you were to go squarely by the numbers, it can make a bad quarterback sound great and a great quarterback sound horrible. For instance, Billy Volek has the 12th best rating for quarterbacks with at least 13 games experience with a 86.8. Does anyone really think that he is the 12th best quarterback in the league? As for Vince Young (I can’t wait to hear my friend, Jason, respond to this) he has a 66.8 rating which ranks him 52nd in the league. In other words he would be on the lower rung of second-string quarterbacks if you went strictly on rating. Granted he’s not in the top tier of quarterbacks in the league, but he leads his team to victory quite often, so the rating doesn’t reflect his play at all. If that doesn’t convince you, then take a look at Brett Favre. Brett Favre ranks 16th in the league for career ratings. That’s right, two Super Bowl appearances, three consecutive MVP awards, eight Pro Bowl trips, and six All-Pro selections, to go along with almost every major passing record, but your quarterback rating is going list you 16th?
I understand that statistics don’t always tell the story, and that’s fine and well, so why create the quarterback rating in the first place? The name itself implies that it is intended to rate a quarterback’s performance, which it does a poor job of. So how does this rating work? You’ll be hard-pressed to find anyone that knows the formula. I had no clue what it consisted of until I began researching it. The information you need is:
1) completion percentage (C)
2) yards per attempt (Y)
3) touchdowns per attempt (T)
4) interceptions per attempt (I)
So far, so good, right? Right off the bat you can see some holes, but they are fairly minor. For instance, a quarterback in a west coast offense is going to have a much lower Y and T values than a quarterback in an offense that attacks downfield, but they will probably have a higher C value. But again, these are fairly minor issues with the system so far. Let’s take an imaginary quarterback and his output for one game and quickly get his rating. Let’s start with his C value. Unfortunately the C value is not simply 75.2 if their completion percentage is 75.2%, you must enter it into a formula to get the C value. You take the completion percentage of 75.2 and subtract 30 from it so it becomes 45.2 and then divide that value by 20 so it becomes 2.26. Don’t ask the why, because I have no clue. A 2.26 is pretty good, because the max value for any of the variables is 2.375. So anyhow, at this point C = 2.26.
Now onto the Y value; let’s say the yards per attempt (ypa) is 7.4. Take the ypa and subtract 3 from it (7.4 – 3 = 4.4) and then divide that value by 4 (4.4 / 4 = 1.1) When you consider the fact that a 7.4 ypa is very good, but it gives you a poor Y value, you see some more holes in this formula. Tom Brady is tearing defenses apart with an 8.75 ypa and that would only give him 1.4375 which is far away from a 2.375. You need a 12.5 ypa to get the max 2.375. But back to our quarterback, his Y = 1.1.
For the T value, you will take the touchdowns per attempt and multiply it by 20, so let’s say he threw 30 passes and had 3 touchdowns, his touchdowns per attempt would be 0.1. So 0.1 * 20 = 2. This value seems pretty fair overall, except when looking at the west coast vs long ball offense. But again, our quarterback’s T = 2
To get his I value, let’s say he threw 1 interception in this game. One interception out of 30 passes gives him a 0.333 interceptions per attempt. Take that value and multiply it by 25 (0.033 * 25 = 0.825) and then subtract that number from 2.375 (2.375 – 0.825 = 1.55). Out of all of the values, I think this one is the most fair as 1.55 is pretty close to 2.375 for giving up one interception. So our quarterback has the value of I = 1.55
So as you can see, this didn’t go near as quickly as we had hoped. But we’re almost done, so no need to stop now. Let’s review the values that we have figured out:
C = 2.26.
Y = 1.1.
T = 2
I = 1.55
The last step is to add all of these values (2.26 + 1.1 + 2 + 1.55 = 6.91). Take that value and divide it by 6 and then multiply that result by 100. (6.91 / 6 = 1.1516) (1.1516 * 100 = 115.16) So we now know that our quarterback has a 115.16 rating for this game in which he had 222 yards passing, 75.2% passing completion, 7.4 yards per attempt, 3 touchdowns and 1 interception on 30 attempts. Not a bad days work! If you’d like to play around with a quarterback rating calculator, visit this site: www.primecomputing.com
After looking at how the rating is figured, I have only one question: Why? The formula is far from accurate, and no other position has a rating system. The system was created based on historical averages, but now that the NFL has changed the rules to increase scoring, it needs updating to say the least. Of the 30 quarterbacks in the Hall of Fame, 17 of them have a rating below 80.0 and that includes John Elway! Yes, I know the game has changed quite a bit, especially with passing, but this just goes to show you how this system is not worth using. In case you’re a statistics dork like me, here is the list of Hall of Fame quarterbacks and their ratings:
Steve Young 96.8
Joe Montana 92.3
Jim Kelly 87.1
Otto Graham 86.6
Dan Marino 86.4
Warren Moon 84.2
Roger Staubach 83.4
Len Dawson 82.6
Sonny Jurgensen 82.6
Troy Aikman 81.6
Bart Starr 80.5
Fran Tarkenton 80.4
Dan Fouts 80.2
John Elway 79.9
Johnny Unitas 78.2
Bob Griese 77.1
Norm Van Brocklin 75.1
Sid Luckman 75.0
Y.A. Tittle 74.3
Sammy Baugh 72.2
Terry Bradshaw 70.9
Joe Namath 65.5
Bobby Layne 63.4
Bob Waterfield 61.6
George Blanda 61.1
Clarence Parker 53.0
Earl “Dutch” Clark 40.3
Jimmy Conzelman, Benny Friedman, Paddy Driscoll did not have enough recorded statistics to properly rate them
As for current quarterbacks, here are their career ratings. Again, I have only included quarterbacks with 13 or more games played:
Peyton Manning 94.8
Tony Romo 94.7
Kurt Warner 94.0
Carson Palmer 91.4
Daunte Culpepper 90.8
Tom Brady 90.3
Marc Bulger 89.7
Philip Rivers 89.4
Chad Pennington 89.3
Ben Roethlisberger 89.2
Jeff Garcia 87.1
Trent Green 86.9
Billy Volek 86.8
Matt Hasselbeck 85.4
Drew Brees 85.4
Brett Favre 85.3
Donovan McNabb 85.3
Jake Delhomme 85.2
Brian Griese 84.3
Mark Brunell 84.2
Jay Cutler 84.0
Damon Huard 83.5
Steve McNair 83.1
Brad Johnson 83.1
David Garrard 83.1
Jamie Martin 82.3
Matt Schaub 82.2
Byron Leftwich 80.0
Kelly Holcomb 79.2
Jason Campbell 79.2
Kelly Holcomb 79.2
Charlie Batch 78.6
Jon Kitna 76.9
JP Losman 76.8
Gus Frerotte 75.2
David Carr 75.2
Patrick Ramsay 74.9
Seneca Wallace 74.3
Eli Manning 74.2
Kerry Collins 73.2
Derek Anderson 73.1
Josh McCown 72.4
Kyle Boller 71.7
Matt Leinart 71.2
Chris Simms 71.2
AJ Feeley 71.1
Trent Dilfer 70.8
Todd Collins 70.4
Charlie Frye 70.2
Joey Harrington 69.4
Rex Grossman 69.3
Vince Young 66.8
Anthony Wright 66.4
Alex Smith 65.7
Bruce Gradkowski 64.9
Ken Dorsey 63.5
Trent Edwards 62.0
Kyle Orton 59.7
Tarvaris Jackson 53.3
And finally, a little record information on the quarterback rating. Steve Young set the NFL record for the highest career rating with 96.8. The single season record of 121.1 was set by Peyton Manning in 2004. Also in 2004, Ben Roethlisberger set the rookie record with 98.1.
So now that math class is over, what are your thoughts on the quarterback rating system? Please feel free to send feedback to chris@phatdragon.com Also feel free to contact me if you have any players or topics that you would like to see covered on Behind the Numbers.
