NFL Passer Ratings

Bryan just called me asking what I thought of Kerry Collins’ negative-seven yard passing performance yesterday. I pulled up the box score for the game to review his stat line (2 for 12 for -7 yards, 0 TDs and a pick) and said “well there have been worse performances.” After Bryan understandably expressed his confusion about my confidence in that remark I read the last number on the stat line: “as bad as it was he still had a 4.9 passer rating and we know that there are definitely players who have had passer ratings of 0.0″ to which we wondered just how bad a quarterback would have to play in order to earn that dreadful 0.0 mark. Well after some brief discussion on the matter Bryan pointed out that this could turn into a pretty good post to which I simply responded “I’m on it.”

Instead of having this post be solely about passer ratings of 0.0, which could certainly have an entire post written about it, I thought I would just take this opportunity to explain exactly how the NFL passer rating works. I’m not going to claim to be an expert on the subject but I do think I understand it significantly better than the average person. This is primarily because I love the QB rating system. It was created in 1973 by Don Smith, who was the Vice President of the Pro Football Hall of Fame, but the rating has been retroactively applied to previous years.

It takes into account the four main statistics when looking at a quarterback’s performance: completions, yards, touchdowns and interceptions. It then weighs them against the number of attempts. This way you can compare the numbers of players who had different amounts of attempts. For example if one quarterback throws 49 touchdowns in a single season and another quarterback throws 50 in a single season then it would appear at first glance that the second quarterback had a better year, but when you take into account (like the passer rating formula does) the fact that the second quarterback had 80 more pass attempts it’s suddenly not as clear cut. This “fictional” example is actually comparing Peyton Manning’s 2004 season to Tom Brady’s 2007 season, in which Brady broke Peyton’s record for most touchdowns thrown in a season, but had 81 more attempts. Peyton’s 2004 passer rating was 121.1, while Brady’s 2007 rating was 117.2. Those two ratings rank first and second all-time for a single season. Steve Young holds the record for best career passer rating with a 96.8.

I’ll jump into this by explaining the basic math behind the formula. This is a five-variable equation, we’ll use the letters C, Y, T, I and A for each.

C – completions; Y – yards, T – touchdowns, I – interceptions, A – attempts. Here is the formula:

NFL Passer Rating

I did my best to draw it up in Microsoft Paint

Now the first thing I feel the need to explain is that there are four components here. The part that is 100C/A – 30 all over 20 we will call the C-component, the part that is Y/4A – 3/4 we will call the Y-component, the 20T/A will be the T-component and the 2.375 – 25I/A can be the I-component. Each component to the formula has a floor and ceiling. The basic way to explain this is saying that the floor for each is zero and the ceiling is 2.375. If any given component of the formula has a value higher than 2.375 or lower than 0.0 the value is to be defaulted back to the maximum or minimum value (2.375 or 0.0) and this is why the 158.3 passer rating is deemed perfect.

The perfect passer rating is rare, having only been accomplished 56 times (four of which were in the post season) since 1960, which averages out to be less than once per year. The first qualifier for a player to be said to have had a perfect passer rating in a game is that he must have attempted at least ten passes, otherwise running backs who pass on a trick play, or punters who throw on a fake punt could be added to the list. The only other qualifier is that they must have a 2.375 in each of the components I mentioned before. This is the trickier part. For a player to get a 2.375 in the C-component they must have a completion percentage of at least 77.5. For the Y-component they must have an average of at least 12.5 yards per attempt. For the T-component a minimum of 11.875 percent of the passes must be for touchdowns. And finally for the I-component they simply must not throw a single interception.

Only 46 different players have had a perfect passer rating in a game since 1960, and only six have done so more than once. Peyton Manning has done so the most number of times with four, and the other five are Ben Roethlisberger, Kurt Warner, Craig Morton, Dave Krieg, and Ken O’Brien. 1989 was the only season in which two different quarterbacks for the same team threw perfect passer ratings in the same year. That year the San Francisco 49ers had both Joe Montana and Steve Young. Only four players have ever had a perfect passer rating and had their team lose that same game: Daryle Lamonica, Bobby Hebert, Mike Buck and Chad Pennington. Lamonica and Buck did not start in those games.

Ken O’Brien has the crown of the most attempts in a game with a perfect passer rating at 32 and is also the only player to ever do it while also throwing for 400 or more yards, which¬† was in the same game. It has only happened six times when a player has thrown for five or more touchdowns, Peyton Manning is the only of which to do so twice. Drew Bledsoe is the only player to have a perfect game as a rookie and Eli Manning is the most recent player to have a perfect passer rating, when he did it on October 11, 2009 against the Raiders. The only four players to have a perfect passer rating in the playoffs are Terry Bradshaw, Dave Krieg, Don Meredith, and Peyton Manning. Manning is the only player ever to have one in January.

Now for the opposite end of the spectrum, the 0.0 passer rating (which doesn’t have a cool nickname like “perfect rating” so I am deciding in this moment that from here on out it should be known as the “epic fail rating”) we start with the same initial qualifier; the player must attempt at least ten passes. And similar (except the exact opposite) to the perfect rating, the epic fail rating’s only other necessity is to have a 0.0 in all four components. In the C-component a player must have a completion percentage of 30 or less to get a 0.0, for the Y-component he must have an average of three yards per attempt or less, for the I-component a minimum of 9.5 percent of his passes must be intercepted and he can’t throw any touchdowns.

The epic fail rating is nearly as rare as a perfect passer rating, having only happened (I think) 66 times since 1960, which is just over an average of once per year. This has been done by 57 players in that time, with Terry Bradshaw, who had three career games of a 0.0 passer rating, being the only one to have done so more than twice. Seven other players did so twice though: Warren Moon, Joe Namath, John Hadl, Dick Wood, Dan Pastorini, Cotton Davidson, and Gary Keithley. Craig Morton is the only player to ever post an epic fail rating in a Super Bowl, when his Broncos lost to the Cowboys but both Tommy Thompson and Otto Graham had done so in NFL Championship games and Thompson’s is the first known player to ever record an epic fail rating in a win; his team won 7 to 0. There was one game on December 9, 1973 that both starting quarterbacks had an epic fail passer rating. Bob Lee of the Atlanta Falcons and Gary Keithley of the St. Louis Cardinals had 0.0 ratings; the Cardinals won 32 to 10, which makes Keithley one of only seven players to ever have an epic fail rating in a winning game.

Randy Fasani had an epic fail rating in his only ever career start, while both Tony Graziani and Gary Keithley joined Fasani for having epic fail ratings in their first career start, but both went on to start more games. In fact Keithley went one-up on the other two by having 0.0 passer ratings in his first TWO career starts and is the only player to do so in back-to-back games. Cotton Davidson is the only other player to have two epic fail ratings in the same season, though Rex Grossman almost joined them when he posted his epic fail rating in 2006 and had a 1.3 rating earlier that season. Both Dave Wilson and Don Gault did so in the last games of their careers; talk about not going out on top. Twice have two quarterbacks for the same team thrown epic fail ratings in the same game, and both times were on December 12. Richard Todd and Joe Namath in 1976 against the Bengals and Cliff Stoudt and Terry Bradshaw in 1982 against the Bills. So if it’s December 12 and your quarterback is having an off day, don’t chant for the backup, because he’ll likely do just as bad. Eli Manning also had his only career epic fail rating on December 12.

The NFL passer rating is also used for the CFL (Canadian Football League) but the NCAA has an entirely different one that I honestly don’t understand at all. Though if there were one thing I could change about the NFL passer rating it would be to eliminate the floor and ceiling marks of 0.0 and 2.375. Let players have negative passer ratings and ratings in the 180s and higher. That way we could compare to determine which perfect passer ratings were the most impressive and which epic fail ratings were the most embarrassing. I understand that the floor and ceiling are in place to make all four components weigh in equally but I think they should do both. Call one the passer rating and one the adjusted passer rating or something because it would give us a whole extra angle to look at this from. If you have any questions, ideas or thoughts in general about the NFL passer rating formula/system feel free to leave a comment.

About these ads

13 Responses

  1. Good lord, you went way above our simple conversation.

  2. Haha, yes I did… It took me the better part of the day, but I think it was worth it.

  3. Nice post!

    Although, admittedly, I’m dizzy and must go to bed after reading through the Passer Rating Manifesto! That was totally awesome and equally well constructed. I learned a great deal, plus I think you’ve coined a new term: Epic Failure Rating. LMAO on that one!

    And I also agree that there should be no floor or ceiling. A negative rating would be a hoot!

    So, does this mean you will do your own with the expanded elements in the formula?

    • Yes, I actually have a spread sheet already set up (no joke) where I can plug in the passing attempts, completions, yards, touchdowns and interceptions for it to kick out the passer rating without a floor or ceiling.

  4. Dude, that was a massive, awesome statistical explanation. Very thorough and detailed, though equally complex and reckoning. You should expose this to rivals.com and espn as well. Personally, I don’t like the passer rating system, it has its flaws and can be deceiving- when you compare some ratings to what a teams record is. Below the top 5 rating/standings the ratios just don’t jive. To me, if a QB is going to be rated by his performance it should directly tied to the teams performance as well. But, I’m in the minority on this one. Regardless, nice job.

    • Thanks man, this post took a lot of work. I’m really glad you appreciated it. Please keep coming back to the site, we post new stuff all the time.

  5. BTW-
    Just wanted to share my appreciation for the Colts too. I have been a fan since 1975, watching Bert Jones, Lydell Mitchell and Roger Carr. To me the Colts will always be the Baltimore Colts. Today, Manning is an amazing QB, very technical, something I am in awe of- with Johnny U like qualities as well. Old school- I think Bert Jones is one of the most under rated QB’s to ever play the game. Some folks recognize his success, but because his career wasn’t too long and because of the dominance of the Steelers and Raiders- Baltimore flew under the radar. The 1976 Colts were one of the best teams in the NFL and so was Jones- MVP too.

  6. Great post! I’ve always loved the passer rating system, too. I found a cool little site that lets you adjust the weight of the four components in the formula:

    http://www.qbranking.com

    Cool stuff!

  7. [...] http://classic17.wordpress.com/2009/10/19/nfl-passer-ratings/ [...]

  8. Excellent explanation. A couple of things I noticed…

    Only 5 of the so-called perfect games have been on the road and none of those since 1972.

    It appears capping of components affects TD’s and Int’s the most.

    The C component has a formulaic-induced cap of 3.5 and the highest (uncapped) among perfect games is 3.222.

    The Y component max (uncapped) among perfect games is 4.214.

    I think credit should be given for achieving excellence with more attempts. Isn’t 20 of 20 completions better than 10 of 10? Isn’t throwing 20 attempts without an interception better than throwing 10?

    Assuming the four component strategy, I think the formula should be tweaked so that each component is maxed at 25 and a true perfect game would then be 100 since most of us are base 10 kind of folks.

    http://en.wikipedia.org/wiki/List_of_NFL_quarterbacks_who_have_posted_a_perfect_passer_rating

  9. Great analysis. Good work on a difficult subject to explain (which isn’t that hard to understand, people just don’t take the time to figure it out).

    I have to disagree with the common problems people have with the formula and thinking it needs to be “tweaked”. The QB Rating is a formula for efficiency. It is one of several tools that can be used to rate or rank QB’s (not the end all). There are many other formulas that include more data (like winning) to rate or rank QB’s. Don’t try to make this formula more than it is. The goal of it is to rate efficiency, not tell you definitively which QB is “better”.

    The perceived min/max issue is not a bad thing but a good thing. Besides, it only affects single game performance (or short windows of time). If you take away the max, the career QB Rating of every qualified QB (min of 1500 att) would be exactly the same. They wouldn’t change at all!! Being that proficient in 1 category is not possible in the long term.

    To be efficient, you have to be good at all phases of the job. When comparing single game performances, should over-proficiency in 1 particular area overshadow weakness in another? Having a max exposes those flaws in a QB’s efficiency. Without it, you’d get a false sense of how efficient a QB is at his job.

    Now, if you want to argue the value to be used for the min/max, that’s a discussion worth having. 1973 was a long time ago and the game has evolved A LOT since then. Though, I have a feeling that if you really analyzed all 4 phases of the game that are taken into account in the QB Rating, you’d find the max levels wouldn’t need to change that much to still be accurate as a ceiling for perfection in each category.

    (BTW: take away the max and the new “perfect game” would be 831.3…that’s a level of perfection that is unachievable and the rating scale would be so out of whack it would be hard to even use anymore.)

    Regarding # of attempts…I see your point…20 for 20 is more impressive than 10 for 10, but how does that relate to efficiency? Depending on the type of offense and game score will determine how many throws in a game you will have. Regardless of the situation, either you were efficient or you weren’t.

    For sh*ts and giggles, let’s say a Dolphin QB playing in a predominantly Wild Cat offense goes 10 for 14 for 225 yards, 3 TD’s, and no picks (he completed 7 passes for only 5 yards a piece, but also connects on 3 bombs for TD’s, mostly because of the gimmick offense). On the same day, Drew Brees goes 31 of 40 for 398 yards, 3 TD’s, and no picks.

    Which QB was “better”? Most would say Brees. Which QB was more efficient in their respective offense? No matter how you slice it, the Dolphin QB was more efficient. The question remains, “how much more efficient?”

    The “uncapped” QB Ratings would be 200.0 to 131.1…in favor of the Dolphin QB (53% difference). As the QB Rating stands now, it would be 153.3 to 131.1 (17% difference). Doesn’t that sound more reasonable? The max limits the outliers. The yards per attempt and TD% are ridiculously high, that doesn’t make the Dolphin QB that much more efficient on a comparative level.

    This example is obviously rare…but if the max was removed from the QB Rating formula, the amount of times you’d see QB Ratings higher than the current max of 158.3 would also be rare. It’s everything in the middle that would be skewed by performances of the exceptional nature in only 1 component of quarterbacking.

  10. [...] for 503 yards, five touchdowns and no interceptions and he was never sacked. That’s a passer rating of 111.8 and we’re talking about situations where the defense knew he was going to pass but [...]

  11. Hi to every one, it’s actually a good for me to go to see this web page,
    it consists of important Information.

Leave a Reply

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

WordPress.com Logo

You are commenting using your WordPress.com 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 )

Google+ photo

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

Connecting to %s

Follow

Get every new post delivered to your Inbox.

Join 179 other followers

%d bloggers like this: