According to the NFL’s official passer rating system, the most efficient quarterback in NFL history is Green Bay’s Aaron Rodgers, with a lifetime mark of 103.1.1 That makes sense: Rodgers is generally regarded as one of the greatest QBs to ever play the game. But if you scroll further down the list, the results become much harder to explain. In the world of passer rating, Kirk Cousins is better than Joe Montana; Derek Carr and Matt Schaub top Dan Marino; and, after one season, Broadway Sam Darnold is running circles around Broadway Joe Namath.
Passer rating is often criticized as Byzantine (have you seen that formula?), incomplete (it does not include data on rushing plays or sacks) and arbitrary (again, have you looked at the formula?). Yet its biggest shortcoming might be the way it is unmoored from changes in the game itself. Passing has never been more efficient than it was this season, in which the league’s average QB posted a rating of 92.9. That is remarkably high considering that a quarterback who posted a rating of 92.9 would have led all qualified passers in 15 separate seasons from 1950 through 1986. Clearly, the scale needs recalibrating.
In the original conception of passer rating, an average rating was about 67. In 2018, only one qualified passer (Arizona Cardinals rookie Josh Rosen) fell below that threshold, and even then just barely (his rating was 66.7). But what if the standards for what makes a good or bad performance had evolved as leaguewide numbers changed? Pro-Football-Reference.com does a great job of adjusting for era with its Advanced Passing indices, which are centered on an average of 100 with 15 points representing 1 standard deviation in either direction. But I wanted to rescale the building blocks of passer rating itself to see how today’s passing numbers would translate to a rating if the NFL had simply allowed its rating system to change with the times.
To do that, I looked at the distribution of stats in each category that goes into passer rating — completion percentage, yards per attempt, touchdown percentage and interception rate — from the sample originally used to craft the formula back in the early 1970s (qualified passers from 1960 to 1970). Specifically, I figured out the spread of values (relative to the league) that, in a given category, led to the minimum number of points (0), the average number of points (1) and the maximum (2.375). Under the hood, passer rating is built around these ranges; it hands out points on that 0-to-2.375 scale in each category, then sums up the four values, divides by 6 and multiplies by 100. (Hence, 67 is supposed to be average — a 1.0 in four categories, divided by 6, times 100.)
For any era, we can rescale what performance “should” lead to a given value in each category to keep the relative leaguewide distribution the same as it was when passer rating was first conceived.2 So while, say, Alex Smith’s 62.5 percent completion rate in 2018 was worth 1.0 point, so was Don Meredith’s 49.5 percent mark from 1962. Do this for every category in every season, and you have a stabilized version of passer rating that no longer spirals uncontrollably upward with each innovation in the passing game.
Some ultra-high ratings change less than you might expect under this new method. Rodgers’s single-season record of 122.5 from 2011 tumbles all the way down to … 121.1. (He was very good that year.) But other seemingly immortal ratings, such as Kirk Cousins’s 99.7 mark this season, get knocked down quite a bit — in Cousins’s case, he falls to a much more reasonable 81.5 rating. (Anyone who watched a Vikings game this year would surely argue that this is more appropriate.) Similarly, Ryan Fitzpatrick’s 100.4 mark this season — yes, that is real, look it up — gets heavily penalized in the interception category (his 4.9 percent INT rate was more than double the league average), taking him down to an adjusted rating of 77.4.
Moving further down the list, Joe Flacco’s decent-sounding 84.2 classic rating properly falls to a mediocre 62.8 after our adjustment, while the 30.7 rating of WOAT candidate Nathan Peterman becomes an 11.6 — perilously close to the minimum possible rating of 0.0. (If Peterman had thrown enough passes to qualify, that 11.6 rating would have “surpassed” Ryan Leaf’s 19.1 from 1998 as the lowest-rated season since 1950.)
All told, the new ratings are once again grounded in a world where an average quarterback scores about 70 — not exactly 67 because the rolling distribution includes multiple seasons for comparison3 — and as a result, the numbers make far more intuitive sense at a glance than the ludicrously inflated official ratings of 2018:
According to the NFL’s official system, there have been 93 qualified quarterback seasons since 1950 with a passer rating of at least 100.0, and nine of those happened in 2018 alone. After our adjustment, though, there have been only 46 such seasons since 1950,4 and only one of those happened this year — the 103.3 mark Drew Brees put up with the Saints. It’s still a golden age for passing, as nearly half of those 46 seasons have happened since 2000, but we’ve also filtered out 51 “false 100s” — seasons that cracked 100.0 on the old scale but not the new one — of which 47 have happened since 2000.
The result of our passer rating adjustment is a much more reasonable career leaderboard that features qualified quarterbacks from a variety of different eras:
|Player||Last Year||Old||New||Player||Last Year||Old||New|
|1||S. Young||1999||96.7||94.2||16||F. Tarkenton||1978||80.4||80.7|
|2||A. Rodgers||2018||103.1||92.5||17||B. Starr||1971||80.5||80.7|
|3||J. Montana||1994||92.3||90.0||18||P. Rivers||2018||95.6||80.5|
|4||T. Brady||2018||97.6||87.2||19||C. Pennington||2010||90.1||79.9|
|5||P. Manning||2015||96.5||87.1||20||M. Ryan||2018||94.9||79.8|
|6||R. Staubach||1979||83.4||86.7||21||J. Garcia||2008||87.5||79.6|
|7||R. Wilson||2018||100.4||85.4||22||B. R’lisberger||2018||94.3||79.0|
|8||D. Brees||2018||97.7||85.4||23||J. Unitas||1973||78.3||78.9|
|9||T. Romo||2016||97.1||85.0||24||D. Fouts||1987||80.2||78.4|
|10||O. Graham||1955||78.2||84.7||25||R. Gannon||2004||84.7||78.4|
|11||K. Warner||2009||93.7||83.7||26||B. Griese||1980||77.1||78.3|
|12||S. Jurgensen||1974||82.7||82.9||27||N. Lomax||1988||82.7||78.1|
|13||L. Dawson||1975||82.9||82.7||28||F. Ryan||1970||78.0||78.0|
|14||D. Marino||1999||86.4||81.4||29||B. Jones||1982||78.5||78.0|
|15||K. Anderson||1986||81.9||81.2||30||J. Kelly||1996||84.4||78.0|
The biggest beneficiaries of our changes are 1950s-era passers like Otto Graham, who originally rated in the 70s (discarding his eye-popping pre-1950 numbers, which were compiled in the upstart All-America Football Conference) but leaps up into the mid-80s after judging him in comparison with his peers. San Francisco 49ers legend Steve Young also gets a boost relative to other great QBs from history, reclaiming the No. 1 slot that he’d held in real life before Rodgers and friends came along.
At the other end of the spectrum, nobody loses more points of career rating than Blake Bortles, who somehow has an 80.6 mark under the classic system but falls to 55.2 with our adjustments. Here are the biggest losers between the old and new QB ratings:
A change like this wouldn’t fix the rest of passer rating’s deficiencies, and it wouldn’t include all the fancy bells and whistles you’ll find in a metric like ESPN’s Total Quarterback Rating. But passer rating itself has always been a surprisingly decent metric within any self-contained era; the team with the higher passer rating (by any margin) in a game wins about 80 percent of the time. It’s the comparisons across eras that have become distorted as the game has changed over time. But a simple fix tethering modern stats to the standards contained in passer rating’s formula would go a long way toward restoring sanity to the metric you still see in every NFL box score and broadcast. The Blake Bortleses of the world might not like seeing their shiny 80-something ratings get dumped into the 50s, but it’s a change whose time has come.
Check out our latest NFL predictions.