Taking the long view of NFL quarterbacks

Comparing NFL quarterbacks to each other has become almost as big an industry as the NFL itself. Endless articles (and this is another) argue the merits of individual QBs. Fans debate how to rank the current crop of QBs, whether so-and-so is “elite” or a “franchise quarterback,” how much a team should pay for their present QB, whether they should try to draft or trade for another team’s QB, and so on.

There’s no definitive way to settle these arguments. But there is a fun way to think about them. FiveThirtyEight has an Elo rating system, named after the early 20th Century physicist Arpad Elo who developed it to rate chess players. Elo models have been used in many ways you may have encountered but not realized. For example, FIFA uses an Elo model to provide rankings for national soccer teams. For a while, even Tinder (!) used an Elo model. NFL Elo uses various standard quarterback statistics to assess each QB as a function of time, adjusts for how offense and scoring have changed between different NFL eras, and ultimately assigns a value for each QB at each point in his career relative to an average starter. Thus, Elo can be positive or negative.

No model or statistic can perfectly answer the question of whether one QB is better than another. But the cumulative Elo of any QB plotted vs. time can be instructive about how QB careers evolved, why one QB wound up with better career numbers than another, how currently active QBs are trending, and so on.

Robby Greer (@greerreNFL) has developed an NFL Elo App for this purpose. If you write code and want to know the gory details, you can find them here. I first got interested in the app when @Doug_Analytics posted this example involving Daniel Jones:

Yikes, Blake Bortles. Let’s hope for a better future than Bortles had after that almost-Super Bowl season with the Jaguars. But let’s ask some other interesting questions and circle back to Daniel Jones at the end. For the record, though, Greer also converts his Elo values for the season into a points added relative to the average starter metric. Here are the top 10 for 2022:

Courtesy of NFL Elo App

So, like metrics such as ESPN’s QBR, Elo was impressed with Jones’ 2022 season.

Who is the GOAT?

Data from NFL Elo App

These (as far as I can tell) are the five highest cumulative Elo totals in NFL history. It’s common these days for NFL fans to refer to Tom Brady as the best QB of all time. But actually Peyton Manning, probably the favorite of those who don’t like Brady, has a comfortable lead over Brady in career Elo. And Drew Brees, always mentioned with the greats but rarely considered THE greatest, is only a hair behind Peyton.

One big difference between Manning and the other two is that Peyton started immediately for the Colts (in fact he started every game of every season of his career other than the 2011 season he missed completely with injury and his final season in Denver). Brees and Brady on the other hand didn’t start as rookies, and neither added much value their first few years: Brees, not until he was traded to New Orleans and played in a Sean Payton offense; Brady, not until Josh McDaniels and Bill Belichick revolutionized NFL offense in the 2007 season. That’s one thing not often mentioned - how even the most talented quarterbacks depend on great offensive minds to tap their full potential.

Of course, Brady quarterbacked three Super Bowl teams before he even became the prolific passer we think of today, he has seven rings overall, which no one can match, and he would have had nine if not for a certain other NFL team. That is not captured by Elo and certainly it counts when assessing who the GOAT is.

The other QBs in the top five, Dan Marino and Aaron Rodgers, have remarkably similar career arcs. The only difference is that Rodgers’ career Elo number is consistently a bit lower than Marino’s because Rodgers mostly sat behind Brett Favre his first three seasons. Both QBs’ performance flattened over time, i.e., as Elo sees it, Rodgers has become more and more like an average QB - not adding much value anymore - in recent years. (Take note, New York Jets.)

Who is the greatest Giants QB of all time?

The New York Giants have had many QBs in their long history. Let’s narrow ourselves to those who started for Big Blue for at least four seasons. Even then, there are too many to fit on one figure, so we’ll break them up into 1950s-1980s and 1980s-2020s.

Data from NFL Elo App

The Giants have two QBs in the Hall of Fame. Y.A. Tittle spent his first two seasons in the All-America Football Conference and didn’t start consistently during a number of seasons in the NFL, so he has only 133 starts in his Elo record despite being active for 17 years. Fran Tarkenton played only one more year than Tittle but started 239 games. During the times their careers tracked each other they were remarkably similar until Tittle’s final 1964 season, when the entire Giants team collapsed. Tarkenton’s career was remarkable in its consistency, an almost linear upward Elo trend indicating that he consistently added value until the very end of his career. (Compare to Marino and Rodgers, who experienced a gradual decline over many years.) Tarkenton’s cumulative Elo is only slightly less than Rodgers’, yet his name is hardly ever mentioned in discussions of the greatest of all time.

Tarkenton was only a Giant for five years, but he is the best QB ever to put on Giants blue. Tittle was great but for a shorter time period. His predecessor, Charlie Conerly, was QB during a more run-oriented NFL era. Phil Simms will always be remembered as the QB who brought the Giants out of the wilderness years, but he was shaky in his early years (negative Elo), was even benched for Scott Brunner for a while, and was never a high-volume passing quarterback.

Data from NFL Elo App

For the more recent era, we repeat Simms’ Elo curve for comparison to the QBs that followed him. Kerry Collins was a mostly unsuccessful QB in the NFL except for his time with the Giants (games 50-117 of his career), during which his Elo curve slopes upward, meaning that he added value to those Giants teams relative to an average replacement. Not coincidentally, during his Giants tenure Collins was under the guidance of an offense-minded head coach in Jim Fassel and (for much of the time) a brilliant offensive coordinator, Sean Payton. Collins’ loss was Drew Brees’ gain.

Eli Manning has the strangest career arc of any QB I have yet seen in my playing around with the Elo app. Most QBs have careers that stay fairly flat or consistently increase or decrease. Some change slope gradually over time when they change teams or offenses (like Collins or Brees), when they slowly improve over time, or in the other direction as they age. Manning’s Elo curve sums up in one figure the controversy about whether he will be admitted to the Hall of Fame. He did not add value his first few years, then beginning around game 55 (the end of the 2007 season), his performance added fairly consistent value through the 2012 season, before declining again. This period spans most of the time that Kevin Gilbride was the Giants’ offensive coordinator, running the Erhardt-Perkins offense with WRs such as Amani Toomer, Plaxico Burress, Hakeem Nicks, Victor Cruz, and Mario Manningham as deep downfield threats. It was also an era during which the Giants often had a good offensive line.

The offense sputtered in 2013, though, and Eli’s performance reflected that (only 18 TD passes and 27 INTs), as seen in the downward trend in Elo. This led to the dismissal of Gilbride and the hiring of Ben McAdoo as his replacement. Eli’s Elo shows that he once again began to add value during the 2014-2015 seasons (which happen to coincide with the start of the Odell Beckham Jr. era) before leveling off in 2016 and beginning another downward trend to end his career. Despite the two Super Bowls, the inconsistency of Manning’s Giants career prevents him from being considered above Tarkenton as the greatest Giants QB ever if we consider the totality of both QBs’ careers.

How does Daniel Jones compare?

Which brings us finally to Daniel Jones. The Elo curves above make several salient points about Jones’ past, present, and future:

• Taken as a whole, the first four years of Jones’ career have been very much like those of Simms and Manning - all three of them did not add net value to their team over the complete four-year period.
• Jones’ Elo evolution was relatively flat during his first season under Pat Shurmur, then declined significantly during the Joe Judge-Jason Garrett-Freddie Kitchens era, and rose sharply during his one season with Brian Daboll and Mike Kafka. In fact, Jones’ cumulative Elo is now higher than either Simms’ or Manning’s after the same number of games. Coaching and offensive scheme matter.

Compare Jones to two other young QBs who experienced a coaching change for the better this year:

Data from NFL Elo App

Both Tua Tagovailoa and Trevor Lawrence benefited from changes this year to coaching on the offensive side of the ball, and the results are clear in Elo. Tagovailoa’s dramatic jump in 2022 (before his concussions limited his performance) also surely benefited from the addition of Tyreek Hill to his receiving corps. Lawrence’s may benefit to a lesser extent from the additions of Christian Kirk, Zay Jones, and Evan Engram. It will be interesting to see how Jones’ Elo progresses if he returns to the Giants in 2023 with a game-breaking wide receiver to throw to and Mike Kafka still calling plays.