Determining NFL Quarterback Archetypes (with stats!)

We’re obsessed with grouping things together. We self-select into groups based on which political candidate we support, which sports team we root for, and which arbitrary country we’re born in. People also spend hours on the internet arguing over “tiers”, or groupings, of their favorite athletes and sports teams. For example, which NFL Quarterbacks are “elite” vs. “great” vs. “just good”? Is Lamar Jackson legit? In these arguments, we typically use statistics like passer rating or yards per attempt to make our point. Ultimately, though, the distinctions and groupings are kind of arbitrary.

But what if there was a way to use Machine Learning methods to statistically sort athletes into groups based on shared characteristics? Enter clusteringa methodology of grouping similar observations into groups, or “clusters”. The theory behind clustering is kind of complicated, but you can essentially think of it as an algorithm that sorts observations into “clusters”, within which observations share similar characteristics (i.e. passer rating, rushing yards per game, etc.). We can think of a lot of applications for this, but this article focuses on NFL Quarterbacks from 2021 to 2023. 

Click to skip ahead to a section: 



[jump to top]

The argument over NFL Quarterback Archetypes has existed as long as the forward pass itself. Where do we draw the line between a pocket passer and a mobile quarterback? Of course, racial biases often seep into our perceptions of NFL quarterbacks, making answering this question that much harder. Black quarterbacks tend to be perceived as more mobile and less reliant on passing, even when that may not be the case. The reverse is often true for White quarterbacks. 

Clustering allows us to take an agnostic approach to this problem. In this post, I analyze all NFL Quarterbacks that finished in the top 15 of total passing yards at some point since the 2021 season. For teams that had no passers that fit this criteria, I considered the quarterback that started the most games in 2023. Just in case you’re wondering why Aiden O’Connell is on this list. 

I considered the following statistics (averaged across 2021 to 2023): 

  • Completion %
  • Interceptions per game
  • Fumbles per game
  • Passing touchdowns per game
  • Passing yards per game
  • Passer Rating
  • ESPN QB Rating
  • Rushing yards per game
  • Rushing touchdowns per game
  • Rushing yards per carry
  • Yards per passing attempt


I collected a number of other statistics (such as total passing attempts), but ultimately they were highly correlated to these variables, so they did not add much information to the clustering analysis.  The underlying data can be found below (feel free to click the button to download to excel): 

I should note that there are actually a few different clustering algorithms. I used the k-means method for this analysis. This algorithm will consider the above statistics and group players who are similar across these variables together. For example, players with good completion percentages and high passing yards per game (i.e. good passers) will tend to be grouped together, while bad passers will tend to be grouped together.


[jump to top]

The results were remarkably on point — they were pretty much what you might expect. The clustering method did a remarkable job at producing 3 clearly defined archetypes, with very little “noise”. Although I considered many variables in the clustering analysis, they can easily be separated into two categories: (1) passing statistics and (2) rushing statistics. As a result, we can pick a “representative” passing statistic and rushing statistic to visually plot the archetypes. I pick Passer Rating (which incorporates many passing statistics, such as passing yards and touchdowns) and Rushing Yards per Game when doing so. The following chart shows the results: 

The archetypes can be fully summarized below. As you can see, the clustering algorithm did a pretty good job of separating out QBs based on their rushing and passing prowess.


Archetype 1: Pass First 

  • Avg. Passer Rating: 98.4
  • Avg. Passing Yards/game: 246.7
  • Avg. Yards per Carry: 3.7
  • Avg. Rushing Yards/game: 10.7
  • QBs: Aaron Rodgers, Brock Purdy, C.J. Stroud, Dak Prescott, Derek Carr, Geno Smith, Jake Browning, Jared Goff, Jordan Love, Matthew Stafford, Patrick Mahomes, Russell Wilson, Tua Tagovailoa


Archetype 2: Rush First

  • Avg. Passer Rating: 88.9
  • Avg. Passing Yards/game: 216.8
  • Avg. Yards per Carry: 5.6
  • Avg. Rushing Yards/game: 44.4
  • QBs: Jalen Hurts, Josh Allen, Joshua Dobbs, Justin Fields, Kyler Murray, Lamar Jackson


Archetype 3: Poor passers and poor rushers (bad at both)

  • Avg. Passer Rating: 84.3
  • Avg. Passing Yards/game: 193.5
  • Avg. Yards per Carry: 3.5
  • Avg. Rushing Yards/game: 10.3
  • QBs: Aidan O’Connell, Andy Dalton, Baker Mayfield, Davis Mills, Desmond Ridder, Gardner Minshew, Jacoby Brisket, Kenny Pickett, Mac Jones, Ryan Tannehill, Sam Howell, Trevor Lawrence, Tyrod Taylor


The following chart helps summarize the difference between the archetypes:


[jump to top]

This is by no means an authoritative list of QB archetypes. That being said, it’s always fun and interesting to add a little statistical backing to a conversation that is usually dominated by subjectivity and bias. Do you find any of these pairings egregious? Feel free to drop a comment or reach out to let me know!

Longtime readers of my blog may find this article vaguely familiar. That’s because it’s an update of an older article that I published in 2020 (back in the days). I had originally largely forgotten about this article, but I noticed that it was getting a few hits every day despite being so old and outdated. It turns out ya boy made it, and this is probably now my most visible post on Google. I realize most people googling this term are probably looking for information about Madden QB Archetypes, but still kinda cool. 

Anyway, thanks for reading, and don’t forget to subscribe below if you liked this article!



* indicates required

Intuit Mailchimp

Leave a Reply

Your email address will not be published. Required fields are marked *