What are the Best Quarterback Archetypes in the NFL?

The quarterback position is the most valuable position in the sport of football. The impact quarterbacks have on their team is massive, and teams can live or die based on who is taking snaps for them. Today, I will dissect the different quarterback archetypes found in today’s league and determine which one is the best.

To determine how many archetypes there are, I performed a k-means cluster analysis in a data analysis program called Orange which analyzed quarterback statistics and automatically separated every quarterback into a cluster with other quarterbacks with similar statistical profiles.

K-means clustering is a type of data analysis that separates data points and puts them into groups with other data points that share similarities. This is one of the simplest machine learning algorithms, but this project wouldn’t be possible without it.

This algorithm determined that the thirty-five QBR eligible quarterbacks from last season can be separated into eight different clusters or categories. By looking at the players’ stats in each cluster, I decided what archetype each group represents.

For this analysis, I used many stats, including per game passing and rushing stats and rate stats like completion percentage, touchdown percentage, interception percentage, and passing and rushing yards per attempt.

Below are the eight clusters with the players sorted by their adjusted net yards per attempt last season. Each quarterbacks’ passer rating and QBR are also listed, which gives you a pretty good picture of who performed the best during the 2020 season.

After this first part, I took the average passer rating, QBR, and ANY/A for each cluster to determine the best quarterback archetype. Without further ado, let’s get into the different clusters.

---

Cluster One: High Volume Pocket Passers

Strengths: Attempts per game, passing yards per game, passing touchdowns per game

Weaknesses: Rushing yards per game, fumbles per game

These are quarterbacks who throw a lot of passes each game and are accurate while doing so. Unfortunately, they don’t provide much value as rushers, limiting their ability to create plays outside the pocket. However, they’re pretty efficient and can lead high-level offenses due to their above-average passing ability.

Average Passer Rating: 96.5

Average QBR: 66.8

Average ANY/A: 6.85

Cluster Two: Low Volume Pocket Passers

Strengths: Interceptions per game, rushing yards per attempt

Weaknesses: Pass attempts per game, passing yards per game, touchdown percentage, rushing yards per game

These quarterbacks don’t attempt many passes per game, which hurts their counting stats, but they are risk-averse and have decent athleticism. They’re not great passers now, but they’re solid, and their athleticism gives them some upside and allows them to survive on the field.

Average Passer Rating: 91.5

Average QBR: 58.7

Average ANY/A: 5.92

Cluster Three: Inefficient, Low Volume Dual Threat Passers

Strengths: carries per game, rushing yards per game, rushing yards per attempt

Weaknesses: pass attempts per game, passing yards per game, touchdown percentage

All of these quarterbacks struggled in 2020, putting up poor passing stats across the board. However, they have strong physical tools, which give them a relatively high ceiling in the right situation. Their risky playstyle and inefficiency give them a very low floor, but they could potentially shine with talent around them.

Average Passer Rating: 75.3

Average QBR: 50

Average ANY/A: 4.59

Cluster Four: Elite, Risk Prone Play Creators

Strengths: pass attempts per game, passing yards per game, completion percentage, carries per game, rushing yards per game

Weaknesses: sack percentage, fumbles per game

These are some of the best quarterbacks in the entire league. What makes them so special is the fact that they can pass and run at a high level. Their ability to pass from inside and outside the pocket makes them incredible and allows them to make throws that most quarterbacks can’t even dream of completing. Their most significant issue is that they’re a bit risk-prone. Due to their aggressiveness, they tend to take more sacks and throw more interceptions. Nevertheless, they’re still elite quarterbacks.

Average Passer Rating: 108.2

Average QBR: 75.2

Average ANY/A: 7.62

Cluster Five: Elite Rushers With Decent Arms

Strengths: Carries per game, rushing yards per game, rushing touchdowns per game, rushing yards per attempt

Weaknesses: Pass attempts per game, passing yards per game

Here are the best rushing quarterbacks in the NFL. They may not be the best passers, but they are elite runners whose dynamic playstyle makes them extremely valuable in the right system.

Average Passer Rating: 92.2

Average QBR: 63.2

Average ANY/A: 6.2

Cluster Six: Inefficient, Low Volume Pocket Passers

Strengths: interceptions per game, fumbles per game

Weaknesses: passing attempts per game, passing yards per game, touchdown percentage, interception percentage, carries per game, rushing yards per game

I’m not going to sugarcoat things. These are some of the worst quarterbacks in the NFL. Their underwhelming efficiency, combined with their poor physical tools, makes them poor passers. If one of these guys is your starter, you’re likely going to be very bad offensively.

Average Passer Rating: 79.1

Average QBR: 38.2

Average ANY/A: 4.76

Cluster Seven: Ultra-Efficient Pocket Passers

Strengths: pass attempts per game, passing touchdowns per game, completion percentage, touchdown percentage, interception percentage

Weaknesses: carries per game, rushing yards per game, rushing yards per attempt

These are the best pure pocket passers in the NFL. They are efficient, high-volume passers, but they’re also some of the worst athletes in the league. They may be extremely accurate within the pocket, but they won’t create plays outside of it. This limits their upside and keeps them from being elite quarterbacks. Nevertheless, you can win with this type of quarterback.

Average Passer Rating: 99.9

Average QBR: 67.4

Average ANY/A: 7.15

Cluster Eight: Ultra-Efficient Dual Threat Passers

Strengths: literally every stat

Weaknesses: literally no stat

This is the cream of the crop. These quarterbacks are very efficient high-volume passers who pair incredible precision with above-average mobility. They really don’t have a weakness, making them extremely valuable. They aren’t as dynamic as the elite play creators, but they’re more efficient and less risk-prone, making them better passers overall.

Average Passer Rating: 112.1

Average QBR: 81.9

Average ANY/A: 8.37

---

Now let’s see which archetype performed the best last season. Here is each cluster ranked by the quarterbacks’ average passer rating, QBR, and ANY/A.

Average Passer Rating

1. Cluster Eight (112.1)

2. Cluster Four (108.2)

3. Cluster Seven (99.9)

4. Cluster One (96.5)

5. Cluster Five (92.2)

6. Cluster Two (91.5)

7. Cluster Six (79.1)

8. Cluster Three (75.3)

Average QBR

1. Cluster Eight (81.9)

2. Cluster Four (75.2)

3. Cluster Seven (67.4)

4. Cluster One (66.8)

5. Cluster Five (63.2)

6. Cluster Two (58.7)

7. Cluster Three (50)

8. Cluster Six (38.2)

Average ANY/A

1. Cluster Eight (8.37)

2. Cluster Four (7.62)

3. Cluster Seven (7.15)

4. Cluster One (6.85)

5. Cluster Five (6.2)

6. Cluster Two (5.92)

7. Cluster Six (4.76)

8. Cluster Three (4.59)

Cluster eight (super-efficient pocket passers with above average mobility) ranks first in all three categories by a fairly wide margin. Quarterbacks like Patrick Mahomes and Aaron Rodgers, who are accurate passers with elite arm strength and above-average athleticism, are extremely valuable in today’s game. Their ability to attack defenses deep down the field and make plays outside the pocket while limiting their turnovers is incredible.

I bet you didn’t expect to see Ryan Tannehill in the same group as Mahomes and Rodgers. Tannehill is one of the most underrated quarterbacks in the NFL, and this proves how good he actually is. Is he really on the same level as Mahomes and Rodgers? I’d say no, but he is a phenomenal passer and an underrated athlete who has been very productive for the last couple of years. He’s a very good player and is probably a top ten quarterback in the NFL.

Cluster four (efficient play extenders) is pretty easily the second-best quarterback archetype in the NFL. These players are phenomenal passers and extremely good athletes who are high-level play extenders. They don’t always make the best decisions, so they take more sacks and throw more interceptions, but you’ll deal with that because they make so many high-level plays. Their versatility is incredibly valuable, and every team is looking for this type of quarterback.

Cluster six (risk-averse passers with limited physical tools) is, in my opinion, the worst quarterback archetype in the NFL. These quarterbacks are inefficient, low-volume passers who don’t have good physical tools. This severely limits their upside and makes them extremely invaluable players.

Nick Mullens and Nick Foles are a part of this group and were better passers than guys like Sam Darnold and Carson Wentz in 2020, but I don’t think they are better individual players. Darnold and Wentz were not good last season, but they have good physical tools and are able to make plays work even when they fall apart. Although they didn’t play well in poor situations last year, I believe they could potentially find success on better teams with more talent around them.

This was a really fun project and I think it produced some very interesting results. For me, it confirms that athletic quarterbacks who are good passers and rushers are the most valuable archetype, and it’s not close. It also showed me that Carson Wentz and Sam Darnold, despite their struggles last season, are probably better than I previously thought.

I’ve been lower on these two players, but this project showed me that their mobility and arm strength give them a high enough ceiling to be good in the right situation. I feel more comfortable saying that about Wentz since he was an MVP candidate in 2017, but perhaps Darnold will breakout in Carolina this year. I still have my doubts about these guys, but I think I’ve been a bit too harsh.

I really enjoyed this, and I hope to do this again at some point using multiple years of data. That’s the biggest weakness of this project. It only includes stats from the 2020 season, so outliers may have skewed the data. Nevertheless, I think the k-means cluster algorithm did an outstanding job creating different quarterback archetypes.