I Have Clustered The 2025 NFL Draft QB Prospects Just for YOU!

Its that most wonderful time of year for football!

One of my personal favorite times of the football year is the NFL Draft. A collision of college football with the NFL, we get to see our favorite players graduate into the big leagues. One of the time honored traditions of this is newsletter is to cluster the current crop of QB prospects with prospects of past to see if we can make any connections and get an idea of how this class stacks up to recent history. As we are only two weeks away from the draft, its time to bust out ole reliable: K-Means Clustering!

What the heck is K-Means Clustering?

If you’re newer to the newsletter, first off welcome! K-Means clustering is a way we can take as many variables as we want, feed it into an algorithm which will in turn group QBs together into clusters based on how similar their numbers are to a point in the middle of a cluster. There are endless variables and metrics that can be used for QB performance, which can make it pretty daunting to try and compare players on your own. So instead of doing that, why not let math do the hard work!

A First Look At This QB Class

Before we get to clustering, let’s take a look at the 2025 QB class to see what we are working with. Compared to recent previous years, this class has been called a “down” class with more middle round options compared to top 10 talent.

If we look at each QBs ranking on the consensus mock draft boards, we can see that those sentiments appear to be correct. While the class isn’t 2022 levels of down, it certainly doesn’t have QBs at the top of the board that we have seen in years past. At the top of this class is Cam Ward, who is projected to be the 1st overall pick to the Titans, and Shedeur Sanders, who has seen his draft odds go from “great chance at the top 3” to “Uhhhh… the Saints at 9? Maybe?”. QB3 in this draft is anyones best guess. Jaxson Dart is currently ranked right at the end of the first round, but Jalen Milroe has been generating buzz as of recent, especially with his invitation to the NFL Draft which is typically reserved for players who may hear their name called relatively early in the draft.

Here is the rest of the draft prospects with ranks on the consensus board. The group from Ohio State’s Will Howard to Notre Dame’s Riley Leonard represent the big crop of QBs from this class. Last year, we didn’t hear a QBs name called after the 12th pick (Bo Nix) until all the way in the 5th round at 150 (Spencer Rattler). The year before that, we had 5 QBs taken within the 2nd to 4th rounds. If I had to guess, I believe we see a return to 2023 with multiple QBs taken in the middle rounds.

The road that Cam Ward has taken from an unranked recruit out of high school to 1st on the consensus draft board (and presumably the first overall pick) is a road that nobody in recent history has taken. When looking at QBs who finished number 1 in the consensus rankings, they all went to big time power 5 schools, even with a couple transfers. If you extend this to the top 3 players on the consensus board, he is only joined by Josh Allen, who went unranked to the number 3 player on the consensus board in 2018.

Our final look at the class before clustering involves taking a peak at the averages of some* of the variables used in the clustering process. All told there are 85 total variables (!) that were thrown into the clustering algorithm. This class compared well in terms of PFF charted accuracy (accurate throw rate). Among the 92 QBs from 2017-2025, Shedeur Sanders (68.3% accurate throw rate) & Kyle McCord (65.5%) rank 3rd and 7th respectively.

The past 5 draft classes have had overall great PFF Rush grades, with the 2025 class leading the way. The main drivers of this high average are Jalen Milroe (90.5 PFF Rush Grade) & Riley Leonard (90.3 PFF Rush grade) which rank 3rd and 4th among this entire pool of QBs.

Where this class struggles is taking on sacks which tanks overall efficiency (both their final year EPA/Play & career average game QBR). If you include FCS years, Cam Ward (165 sacks) & Shedeur Sanders (150 sacks) have been sacked more than any other QB in this pool of QBs in the CFB playoff era (2014-Present). Over the past two seasons, these are your most sacked QBs in the FBS:

1. Shedeur Sanders - 90

2. Brayden Schager - 81

3. Jalen Milroe - 67

4. Cam Ward - 66

Now that we have a little bit of an idea of where this QB class stands, its clustering time!

2025 NFL Draft QB Clusters

Here is how the clusters fell for this draft cycle. As mentioned before, there were 85 total variables thrown into this clustering algorithm. While it would be tedious to list them all, the variables all fell into these buckets:

• Career & Final Year efficiency (run and pass)

• Charted Accuracy & Completion Pct Over Expectation

• Tape based metrics (PFF Grades & NFL.com grades)

• Scheme metrics (Play action rate, screen rate etc.)

• Measurables & recruit rating

While some QBs stick out, for the most part this is an entangled web of mess. Luckily, we have another way to visualize this by using tiers. Here is what the clusters look like when they are ranked by the clusters average total PFF Wins Above Average:

This gives us a little bit cleaner look at the clusters. At the top of the list you have QBs that, as a group, averaged the most PFF Wins Above Average. You can see Jaxson Dart and Dillon Gabriel occupy the first tier, as both were high productive college QBs. While they weren’t as efficient on a per snap basis as the next tier down, they still were at the top of college football for a decent chunk of time.

Kyle McCord and Shedeur Sanders occupy the “Tape grades over production” tier. As an entire group, this cluster tended to do better in more tape based metrics (PFF Grades, NFL.com grades) vs. raw production.

To me, the most intriguing cluster here is the “high risk, high reward” cluster.

You have a multi time MVP, a reigning superbowl champion, reigning rookie of the year and two backup QBs (one of which, Malik Willis, performed well in Jordan Love’s absence in Green Bay). You also have the consensus number one player, Cam Ward, and Jalen Milroe who appears to be everything from an exciting QB prospect with limitless potential to a potential RB at the next level. As I said before, this class doesn’t really have a true QB3. Perhaps this can be filled by Milroe, who has immediate top 5 rushing upside and a deep ball that can punish the defense if they stack the box.

The bulk of the rest of the 2025 QB class is in the “The QB Potluck” tier. This tier includes everything from Justin Herbert & Brock Purdy to practice squad QBs. Unlike last year, we do not have any QBs that are included in the Josh Allen tier of craziness. Similar to what we did earlier with the QB draft classes, here is how the clusters stack up in key metrics:

This shows that while tier 1 produced more PFF WAA, they also played a lot more snaps on average and were not as efficient as the 2nd tier. Our highly intriguing Cam Ward/Jalen Milroe tier were not the most accurate QBs in college, and they took a LOT of sacks, but they were still highly explosive and efficient thanks to their legs.

Quick Hitters & Final Thoughts

Part of my day job involves coming up with short, succinct evaluations of players. As a final look at this QB class, I will attempt to summarize their profile in one to two sentences (in a little bit more casual way):

Overall, this class doesn’t have as many heavy hitters as previous classes, but there is a core group of guys that should find themselves dressing out on Sundays this fall. So much of this process is where these QBs end up, so we will have to see where they end up in a couple weeks. Either way, I am still VERY much excited for the NFL Draft!

Lily Donation Fund

In October our lives upside down for a moment as our little cat was sick. After some extensive care and a really rough surgery, she has been recovering well and back to being our tubby little girl.

I hate that I have to do this but this whole ordeal was rough financially for us. If you enjoy the articles and wish to help (which you 10000% do not have to if you do not want to) I left the link to the GoFundMe up above.

If you want to dive in to the data like I do, check out @CFB_Data and @cfbfastR on Twitter, where you can learn how to get started in the world of College Football data analysis!

If you want to see more charts and one off analysis, follow my twitter page, @CFBNumbers