Drake Maye’s play through the first three weeks of the NFL season has been nothing short of fascinating to witness. Through three games, the future of the New England Patriots has largely been the most captivating athlete the team’s had in years. Although his occasional predilection to hold the ball too long does personally drive me crazy, any Patriots fan should feel extremely bullish about a future where Maye stays our starting quarterback.
Most advanced metrics have been relatively favorable to Maye too, which highlight his own improvement from his rookie season. In 2024, Drake Maye finished 24th in league-wide combined Expected Points Added per Play and Completion Percentage over Expected (EPA+CPOE) among quarterbacks with 200 plays or more (0.067). This year, however, and among quarterbacks with 40 plays or more, he’s eighth in the league (0.159). Although this season has only given us three games’ worth of data, by 2024 numbers, Maye has produced like a Top 10 quarterback.
Will these numbers hold? To assess how good Maye has been in a variety of quarterback-specific statistics, I’ve jumped into the Pro Football Focus Premium archives to analyze how he stacks by the numbers in key areas. In order, they are passing pressure splits, time-to-throw splits, passing-concept splits, passing-depth splits, and sack attribution.
EDITOR’S NOTE: This column was written at the conclusion of Week 3 games, but it will be published after Thursday Night Football.
Pressure Splits
One way to get beyond raw totals is to look at contextual splits. These tell us how quarterbacks perform under two types of circumstances: blitz-or-no-blitz and pressure-or-clean pockets. Although passer rating is not necessarily the by-all-end-all authority on how well a quarterback performs, I will be using it for this section (as well as the following ones) because it’s easy to understand.
Compared to last year, when Maye struggled heavily against the blitz (75.4), Maye has been the third most efficient quarterback in the league in that same category (120.0). Only Jordan Love and Lamar Jackson currently are ahead of him. Against non-blitzing opponents though, he’s been closer to above average (91.3) than outstanding. This is roughly around where he was last year (93.4).
Performing well in a clean pocket is one of the most stable indicators of quarterback performance. Here, Maye has largely continued his excellent play (113.4). With that said, he tends to struggle when under pressure (58.4), though it’s not much different from the average starter (64.8) in our dataset. Of quarterbacks under pressure, only Lamar Jackson has a passer rating above 100.
Time to Throw: Quick Decisions vs. Extended Plays
In this section, I’ve looked at quarterback performance under two different time-to-throw windows. One of them is how quarterbacks fare when they get the ball out early (under 2.5 seconds); the other one is when they hold it longer (2.5 seconds or more).
If you live in New England, you’ll know that there was a media train around a possible conflict between offensive coordinator Josh McDaniels’ style and Maye’s gunslinger tendencies. For most of this season though, Maye has excelled in the quick game, with the eighth highest passer rating (111.9) among qualifying quarterbacks. If anything, it’s when the play stretches past 2.5 seconds where Maye has relatively struggled (83.7). Per PFF, he has had three turnover worthy plays in such situations versus one in the former category. Moving forward, Maye will need to improve in these spots and showcase more patience.
Passing Depth Splits: How Far Downfield Does Maye Thrive?
For this part, I’ve broken down how Maye has performed in four notable categories: throws behind the line of scrimmage, throws under 10 yards, throws under 20 yards, and throws over 20 yards. After doing that, I then combined the former two and the latter two into two subsets: “quick game” and “long game.”
Maye has been nothing short of outstanding as a long-game passer, and as mentioned before, he’s done his job in the quick game. While he’s only had seven pass attempts down the field (20+ yards), he’s completed four of them for 117 passing yards and a touchdown for the third highest passer rating (141.4). Combined with how well he’s performed on quick plays, this demonstrates that Maye can kill teams down the field and methodically pick apart a defense.
Passing Concepts: Screens and Play-Action
The next step in understanding Maye’s early success is his performance in specific passing concepts. For this part of the column, I’ve broken done how he and other quarterbacks have performed on screens, non-screens, play action, and non-play action calls.
Some quarterbacks, like Tua Tagovailoa and, surprisingly, the resurgent Baker Mayfield have largely seen success through screens this year. With Maye, however, it’s actually been a relatively weaker spot for him in terms of production and efficiency (88.0 passer rating). In my opinion though, it’s a mix of column A and column B;. Though Maye’s largely been an accurate passer, he does tend to whiff throws every now and then, and the Patriots as a whole have surprisingly called only eight screen pass attempts all year.
While Maye definitely sees a slight improvement in play action (108.3) vs. non play-action (99.2), the difference is largely negligible – hardly what you’ll find in the tier of players that includes Dak Prescott, Bo Nix, Jake Browning, or even Matthew Stafford. Side note, however: it is very strange to see each of Justin Fields, Geno Smith, and Brock Purdy so low as play action passers, especially as all three of them are threats on the run as well – it’s especially weird to see Purdy, at least in two games this year, confusingly excel as a passer without play action.
OL% vs QB% Pressure Responsibility
To take a brief detour, the 2024 Patriots were largely defined by a horrific offensive line. And yet after two expensive swings in free agency, as well as drafting left tackle Will Campbell, the 2025 Patriots currently rank fourth in offensive line efficiency (87.8) per PFF, which is a leap from 31st (80.1) the prior year. As it pertains to Maye though, I was curious to see how he stacked compared to his peers in terms of whether he ‘owns’ the pressures induced on him or not this season.
EDITOR’S NOTE: For simplicity, I’ve excluded the rare “Other%” category, where pressures are assigned to running backs, tight ends, or schematic quirks. Those show up in the data occasionally, but they’re marginal and muddy the bigger story.
Unlike our other charts, which are fairly unambiguous,this one is tougher to decipher given the noisy dataset. Michael Penix Jr., who has had no pressures attributed to him, is surely not Tom Brady reincarnated; this clearly reflects a very small sample size of total pressures. And in a way, owning pressures is not inherently a bad thing – take Lamar Jackson, who leads the league in pressures owned (12). For some quarterbacks like Jackson and Patrick Mahomes (third with 9), this is a miniscule downside in exchange for having them potentially make magic out of a bad situation.
Meanwhile, Maye falls closer to the middle (11.8%). This is a slight improvement from last year’s numbers (14.3%), but it’s made a huge difference for New England’s offense, as both he and the situation around him have improved drastically.
The Bigger Picture: Archetypes in Play
As a final step, I wanted to see if we could reduce all the passer rating splits into a single visual that describes quarterback “archetypes.” Some of my long-term readers might remember that I’ve done principal component analysis (PCA) before for analyzing cornerback play in decades of data analysis. For this column’s purposes, I’ve done it for every passer rating split I’ve tracked.
Because different splits are on different scales, I first normalized each quarterback’s numbers relative to the league average. After that, I then used PCA to compress those 14 dimensions down to two axes, PC1 and PC2, in order to capture the largest variance in playstyle. Finally, I applied K-Means clustering to highlight natural groupings. The results were interesting, to say the least. The way I see it, any good beginner model isn’t necessarily 100 percent accurate, but it might give a little bit of insight more than half the time while huffing crack the other percent of the time.
Remember: PCA is purely descriptive, not predictive or adjusted for domain expertise. I do not think, for example, that Patrick Mahomes is a remotely comparable player in terms of value to JJ McMarthy, Cam Ward, Joe Flacco, or Bryce Young. However, it is encouraging to see Maye close to players like Josh Allen and Baker Mayfield from a purely numbers perspective. Fingers crossed that this holds up.
Appendix
My Spreadsheet of all these Stats