Through eight weeks of the NFL’s 2025 season, Drake Maye has been the greatest deep passer in recorded football history. This is no exaggeration – out of qualifying quarterbacks (with 50 percent or more of the max number of passes over 20 air yards), he is the only one to boast a perfect 158.3 passer rating. In fact, since 2006, when Pro Football Focus began tracking deep passing data, there has never been a perfect passer on deep passes (defined by being 20+ yard throws) this early into a season.
To see if this was just a deep passing trend, I used nflR to dive through the history of quarterback play-by-play data, examining one particular advanced metric: Completion Percentage Over Expected (CPOE). For those not familiar with this statistic, it uses proprietary player tracking data to evaluate passing accuracy adjusted for context. By this statistic, 2025 Drake Maye’s first eight weeks are the third most accurate passing stretch in recorded history, just behind 2013 Phillip Rivers and 2007 Tom Brady.
While CPOE is not necessarily one-to-one with being the ultimate quarterback metric, it’s still safe to call Maye’s current stretch historically productive. What I’m curious about, however, is how Maye tracks against contemporaries – do players like Maye typically maintain their production throughout the second half of the year? If not, what changes and what can New England Patriots fans expect from their budding star quarterback? I decided to look into the numbers behind what appears to be Maye’s Year 2 breakout.
Methodology
Longtime readers of my portfolio (all five of you) will know that I love a good k-Nearest Neighbors algorithm. This is basically a way of taking all the columns and rows of a dataset containing every quarterback season since 2006 (or more specifically, the first eight weeks of them) and finding five statistical comparisons and finding which ones have the most statistically similar scores to Maye without adjusting for feature importance. To keep things simple, however, I just looked at EPA/play and CPOE, essentially measuring his overall production and accuracy.
Here, we get a fascinating group of five elite quarterbacks during their primes. 2013 Philip Rivers stands out as the closest match (12.57% CPOE, 0.319 EPA/play), having one of the best statistical starts of his Hall of Fame-caliber career. Joining him is 2008 Kurt Warner (9.61% CPOE, 0.230 EPA/play) and MVP winner 2009 Peyton Manning (9.45% CPOE, 0.367 EPA/play). Interestingly enough, Russell Wilson – another dual threat quarterback with a penchant for deep shots – appears twice here in his 2020 season (10.29% CPOE, 0.234 EPA/play) and his 2019 season (10.16% CPOE, 0.184 EPA/play). The question, however, is how these comparison points ended up performing in the second half of the year.
Unsurprisingly, all of these players ended up regressing a bit more in the second half of season. This didn’t necessarily stop these players from obtaining any type of success – as both Warner and Manning ended up leading each of their respective teams to Super Bowl appearances – but it does suggest that every element of their hot starts perhaps won’t stay constant.
Although the relatively modest R-squareds for the above two models shows that they aren’t entirely explanatory (as strength of schedule, supporting cast availability, and injuries can shape a player’s success), they suggest that eventually a dropoff will happen. In fact, across the broader 515 qualifying quarterbacks in our data set (those with at least 100 attempts in the first half and second halves of each respective season) , 291 of them regress in CPOE (-0.67% on average) and in EPA/play (-0.013).
Given Maye’s incredible start (12.31% CPOE, 0.289 EPA/play) to the season, our kNN model predicts a significant dropoff in the second half of the season if he follows the trends of his statistical comparison points. More specifically, it predicts him to have an above average 3.70% CPOE and a similarly above average EPA/play of 0.116 in the final nine games of the season. Projected across the full season, this would leave Maye with a still-elite, though not earth-shattering CPOE (7.75%), and a still-great EPA/play of 0.197, which would roughly make him a Top 10 quarterback by the end of the year.
Adding More Context
At the same time though, what if these are not the right players to compare Maye to? After all, Maye currently sits at 3rd in betting odds for winning the NFL MVP award this year. Out of curiosity, I decided to examine how our version of “Projection A Maye” stands up against previous MVP winners and runner-ups (in cases when non-QBs won) of each year since CPOE and EPA were tracked.
What’s interesting to note about each of these quarterbacks is how they trended from the first half of their respective seasons to the second half. What I found was that MVP-caliber quarterbacks typically retain their accuracy throughout the full season, with the typical change being only -0.02% in CPOE, while the average shift in EPA/play trended marginally positively (+0.012). And yet even if Maye ended up declining in the second half of the year in line with his historical comparison points, he would still have the most accurate MVP-candidate-worthy season since 2011 Rodgers, as well as an EPA/play that, bare minimum, won’t disqualify him from the race.
Out of curiosity, however, I decided to compare Maye’s current pace in our two different statistics, if it exactly holds up, with three different outcomes: the expected projected regular season stats in Projection A (his initial kNN comparisons), his Projection B (assuming he follows MVP candidate trends), and the MVP average. The results showcase a wide range of outcomes, but my personal takeaway is that anywhere from a regression to “very good play” to “sustainably game-breaking” could remain in the picture.
It’s worth noting Maye also has several factors going in his favor for the rest of the year. For starters, his schedule continues to be fairly light, as the Patriots currently have the third-easiest remaining schedule out of current teams. It’s worth also noting the unusual bit of luck he’s had when it has come to player availability as well – on the offensive end, none of his best receivers (Stefon Diggs or Kayshoun Boutte) or any of his best offensive linemen have missed significant time.
Of course, we can’t talk about Maye’s chances at winning the MVP without addressing the elephant in the room.
Maye vs. Mahomes
No one can ignore the presence of Patrick Mahomes, the best quarterback of the last five years, in the 2025 MVP race. Although the Kansas City Chiefs started the year off on the wrong foot, they’re currently amid a three-game win streak in which Mahomes’ play has taken off to another level. Save for a stray loss to the Jacksonville Jaguars, Kansas City’s return to form has been as big of a storyline over the last five weeks as Maye’s breakout party. While Josh Allen is still technically ahead of Maye in the 2025 betting odds, given his position (second to the Patriots in the same division) and worse overall statistics, I’m going to focus on Maye vs. the closest thing we have to a consensus best quarterback in the league.
By the statistics, it Maye (12.30% CPOE and 0.284 EPA/play) has a pretty commanding lead over Mahomes (2.85% CPE and 0.244 EPA/play) in eight games. Combined with the fact that Maye’s team (6-2) currently has a better record than Mahomes (5-3), and it’d be safe to call this debate a done deal if the season were to end today. With that in mind though, we have nine weeks left in the regular season to see how it shapes up.
In similar fashion to how I used a kNN algorithm to create five statistical comparison points to Maye, I compared Mahomes’ first eight starts of this season to each of his previous five seasons, as well as their second halves to examine what – if all else were held equal – we can expect to see from him in the race.
Here is where we discover something shocking; if we assume that Maye regresses to the mean in line with his original statistical comparison points, and assume that Mahomes’ second half of the year falls in line with his previous five seasons in terms of how he’s trended, then Maye still ends up with a lead in both EPA/play and CPOE.
There are several factors that have undeniably shaped a role in Maye’s historic production. But the data says that even with a regression to the mean, we can expect to see Maye in the MVP race for the rest of the season, and arguably ahead of the betting favorite too. For Patriots fans heading into the 2025 season with any doubt about the long-term potential of their franchise quarterback, it’s safe to say that even with a regression, he’s not going anywhere any time soon – and that maybe, just maybe, the best way to counter Mahomes Magic is a little bit of Maye Magic.
Appendix