TPRR check-in: Who has made strides earning volume so far?
Team-by-team breakdown of every target-earning profile
Long-time readers know what this is. It’s time for the in-season TPRR check-in, which is just such a fun exercise.
I want to start with a little discussion of routes data, because it’s something I haven’t talked about for a while, and it’s been a hot button topic on social. In the past, I’ve voiced frustration that we don’t have consistent routes data, because it’s so integral to understanding receiving profiles. Not only are routes the obvious measure of opportunity for pass-catchers — that used to be thought of as targets, but we now know targets are earned through skill — but they are also the denominator in several key stats, which creates huge scaling issues when we talk about different sources having different figures.
These scaling issues I am referring to specifically relate to anything that ends in “per route run,” like TPRR, wTPRR, and YPRR, which are in my opinion the best WR stats that exist and the ones this column is based on. For a long time, PFF was the central source for routes data, but it’s become clear that several other sources think they are far too liberal with their charting of what constitutes a route, something the guys at Fantasy Points Data have been discussing heavily recently. I wanted to shout these guys out specifically because I find it commendable they are trying to improve data by charting it in more accurate ways.
At the same time, it is not necessarily the case for me that there is a “right” way to chart routes, and I’ve seen commentary that suggests this. I do think in a vacuum, the way Fantasy Points is striving to do it is probably best. But as long as there is consistency, we’re able to layer in other important information.
Where they are drawing a line is on routes where it is possible to draw a target, meaning the disputed routes tend to be plays like WR screens or I think some RPOs. But the case I would make would include even RPOs, which is to say that we don’t know the specific play design, and we don’t know who actually really was in the progression. One of the things I’ve come to understand as it relates to some first-read target charting, which I’ve written before is necessarily inexact, is that there are a lot of plays where we just don’t have the information. We don’t know what play was called initially, how it was switched at the line, how the progressions were drawn up, how those might have been switched at the line — there’s a great video from early this offseason where Yahoo’s Nate Tice sat down with C.J. Stroud and used the opportunity to ask questions about a few plays, including one where he threw a backside out that Tice was sure was not part of his initial progression on the front-side of the play, and Stroud essentially confirmed that and confirmed that he more or less just went off script because of reads he was making pre-snap about the front side defenders — etc.
So if you have seen commentary that certain routes sources are more right because they have delineated which routes are real and which aren’t for “earning” a target, I would simply make a case that it’s a lot more opaque than that. When you watch former quarterbacks break down plays, they frequently lay out what they think the progression is, which includes possible “alert” routes and then the checkdowns at the end, and still, relatively often, there will be a route that isn’t even given a number in the progression. It’s a clear-out route where, to that analyst, there’s no world they can see where the ball would go there. This was the route Tice asked Stroud about — he is a former collegiate quarterback who couldn’t understand how that was the route Stroud got to, and Stroud more or less confirmed that it was a very rare play (and a good question).
Further, what do you do when a player is the second read in a progression but the first read wins? See, this is an assumption we’ve always made that, yes, this is a route where it’s plausible to earn a target. Except in practical terms, it’s not. This gets back to the increased reliance on first-read targets, and how something like 80%+ of all targets are deemed “first read,” which I also wrote about having real issues with last year (there’s a section in this mailbag from after Week 2).
I’m not trying to have that both ways; I’m trying to emphasize that the more you think about it, the less clarity there is here, and while there are reasonable ways we can try to improve, all of this gets back to the difficulty with quantifying football data across the board, and why I would reject absolutist commentary that states there is a “right” or “wrong” way of doing basically any of it. If you went through that link to first-read commentary in the last paragraph, you’d see it was a section about all “advanced” data, and how “You can fall into an issue with much of the advanced data that’s cited of believing it’s moving the needle more than it actually is, because it’s new and interesting and feels like it’s giving us more than it is.” If you ever wonder what I believe about how amazing some of this stuff is, or whether something is the right or wrong way to do something, as it relates to football stats, you can be confident my answer will be that it’s not as simple as the people incentivized to make it out to be that way are telling you.
Again, this is not to say that there is not virtue in trying to improve things, and I commend anyone working toward that end. But this is not just a semantical point, either. With any data, a major point of emphasis that needs to be remembered at all times is how we use it. That’s literally why it exists — we’re trying to quantify things so we can observe them in numbers.
And that’s a huge sticking point with routes, actually, because of how routes are the denominator in a lot of rate stats. As far as routes data is concerned, the way PFF has always done it, while probably overly inclusive, has been the standard for the per-route metrics, which creates a standard for how we interpret those pieces of data. For example, I know inherently what a 20% TPRR means, or how a 2.0 YPRR season is a great benchmark. When we start pulling out routes, suddenly 2.5 becomes the new 2.0, or something. It’s not exact, and while that doesn’t mean we can’t adjust, it does impact our ability to use the data for its intended purpose.
So I’ll be using PFF data today, as I have for years, because I have found success with it, and I don’t have the incentive right now to relearn the scales of the stats that are impacted, especially because I do believe it’s largely just edge cases that are impacted here, and in many of those cases I’ve shown an ability to be delicate with what the data is saying, because I’m always using a stat like TPRR as just a foundation and then adding a ton of what I believe to be relevant context.
But I also wanted to note something cool that will impact this going forward. As I noted, I’ve voiced frustrations in the past about a lack of consistency with routes data. Everything I just explained is a major problem that creates complications across those types of per-route stats when there’s not a consistent “routes” metric, which is truly crazy if you step back and think about it. I can’t think of other stats we dispute like this, especially when you consider how routes have become this super important stat that other stats are based on. The idea that you’d have different stats, which would have the same name, solely based on your source for routes data, because we’ve never had a uniform source, has been a preposterous issue going on a decade now.
For the last few years, I’ve been calling on the NFL to release their own routes data as an answer to this obvious issue. And I hadn’t really realized it yet, but they have. The new NFL Pro subscription that I’ve been mentioning does include full routes data, and while it doesn’t seem downloadable (yet) for ease of spreadsheet use, coders can access the API and pull it and do whatever it is those modern magicians do.
They also appear to be pretty close to the PFF numbers — definitely not the same, but similarly inclusive. Again, the inclusivity is probably not ideal, but it also means the TPRR/YPRR scales should be similar, and the flip side is because it’s done based on the GPS tracking rather than charting, we get a uniformity that could be described as a positive.
As I’ve tried to be clear to say, the charting sources should almost certainly be considered more “accurate,” particularly those that are considering tough questions and striving toward that exact goal. But that doesn’t mean they are right and other sources are wrong, necessarily. And going forward, I’ll actually be starting to shift everything over to the NFL data, because we now have the league giving us their official definition of what constitutes a “route.”
Before I jump into the teams, let’s also talk more about what it is we’re going to do today. Er, actually, let’s just quote and link to a bunch of shit. Here’s what I wrote in last year’s in-season column.
For those of you newer around here, I started doing work on TPRR back in 2020, when I argued the fantasy industry should be spending a lot more time with it as a stat. It’s been great to see that in the years since, but there’s also been the (expected) attempts to delegitimize it, and so I’ve written a lot of theory about the stat, and how I prefer to use it. That first link is the original post where I introduced it here at the newsletter (I’d previously that summer introduced it through a Twitter thread, while still at CBS), and it still contains some of my best thoughts on it, but then there’s:
More early thoughts on how air yards fit in from October 2020
General thoughts and how WOPR relates to it from September 2021
Even more general thoughts as I started looking at college data in April 2022
The “All roads lead to A.J. Brown” post in July 2022 (that starts “I’ve gotten some questions this offseason about my reliance on TPRR as a metric…”)
If you’re looking for more information about the stat, those are some great places to read up on how I use it, and why I love it (huge hat tip to reader Scott Wolf for helping me catalog my archives this offseason). One short note I’m sure I emphasized multiple places there is that I almost couldn’t care less about the r-squared — even though it’s strong — especially relative to other (blended) stats with lots of stuff thrown into them, because TPRR is a very simple stat, that literally measures what it says, which is the number of targets a player earns per route he runs.
For so many of the stats people want you to believe matter, there is more complexity. Maybe it’s a model with layers of inputs, so you’re never really sure when parts of it are unsustainable small samples.
…
Which isn’t to say TPRR is perfect, and then I’ve also created one of those complicated stats — wTPRR, which adds air yards and weights targets and air yards such that targets are more notable — but I tend to use this stat a little less frequently, because it isn’t so simple to understand. But that short note I wanted to emphasize about TPRR and its r-squared is that what we’re not able to glean in large-sample testing is referred to as “unexplained variance,” and with TPRR I always know what the unexplained variance is. It’s everything else.
So many of the other stats try to throw everything into a model, but that model still isn’t predictive enough to be the only thing you can use (not even close, because it’s football), and it becomes so opaque a stat that you also can’t use it with anything else to do multi-variable analysis, which is really what we need to be doing in all of these spots. We need to be looking at things from different angles to read them right. That’s why we all love analyzing football, because there are no clear answers and each little debate is like a new puzzle where the key piece might be something different.
So with TPRR, I have this great foundational piece, and then I know that I need to add in context about the size of the player’s routes role, and his aDOT, and whether the offense runs a lot of two-WR or three-WR sets (and the logical extension of what his competition for targets is across the board, in terms of the other four possible eligible receivers on any given route), and whether his QB is mobile or takes sacks at a high rate such that a high number of dropbacks (and thus routes) aren’t even becoming pass attempts (and thus targets).
That’s a whole bunch of stuff that sounds vaguely complicated but really isn’t, at least not if you’re reading Stealing Signals each week (or writing it). We know what these offenses are, and what the players’ roles are, and we’re not actually comparing the TPRRs across teams like it’s apples to apples.
In the years I’ve used TPRR, I’ve done extensive early offseason looks at the full-season data, that have been integral to my offseason process, and then perhaps more importantly I’ve done a couple in-season checks each year. And since the beginning, I’ve done it by going team by team and looked at the ecosystem of that individual offense, where I can apply my own context to the unexplained variance of what TPRR is not catching.
For the in-season stuff, I like to wait a few weeks so we can get a decent routes sample. Say, four weeks, so we can get a good number of players over 100 routes. Oh hey, that means we’re up to do that today. Let’s get it.
I always like to start with a look at the league leaders, so you can get an idea of the scale of the different stats. Let’s start with the top TPRRs so far this year with at least 100 routes.
Here’s where I’ll cut in and instead of just copy/pasting the 2023 leaderboard, let’s show 2024.
The displayed data is - TPRR, wTPRR (routes)
Malik Nabers - 0.34, 0.86 (152)
Wan'Dale Robinson - 0.30, 0.62 (123)
Diontae Johnson - 0.30, 0.76 (127)
Amon-Ra St. Brown - 0.29, 0.67 (130)
Michael Pittman Jr. - 0.28, 0.72 (101)
Nico Collins - 0.27, 0.71 (157)
Justin Jefferson - 0.27, 0.74 (113)
Jauan Jennings - 0.26, 0.73 (104)
Tyreek Hill - 0.26, 0.70 (116)
DeVonta Smith - 0.25, 0.67 (110)
Drake London - 0.25, 0.63 (124)
George Pickens - 0.25, 0.68 (112)
Courtland Sutton - 0.25, 0.70 (138)
Chris Godwin - 0.24, 0.52 (133)
Christian Kirk - 0.23, 0.65 (124)
Terry McLaurin - 0.23, 0.63 (124)
Garrett Wilson - 0.23, 0.54 (151)
Brian Thomas Jr. - 0.22, 0.61 (117)
CeeDee Lamb - 0.22, 0.56 (140)
D.K. Metcalf - 0.22, 0.59 (159)
Keep in mind, the point of these stats is not to compare across teams necessarily, so when we see two players from the same team at the top, there’s probably a correlated reason for that. (It also has to be said it’s insane a rookie is at the top.)
One thing I always reference here is you can see how someone like Wan’Dale Robinson has a lower wTPRR because of a lower aDOT and fewer total air yards. We can also sort this by wTPRR, to show how it shifts.
Malik Nabers - 0.34, 0.86 (152)
Diontae Johnson - 0.30, 0.76 (127)
Justin Jefferson - 0.27, 0.74 (113)
Jauan Jennings - 0.26, 0.73 (104)
Michael Pittman Jr. - 0.28, 0.72 (101)
Nico Collins - 0.27, 0.71 (157)
Tyreek Hill - 0.26, 0.70 (116)
Courtland Sutton - 0.25, 0.70 (138)
George Pickens - 0.25, 0.68 (112)
Amon-Ra St. Brown - 0.29, 0.67 (130)
DeVonta Smith - 0.25, 0.67 (110)
Marvin Harrison Jr. - 0.22, 0.65 (123)
Christian Kirk - 0.23, 0.65 (124)
Drake London - 0.25, 0.63 (124)
Terry McLaurin - 0.23, 0.63 (124)
Wan'Dale Robinson - 0.30, 0.62 (123)
Brian Thomas Jr. - 0.22, 0.61 (117)
D.K. Metcalf - 0.22, 0.59 (159)
Alec Pierce - 0.15, 0.57 (100)
CeeDee Lamb - 0.22, 0.56 (140)
Robinson falls to 16th, behind someone like Marvin Harrison, Jr., who narrowly missed the top 20 in the first list but settles in at 12 when we consider his downfield profile.
For the below, I’m going to lower the routes minimum to 20 (5 per game) to bring in RBs and secondary players, because a big thing we want to look at is guys who haven’t had a full role yet, and how they are doing.
Game on.
Arizona Cardinals
Trey McBride - 0.24, 0.56 (84)
Marvin Harrison Jr. - 0.22, 0.65 (123)
Greg Dortch - 0.21, 0.49 (94)
Michael Wilson - 0.18, 0.44 (114)
Elijah Higgins - 0.13, 0.27 (55)
Emari Demercado - 0.11, 0.19 (27)
James Conner - 0.10, 0.15 (61)
Trey McBride is earning volume well for a TE, as we’ll see with several more names below, but his after-the-target efficiency hasn’t been there yet, and he obviously has now missed a game. Marvin Harrison, Jr. has the much stronger wTPRR thanks to a 15.6 aDOT, and his first NFL month has been an unquestionable success.
The rest of the offense seems fairly insignificant, if I’m being honest. Greg Dortch and Michael Wilson are in a range where they matter but, again, aren’t significant. Wilson’s run solid routes, which is nice. The RBs are barely seeing volume.
Atlanta Falcons
Drake London - 0.25, 0.63 (124)
Bijan Robinson - 0.19, 0.28 (84)
Darnell Mooney - 0.19, 0.50 (123)
Ray-Ray McCloud III - 0.18, 0.42 (112)
Tyler Allgeier - 0.14, 0.23 (21)
Kyle Pitts - 0.14, 0.35 (100)
Oh, hey, Kyle Pitts. For reference, he was 21.1% as a rookie, 26.5% in Year 2, 18.5% last year, and now we’re down to 14.0% at a career-low aDOT of 10.2 for an even bigger wTPRR dropoff. The routes aren’t really the issue. He just probably isn’t very good anymore.
Drake London is earning volume, but hasn’t been productive enough thanks to a 6.5 YPT. Darnell Mooney and Ray-Ray McCloud aren’t really that high, but the routes consolidation does help.
Baltimore Ravens
Justice Hill - 0.24, 0.40 (66)
Isaiah Likely - 0.24, 0.58 (76)
Zay Flowers - 0.20, 0.46 (118)
Derrick Henry - 0.17, 0.25 (36)
Nelson Agholor - 0.15, 0.39 (54)
Rashod Bateman - 0.13, 0.38 (111)
Mark Andrews - 0.11, 0.27 (81)
Oh, hey, Mark Andrews. For reference, Andrews was at 17.5% as a rookie in 2018, then in Year 2 had a ridiculous 32.9% figure, then was between 23.9% and 25.3% the next three years (2020-2022), and then when it dipped last year, that was to 21.6%. For him to be sitting at an 11.1% TPRR is almost unfathomable to me, but as I alluded to, we’re going to get more examples of this today at the TE position.
Meanwhile, Isaiah Likely had fallen to 13.0% last year in the first year under Todd Monken, after 20.9% as a rookie, but he’s back up to 23.7%, with his rise and Andrews’ fall feeling correlated.
Zay Flowers was at 19.9% as a rookie, and has had a very slight bump to 20.3%, but his YPT has dropped from 8.3 to 6.6 for a YPRR drop from 1.64 to 1.34, when the thesis would have been a YPRR climb. He’s unquestionably been disappointing so far, but the efficiency side is easier to see rebound, and the TPRR is at least in an OK range. Rashod Bateman’s is far more concerning.
Justice Hill likely won’t maintain a 24.2%, but that — plus the run lean — has been a major factor in limiting the other guys so far.
Buffalo Bills
Curtis Samuel - 0.22, 0.43 (41)
Dalton Kincaid - 0.22, 0.48 (78)
Khalil Shakir - 0.21, 0.45 (89)
Ty Johnson - 0.19, 0.49 (31)
James Cook - 0.16, 0.32 (61)
Marquez Valdes-Scantling - 0.14, 0.47 (43)
Keon Coleman - 0.13, 0.39 (83)
Mack Hollins - 0.11, 0.31 (93)
Dawson Knox - 0.08, 0.17 (50)
Right away, you can see the lack of concentration by the length of the list, which hasn’t really been injury-impacted, but is just rotational.
Khalil Shakir’s issue was always target-earning, but he’s up over 20% thanks in large part to a lot of designed swing passes. That’s fine, and it gives him value, but I wouldn’t look at this as a big skill increase personally.
Dalton Kincaid is in an OK range for a TE, but he’s run 15 fewer routes than Mack Hollins, which is how I’d frame his bigger problem. Through four games, 78 routes is just not enough for a guy to hit a receiving ceiling.
Curtis Samuel’s per-route stuff is also boosted by some short passing, and you can see how the team leader is wTPRR is actually a running back, Ty Johnson, which is just absurd. One thing I’d say is this all seems fairly unsettled, although it could just stay this fluid all year.
Carolina Panthers
Diontae Johnson - 0.30, 0.76 (127)
Miles Sanders - 0.21, 0.33 (48)
Xavier Legette - 0.19, 0.50 (95)
Chuba Hubbard - 0.19, 0.29 (69)
Jonathan Mingo - 0.16, 0.34 (105)
Adam Thielen - 0.15, 0.39 (74)
David Moore - 0.13, 0.33 (23)
Ja'Tavion Sanders - 0.10, 0.17 (61)
Tommy Tremble - 0.09, 0.20 (76)
Diontae Johnson was top three in both TPRR and wTPRR in the league-wide lists above, because he earns volume like that, and the target competition isn’t great. Since Andy Dalton took over, things have obviously been looking up. These are career-high numbers for Diontae, but I kinda think that has a shot to stick given the situation.
Xavier Legette is earning volume well for a rookie, and is clearly the upside play over Jonathan Mingo rest of season for a No. 2 play. His 7.9 YPT has been OK, while Mingo’s been at 4.9 after 5.0 last year and is just hilariously bad. His 0.79 YPRR is showing real improvement over his 0.78 as a rookie, /sarcasm.
Adam Thielen saw his targets dip quite a bit even before his injury, on a small sample of 74 routes. He was solid at 21.4% last year but had fallen to 14.9% and now obviously needs to rehab. He’s off the radar for now.
Both RBs have seen nice volume here, which is great for them and could also eventually be good news for Jonathon Brooks. This is another spot where the TEs haven’t, though, which is a bummer for fellow rookie Ja’Tavion Sanders.
Chicago Bears
Keenan Allen - 0.23, 0.56 (48)
DJ Moore - 0.20, 0.48 (157)
Cole Kmet - 0.20, 0.46 (101)
D'Andre Swift - 0.18, 0.28 (91)
DeAndre Carter - 0.16, 0.43 (75)
Roschon Johnson - 0.14, 0.27 (28)
Rome Odunze - 0.14, 0.44 (151)
Gerald Everett - 0.08, 0.15 (63)
Keenan Allen only has 48 routes so far, so I’d hesitate to draw strong conclusions, but the TPRR has been pretty solid so far. D.J. Moore has stayed in a strong range as well, and Cole Kmet continues to show that he needs to be out there a ton.
My guy Rome Odunze is a bit of a concern here. He’s run a lot more routes than I would have anticipated out of the gate, but is struggling to earn volume on those routes, looking like a distant fourth among their main receiving weapons. His yardage efficiency is also just meh, at 7.9. This TPRR figure isn’t necessarily a death knell or anything, but you’d definitely like your star rookies to be hitting early where possible. At the least, I need to apologize to Siegele after giving him crap when he said Brian Thomas needs to be ahead of Odunze on Sunday night’s Stealing Bananas. While Thomas has unquestionably been great, I honestly didn’t realize Odunze’s limited early production was this routes/volume-dependent.
Cincinnati Bengals
Erick All - 0.32, 0.58 (38)
Chase Brown - 0.26, 0.44 (34)
Mike Gesicki - 0.25, 0.57 (77)
Tee Higgins - 0.22, 0.59 (67)
Zack Moss - 0.17, 0.28 (87)
Ja'Marr Chase - 0.16, 0.39 (143)
Trenton Irwin - 0.16, 0.39 (44)
Andrei Iosivas - 0.13, 0.33 (128)
Drew Sample - 0.11, 0.21 (36)
Ja’Marr Chase sitting at 16% on 143 routes through four weeks is the evidence you need that what I just said above about Odunze isn’t a death knell. But part of the thing for Chase is a 13.0 YPT is still pushing him to a 2.10 YPRR. His own efficiency is limiting passing volume in some spots, and that’s not a major issue. Expect him to be over 20% at minimum on our next check-in, which means well over that mark over the next 150 or so routes. It’s still a little concerning, though, when we thing about his ultimate ceiling.
Tee Higgins has barely played, but sits at a comfortable 22%. Mike Gesicki’s routes have been limited, but he’s earned targets when on the field, at 25%. And with RBs, the high rates are usually a sign of intent — Chase Brown sitting at 26% while Zack Moss is down at 17% does seem to signal to me an intent to get Brown the ball in the passing game.
Oh, one last name I should mention is Andrei Iosivas, who I do like, but who is too low to be anything more than a TD flyer anytime soon.
Cleveland Browns
David Njoku - 0.25, 0.50 (20)
Jerome Ford - 0.20, 0.33 (99)
Pierre Strong Jr. - 0.20, 0.36 (25)
Amari Cooper - 0.20, 0.51 (173)
Jerry Jeudy - 0.18, 0.47 (162)
Blake Whiteheart - 0.13, 0.27 (23)
Elijah Moore - 0.12, 0.25 (153)
Jordan Akins - 0.10, 0.19 (101)
Cedric Tillman - 0.08, 0.16 (49)
David Njoku started hot on his first 20 routes, looking like he might keep last year’s gains, before the injury. He’s barely played, but they could use him. Amari Cooper and Jerry Jeudy have been the only reliable pass-catchers, as the other TEs and Elijah Moore aren’t earning any volume at all.
Njoku’s a nice add and might be a TE who can do some stuff, while Cooper and Jeudy are otherwise the significant pieces of the passing game.