Stealing Signals, Week 14, Part 1
Offenses' 2023 counterpunch, plus TNF and the early Sunday games
Shawn and I recorded a fun Sunday night Stealing Bananas last night, and one thing I got into was the impact of Pass Rate Over Expectation. The league has shifted a lot over the past couple seasons, but a couple of years ago I used to spend a lot more time talking about throwing with intent on early downs and in other nonobvious pass situations (like short-yardage).
That took a bit of a back seat in a 2022 season that saw rushing efficiency tick back up as defenses shifted, and we did see some of the overperforming offenses be run-leaning ones (particularly those with mobile QBs). But one of the interesting trends of 2023 I’ve noticed while tracking this stuff on a weekly level is how that’s shifted back; teams that are throwing with intent — passing to set up the run — have really shown the effectiveness of that this year, again.
And then we have the recent examples of the inverse, like our discussion of the Chiefs against the Packers last week, where their rushing efficiency likely led to suboptimal play-calling and their own demise. We saw similar again this week with the Lions, who got both of their main RBs over 6.0 yards per carry on double-digit rushes, resulting in a legitimately impressive combined line of 21 carries for 132 yards and a score, but in a double-digit loss.
An easy comparison right now is what the Packers are doing, as there has been a lot of recent talk about Jordan Love’s development thanks to back-to-back wins over the Lions and Chiefs, helping them climb back from a 2-5 start to a 6-6 record. In those two huge wins, the Packers posted PROEs of +12.2% and +9.5%; over their first 10 games, their previous season high was +6.6%, and those two recent games have only now brought their season total barely into the positive at +0.7%.
Part of what has stood out about the Packers over those two games — and we’ll see if it continues on Monday night — is that several teams set season lows in PROE in Week 13. In fact, nine teams had PROEs of -10% or lower, and another three were below -8%. Despite 2022 being a season where play-calling clearly trended toward the run — in a way that made some question whether the “Expectation” part of PROE’s formula might need to be adjusted — I don’t remember such an extreme week of teams all shifting toward the run at once.
Typically speaking, about half the league sits around neutral PROEs, probably between plus-or-minus five percentage points. We got back to that in Week 14, with only four teams in negative double figures, and 14 teams within five percentage points of zero (only 26 teams have played), highlighting how uniquely run-heavy Week 13 was. And just to give a quick sample of the Week 14 data, the Jets led the NFL at +9.0% in a game where Zach Wilson looked better than he has in his career, both Buffalo and Kansas City were next in their matchup, and then the Browns were fourth, with Joe Flacco guiding them to victory (the Jaguars were fifth, to be fair). Baltimore came in a close sixth in their strong showing, as well.
Meanwhile, Detroit, Carolina, and Philadelphia were among the four teams in negative double figures, all in double-digit losing efforts. The Texans were at -8.9% in their worst offensive game in some time. Obviously they were in a tough matchup, as were some other teams on this list, and they also lost C.J. Stroud to a concussion. I’ve heard the arguments that PROE misses where some offenses need to protect their QBs and you can’t just throw every down, always. And I don’t disagree.
Likely some of what impacted the very run-heavy Week 13 — and even some of those Week 14 results — was weather, and some was just variance. But I do think even controlling for those elements, there’s very interesting data to track with PROE right now, that helps explain some of the successful and less successful NFL teams.
My hypothesis is this: In 2022, when rushing efficiency spiked alongside the pass defense shifts, offensive coaches learned they needed to get back to some basics in rushing the ball. And now in 2023, defenses got even better, by evolving to also take away some of that rushing efficiency. Meanwhile, QB health has been a huge story, and passing offenses have still be horrendous.
But my hypothesis follows that creatively throwing — in nonobvious situations and with the intent to pass to set up the run — has become a huge part of the “offensive counterpunch.” Or, perhaps more accurately, the counterpunch is there somewhere in the Xs and Os of the passing game — in a way that I can’t articulate — but we’re seeing results from the coaches willing to lean into that.
A key part of my hypothesis is that many of the other coaches are more or less still treating things like it’s 2022, having forgotten some of the lessons of, say, the offensive explosion in 2020, where so many offenses were willing to lean into pass-first tendencies. That was only a brief glorious moment for NFL offenses in the long-term ebb and flow of these things, before defenses quickly understood that focusing on the pass applies to both sides of the ball.
[As an aside, it’s always important to keep in mind there are just 32 teams and thus individual decision-makers can really impact results, and part of the evolution of the sport is how coaches get replaced and new ideologies spread, so these things do take time. For example, after the Seahawks’ successes in 2013-14, their Cover-3 defense spread through Gus Bradley and Dan Quinn and more.
But offenses found counterpunches when the personnel didn’t include an Earl Thomas-level talent at free safety, and those defenses got picked apart for a couple years before those systems eventually got replaced (by coaching changes mostly, which again take time). The more recent defensive trends are credited to Vic Fangio mostly, and his disciples like Brandon Staley. The recent trend on offense is the Shanahan/McVay tree, obviously.]
So my hypothesis is that offensive coaches were starting to learn the lessons of throwing on early downs, and you — as a football fan who is plugged in enough to still be reading Stealing Signals intros in December — were starting to hear more and more about PROE and the positives of throwing with intent, and then defenses quickly found answers, and the 2022 counterpunch did feature more running (including mobile QBs) — but that counterpunch was a band-aid on a major wound as these teams like the Giants and Falcons who overperformed last year with those philosophies were mostly just good by comparison to other struggling offenses, not actually legitimately dynamic in a way that could win playoff games — and now here in 2023 we actually do have some passing counterpunches, but we’re not quite seeing the results in the league-wide data yet because too many coaches didn’t really commit to what drove the 2020 scoring environment (those pass-first lessons didn’t have enough time to get deeply engrained enough, and it was easy to revert back to old pre-2020 run-heavy tendencies, especially given turtling up was a lot closer to optimal in 2022).
(Takes deep breath.)
Now one thing we’d expect if all this is true is passing efficiency returning generally, and as I mentioned we haven’t seen that. But the theory goes that offenses aren’t taking advantage of clever ways to throw on nonobvious passing downs enough — the big talk back in 2020 and “The years of PROE” was how your mix of pass attempts leaning more toward nonobvious situations than obvious situations would really improve overall passing numbers because of the massive gap in efficiency between the two situations. That gap in efficiency is obviously the result of how the defenses are aligned, and what they are trying to stop, where catching them off-guard naturally raises efficiency, but they’ve always been good when they know passes are coming.
And so much of this gets back to the talk of running effectively to try to get defenses out of two-high, with the hope being they’ll bring more guys into the box and then you have those advantageous passing situations, and how defenses weren’t really willing to do that in 2022. Maybe they are more frequently in 2023? Maybe the ways they disguised some of these things have been found out by QBs and offensive coordinators? I’m not sure.
All these tenets of football are interrelated, and I’m not trying to deny any of them, rather trying to bring them all together in an understanding of what is going on here. And it’s complicated and I’m probably wrong on parts of this, but I do think there has to be some validity in here as well. Or at least that’s the optimist in me.
Regardless, what I’m seeing more and more right now is that the surprising offensive outcomes we see on Sundays are probably going to show up with strong PROEs in my data on Monday, and vice versa. And as that correlation gets stronger again here in 2023 — in a way that I didn’t feel like it was in 2022 — I’m also considering that there are coaches succeeding by leaning into it. It’s been a staple of Mike McCarthy and Dallas since the bye. The Packers just started doing it aggressively. The Colts are in playoff position with Gardner Minshew in part because of a big recent shift in PROE that has them fourth in the NFL over the past month.
This is a stat that does vacillate a lot week to week, in part because expected pass rates vacillate a lot, and we’re talking about 60-some-odd playcalls that make up a whole game, where maybe a dozen or two key playcalls really dictate a lot of this over expectation element of it (every team has some degree of “balance,” is what I’m saying). And there are game situations that force teams away from this — just as it’s obviously not the case that getting your running back 20 rushes is an automatic win, it’s also obviously not the case that throwing every down automatically wins, either. Maybe there’s even some correlation-versus-causation stuff here where the coaches willing to throw with intent are also the ones able to do it creatively enough that it’s effective (I certainly don’t think PROE was Matt Canada’s problem, for example).
But I do think we’re trending back toward something offensively, and away from something else that defined 2022. And that’s been a major item to look for in the 2023 season — this offensive counterpunch — so I wanted to indulge myself a little this morning to talk through one way I think I’m seeing it, and how I’m tying that all together in my brain to explain what’s happened in the offense vs. defense chess match of the past three or so years.
The reason it’s fun, and was worth the time to think about, is the whole idea suggests that as coaches do get more aggressive again, we might start to see numbers improve. It might also be the case that defenses will just start to give up rushing efficiency again, and go away from the “this and that” defenses we’ve talked about (hat-tip The Playcallers, as always). But I guess that’s my whole point — for the first part of 2023, it did feel like defenses just got even better across the NFL. But Love in Green Bay and Flacco in Cleveland and even Zach Wilson looking legitimately good in Week 14 for the Jets — these success stories do suggest defenses can’t get away with this forever. They can’t take away everything. That’s at least something.
And my optimism again lies with the offenses that are willing to throw. They need to be healthy, functional offenses, of course. The Chargers throw a good amount, but that offense is broken. But we’ve seen offenses like the Commanders go away from the high PROE stuff, and struggle. After a 4-5 start, that team has lost four straight while settling in as a middle-of-the-pack PROE team in that stretch.
Again, only 32 teams. Sometimes tough to draw sweeping conclusions from league-level data, and each of those 32 data points does have confounding variables. I’m 2,000 words into this, so let’s get to the games.
Data is typically courtesy of NFL fastR via the awesome Sam Hoppen, but I will also pull from RotoViz apps, Pro Football Reference, PFF, Next Gen Stats, Fantasy Life, the Fantasy Points Data Suite, and I get my PROE numbers from the great Michael Leone of Establish The Run. Part 1 of Week 1 included a glossary of important statistics to know for Stealing Signals.
You can always find an audio version of the posts in the Substack app, as well as easier-to-see versions of the visuals at the main site, bengretch.substack.com.
Patriots 21, Steelers 18
Key Stat: Ezekiel Elliott — 7 HVTs (tied fourth most in Week 14, through Sunday)
The Patriots are another great example of a team that came out passing to help a young, unproven QB. They only finished with a PROE of +2.6%, but they had 23 first-half dropbacks against just 10 rushes, and per RBsDM, posted a +17% PROE when win probability was lower than 70% (more neutral situations, specifically earlier in the game, before they got out to a 21-3 lead). With Mitchell Trubisky struggling, New England eventually ran the ball quite a bit in the second half.
Bailey Zappe was kind of dealing in the first half, including a great throw and catch on a second touchdown connection with Hunter Henry (3-3-40-2) for his third TD of the first half. JuJu Smith-Schuster (6-4-90) also came alive for a revenge game, Tyquan Thronton (5-3-17) was uninspiring, and Zappe did plenty of swinging the ball out to Ezekiel Elliott (22-68, 8-7-72-1) as well.
Elliott dominated the snaps with a 91% share, and ran routes on a massive 78% of dropbacks. I’d argue this was evident in the way they’d used him behind Rhamondre Stevenson all season, and the usage after Stevenson’s injury in Week 13 (I’ve been out on Elliott for about three years but still had him as the third biggest “Signal” last week behind only De’Von Achane and Zack Moss). Elliott did struggle to run efficiently, but the receiving role was massive, and his 7 HVTs got him there. The Patriots cut Ty Montgomery (2-1-17) after the game, and it’s possible newcomer JaMycal Hasty will have a bigger impact on Zeke’s role. But I kind of doubt that, and while I definitely wouldn’t rank Zeke as a high-end option this week, there’s only so far you can push him down. It’s also an advantage for these RBs that are pretty fresh to be taking over in December when defenses are worn down. You just don’t see a lot of RBs with the potential for 90% snaps, and the touches that go with that give you a ton of paths.
For his part, Hasty becomes a reasonable stash with some HVT upside; I’m always talking about how these guys come out of nowhere, and it’s often these really thin RB rooms where the last man standing can wind up with a massive role, like we saw with Zeke.
Though this game wound up 21-18, the Steelers had just 112 yards in the first half, nearly all of which came on a 75-yard touchdown drive after they had already fallen behind 21-3. Trubisky really struggled, throwing one interception that got wiped away by penalty, another near pick that was dropped, and a third that was an actual interception, and a really poor one off his back foot where he didn’t step into a hit. Look, I’m not saying I’d have the courage to do that, but the end result was a floated ball into the middle of the field, which just can’t happen (a safety cleanly and comfortably caught it right in the middle of his frame like it was thrown to him). The chatter on Twitter was Trubisky needed to be benched, but he did lead a TD drive before half, so he got the second-half nod, and apparently he did enough there because it sounds like he’ll be starting again this week. None of that is good for the Steelers’ offense.
Diontae Johnson (7-3-57-1) hit for a nice TD, while George Pickens (6-5-19) had a bizarre route tree with all underneath catches, a rarity for him. Pat Freiermuth (7-3-18) drew volume but was inefficient, which is easy to chalk up to the QB play.
Najee Harris (12-29, 3-3-19) got more work than Jaylen Warren (7-11, 5-4-29), but I can’t really say Warren was better, because neither did much. It was close to an even split, which since the bye it’s been an even/odd thing where one week it’s basically even, and the next Najee gets a 55/45 or even 60/40 edge. There’s probably nothing here, but these are their snap rates over the past eight games, with Najee on the bottom:
Signal: Ezekiel Elliott — 91% snaps, 78% routes, 7 HVTs (the rushing efficiency is a minor issue, but 78% routes and 7 HVTs is very positive); JaMycal Hasty — becomes a reasonable RB stash after Ty Montgomery’s release as only RB on Patriots’ active roster behind Zeke (assuming Rhamondre Stevenson is out)
Noise: JuJu Smith-Schuster — 90 yards (strong game against ex-team but a lot of pass-catchers were out and this was just his second game over 35 yards all season)
Buccaneers 29, Falcons 25
Key Stat: Bijan Robinson — 7 HVTs (season high; tied fourth most in Week 14, through Sunday)
Tampa took down Atlanta after a back-and-forth final quarter that featured two scores from each team, and a couple of late lead changes. These types of late-game situations where both teams are converting can keep both offenses aggressive and really lead to fantasy points racking up late (as well as actual points, with the 25 total points scored in the fourth quarter being evidence of that).
Mike Evans (6-1-8) was a clear focal point of the Falcons’ defense, and got locked down by A.J. Terrell, something Tampa didn’t have great answers for early. Chris Godwin (11-5-53) ultimately dominated volume but wasn’t great, and Cade Otton (5-2-16-1) caught a fourth-quarter TD. Evans had a near-miss of his own, where a hand came down out of bounds just before his second foot came down in. The Bucs routes are more concentrated than maybe any team in the NFL, with the same five guys all being consistently strong for nine straight games since the Week 5 bye. Trey Palmer (1-1-5, 1-3) is the fifth.
Actually, Rachaad White (25-102, 2-2-33-1) is the fifth, and his routes remain elite. A big reason I would argue the Bucs have such a high concentration is they have a RB who consistently runs routes on over 75% of dropbacks, reaching 87% in this game. The role is killer, and his rushing efficiency has even improved of late, though he’s still averaging just 3.7 yards per carry for the season. But as a really efficient receiving back in a high-snap role with a ton of HVTs, he’s been fantastic, and my lack of enthusiasm on his ability to maintain this role from my early-week writeups is one of my bigger in-season regrets/misses.
Another great RB role continued to blossom in this game, as Bijan Robinson (10-34-1, 7-5-54) has finally really started dominating the key snaps. His 7 HVTs included 3 green zone touches, and he could have scored twice; an earlier sequence — before the drive he capped with a later touchdown — included him getting tackled on the sideline at the 2-yard line where he seemed close to breaking free for a score, and then being wide open in the flat on the next play, only for Desmond Ridder to throw wide when a defender was vaguely in his throwing lane. I’m expecting Robinson to play a key role in the fantasy playoffs wherever his teams have made it, as he’s started to consolidate work over Tyler Allgeier (9-40) and Cordarrelle Patterson (3-7), and his schedule is nice. One fun Bijan stat is he is at 4.7 yards per carry despite only one carry longer than 25 yards (and even that was “only” 38). He pretty clearly has the skill to break big plays, and when those come — whether this year or next — he’ll be a 5.0+ YPC kind of back who also has the receiving ability he’s displayed. In other words, he’s lived up to the hype, despite a lack of massive fantasy performances.
Kyle Pitts (6-3-57-1) got loose for a lengthy touchdown on a downfield reception, which I’ve been waiting about two years for, and Drake London (11-10-172) wound up with a massive line after racking up catches of 45, 16, and 28 on the final two drives. London had already posted a strong day at that point, but he more than doubled his yardage with those three late big plays (which is also the kind of thing we’ve been waiting about two years for, i.e. it was nice to see Atlanta throw 40 times, and Ridder did ultimately have a pretty strong game).
Jonnu Smith (7-4-27) saw some decent volume, as well. Ultimately, this game was huge in that Tampa came from behind to win when a Falcons’ victory would have gone a long way to securing their spot in the playoffs, given they have a soft remaining schedule relative to their two division rivals (I’m of course rooting against the Falcons because I really can’t do another offseason and season writing about Arthur Smith, and while I’m not sure if they’ll move on, it’s far less likely if they win the division).
Signal: Bijan Robinson — season highs with 7 HVTs, 3 green zone touches (role looks very strong heading into fantasy playoffs)
Noise: Mike Evans — one catch (was a clear focal point of Atlanta defense, still nearly caught a late TD to salvage some production)
Ravens 37, Rams 31
Key Stat: Lamar Jackson — 546 air yards (most by a QB in a game this season)