clock menu more-arrow no yes mobile

Filed under:

Duck Tape: Film Analysis of USC 2023

A preview of Oregon’s week 11 opponent in Autzen

Utah v USC Photo by Meg Oliphant/Getty Images

Special thanks to Alicia de Artola Castillo of Reign of Troy for joining me on the Quack 12 Podcast to discuss the Trojans’ roster: LISTEN HERE

The central question about USC is why, despite returning the entire staff and nearly all key personnel from last year, they appear to be struggling so much more than in 2022. The tempting answer offered in popular media is that the defense had become such a liability that it was no longer survivable. While it’s certainly true that the Trojans have significant defensive problems and this is the team’s main overall handicap, I don’t think it’s the right way of looking at the above question. After charting all of USC’s games, performing a statistical regression, and enjoying a long and productive conversation with Alicia, my conclusion is that a marginal decline in offensive performance has greater explanatory power regarding 2022 vs 2023 game outcomes, and this article will demonstrate how.

As discussed on the podcast, the equation is the same as in 2022 – the explosive qualities of the offense and the improvisational genius of the quarterback mask the offense’s efficiency and blocking problems, and allow the Trojans to outscore opponents despite the defense giving up vast amounts of yards and points. The variable within that equation which has changed is not the value of the defense degrading – if anything, it’s improved slightly on a couple metrics, though obviously it was still appallingly bad and as I’ve spent quite some time documenting that was former DC Grinch’s responsibility so his dismissal this week was appropriate.

Rather, in my opinion the reason USC is having a worse season than in 2022 is that the offense has gotten even less efficient, and playcalling has become significantly more predictable with an even more irrational mismatch between individual playtype frequency and its success rate. As such the offense can no longer automatically outscore every opponent, even though it still performs at a very high level compared to virtually every other FBS team.


Head Coach Riley continues to operate his schematically fascinating power-RPO offense at a high level, with reigning Heisman winner #13 QB C. Williams an excellent fit for it. This year USC’s offense is ranked 5th in F+ advanced statistics, and finished last season in 1st place.

Williams’ performance as a pocket passer is very high level, with elite arm talent and pocket manipulation tools. Some examples:

(Reminder – you can use the button in the lower right corner to control playback speed)

  1. :00 – This zone blitz with the DB coming from the boundary and the linebacker moving over to fill in was never going to work, Williams reads the field much too quickly and can fit the ball in that lane much faster than the defense can occupy it.
  2. :06 – Misdirection plays are a big part of the passing offense, here getting the defense to bite on an initial screen block from #87 TE McRee with Williams fooling the safety with a shoulder fake, then a tight spiral to McRee wide open down the sideline.
  3. :22 – Williams doesn’t require ideal mechanics for throwing power, as here with the LG beat and LT getting worked back, so he has no room to step into the throw and has to do it all with his upper body – it’s still placed perfectly on the 25-yard out pattern.
  4. :31 – This drag route is all over USC’s playbook, Riley uses it as both a man- and zone-beater (the former because his receivers will outpace coverage, the latter because defenses rarely have midfield bisection rules). It requires a longer pocket than the protection can typically deliver on their own, but Williams is a natural at working up and down the pocket to keep rushers from getting a good angle and buying himself the critical extra seconds for a receiver like #16 WR Washington to come open.

On my tally sheet this year, USC has an aggregate per-play efficiency rate of about 52.5% excluding garbage time (281 successful plays vs 255 unsuccessful ones, given the down & distance), averaging 8.0 adjusted YPP with a little over 20% achieving explosive yardage. Those are exceptional yardage and explosiveness figures, but just a bit above average efficiency rate, in my experience.

That separation – more explosive, less efficient – expands on the small but noticeable discrepancy in their 2022 numbers, in which they had a 58% efficiency rate (only a couple points under the typical 60% championship threshold, unusual for an elite offense but not by a huge amount), but averaged an incredible 7.7 YPP and 19.5% explosiveness. So in 2023, USC has fallen even farther in efficiency by a pretty significant 5.5 percentage points, but have actually gotten more dangerous when they do connect by a third of a yard per play and half a point in explosiveness.

It’s also the case that USC’s offense has declined in effectiveness since the beginning of this year, with a clear breakpoint between their first five and most recent five games of the season. Here are the metrics I track from charting, broken out between those two periods:

Much of that has to do with simply playing tougher defenses – USC’s first five opponents averaged a 96th ranking in defensive F+, while the last five averaged 33rd. But as Alicia and I spent the balance of the podcast discussing, the offense this year also has some issues with predictability as time has gone on, increasingly relying on Williams’ improvisational ability and long-ball passing which has made it significantly harder for the Trojans to operate if they get behind schedule.

Due to those issues and protection problems, called passing plays this year have resulted in a sack, scramble, or throwaway on about 37% of dropbacks, the highest rate in the conference (with the potential exception of UCLA, I haven’t had occasion to keep up on charting the Bruins as I have with the rest of the Pac-12, but I’ll catch up this offseason). Watching film, it’s easy to understand why this is not just a survivable issue but might even be an asset: Williams is quite possibly the most impressive scrambler I’ve ever seen, with instinctive pocket manipulation skills to extend plays and, astonishingly, greater accuracy throwing deep downfield on the run than on designed plays with clean pockets.

Bringing him down or inducing an unproductive play requires the defense winning in not just a single way, but getting through the protection from multiple directions and sticking with coverage for an almost absurd amount of time. Some examples:

  1. :00 – This sack required not just the LG getting beat up the middle, but the escape hatch being closed by the LT getting beat too. The RT’s lost his guy, and this play follows another sack in which the C and RG were beat, to complete the sweep.
  2. :11 – This pocket manipulation is just astounding, and against one of the best pass rushing teams in the nation. Watch Williams’ helmet, he’s doing this all by instinct and peripheral vision while keeping his focus downfield and begging anybody in his receiver corps to get open, until he finally does it himself.
  3. :29 – Here the RG is beat, but Williams buys a bit more time drifting back and away while keeping his eyes downfield, and it’s not till the RT is beat too and he can’t make it out of the pocket that he releases, while falling away … it’s even on-target, he just doesn’t see the DB come off the TE crosser to break it up, which is a very Cal thing to do.
  4. :41 – I count seven seconds from snap to release here, and the accuracy of this throw to a small sideline window while on the hoof is incredible.

For the entire year, USC’s passing offense is underwater in terms of efficiency at 49% (167 successes vs 174 failures), though still at championship-caliber yardage at 8.7 adjusted YPA and 19% of passes gaining 15+ yards. Per the above chart, however, in the last five games they’ve operated as an almost perfectly average FBS passing offense in terms of yardage and explosiveness, and significantly below average in terms of efficiency. That reflects a higher than expected incompletion and sack rate, and incorporates a screen pass success rate of only 33%.

A good deal of USC’s efficiency issues stem from their offensive line, which needed to replace multiple starters from last year and did so through three portal transfers. Longtime readers will know I’m highly skeptical of such projects, but I’ve been watching USC closely as a potential counterexample to the well documented trend in the 2018-2022 dataset of heavily portal-based lines being unproductive. That’s because they got two of them, #52 OL Kingston from Wazzu and #71 OT Tarquin from Florida, very early in the process, and the third, #70 LG Pregnon, was one of the most valued transfers in the entire cycle.

Unfortunately for the Trojans, math is undefeated. USC’s cumulative o-line error rates on my tally sheet are two to four points worse than last season, and they’ve undergone quite a bit of shuffling on the right side. At first they had Kingston (a natural guard who played left tackle, poorly, for the Cougs) at RG, but he was in a three-man rotation – Tarquin and returner #76 OL Murphy would switch back and forth at RT, and then Kingston would move out to RT and Murphy in to RG. Last week they seemed to finalize it with Tarquin on the bench, Kingston at RT, and Murphy at RG all game long, with exactly the issues one would imagine from playing two guys out of position without much experience.

There also doesn’t appear to be a “mistake-eraser” in the receiver corps, which is very talented but as Alicia and I discussed, has been surprising in that the most productive was a previously pigeon-holed possession receiver. Both of us are hard pressed to explain the falloff in production and explosiveness in the other Oklahoma transfer, #4 WR Mar. Williams, compared to the previous seasons I’ve watched his film, and while it’s clear they’ve got some extremely promising true freshman like #1 WR Za. Branch, #19 WR Robinson, and others, as is typical for receivers they’re probably a year a way from consistently contributing.

As such, the designed progression passing attack – setting aside RPOs and scrambles – is actually the least effective part of the Trojans’ offense, and the fact that it makes up such an outsized part of the playcalling (on all downs, and even more so in predictable situations) is the largest explanatory factor in understanding USC’s recent offensive downturn and as such their performance record compared to last year.

Here’s a representative sample of failed designed passing plays:

  1. :00 – Early in the season I was startled with just how easily Stanford’s DBs – hardly a bluechip crew — had USC’s receiving targets locked up across the board when they tried progression passing. The Trojans shredded them on the triple-option RPO as we’ll see below but they weren’t getting anything from the pocket until well into garbage time.
  2. :14 – Notre Dame’s approach was to go right through the o-line, sacking Williams six times and picking up three interceptions (and nearly a fourth) off of scrambles. Here’s the last meaningful sack of the game, in which the entire line collapses on the blitz and the man coverage only needs to hold up for about two seconds.
  3. :27 – Miscommunication here, Robinson has likely overrun the route but Williams has time without pressure to read his momentum and put it on him instead of demanding he make the tougher catch going back into the corner he’s beat. Robinson has a bright future but Williams keeps giving the freshman throws that he’d need to be a senior to catch.
  4. :40 – Even with the benefit of looking the other way on how grabby #79 LT Monheim is (the o-lineman with the most holding flags in the Pac-12 last season), Williams has been oddly hot-and-cold from the pocket on his deep ball this year. He has a few jaw-droppers of course, but several more like this where he has time to set up and an open man but just misses him. His platform is really twisted, it’s like every throw has to be a big whip to beat pressure instead of a mechanically smooth delivery.

The designed rushing offense has a 58.5% efficiency rate (114 successes vs 81 failures), with a 6.8 adjusted YPC average and 21.5% gaining 10+ yards over the entire season. It’s come down a bit from the elite rushing numbers over the first five games, but remains green across the board during their last five with above average efficiency while sustaining championship-caliber yardage and explosiveness in the run game.

Another offensive trend that’s continued and expanded from 2022, which Alicia and I have discussed with some frustration for two straight podcasts, is the mismatch between rushing effectiveness and frequency – the Trojans were more than eight percentage points better at running the ball vs throwing it last year, and the gap has expanded to over nine points this year. And yet they only run about 38% of the time, if anything slightly less this year. On the podcast Alicia pointed out that Riley seemed to abandon the run at the first sign of trouble; I dove into my charts after we talked and found a strong correlation between drive count and run rate – throughout 2023, during the first four possessions the Trojans have called run plays 43.8% of the time, but only 30.7% for the rest of the game (excluding garbage time), an increase in the already strong passing bias of more than 13 percentage points.

One of the clearest examples of playcalling issues we looked at on the podcast is how 2nd downs in 2023 look like 3rd downs in 2022 – the Trojans are almost always going to run in short-yardage, and pass the rest of the time. Whereas in 2022 USC had a healthy mix of between 35-45% rushing in any 2nd down situation and were well above water in success rate no matter what, in 2023 their 15% rush rate on 2nd & medium or long has meant they’ve fallen at least six percentage points in efficiency whether rushing or passing in those situations, with some of them falling underwater. Another dramatic example of situational playcalling problems compared to last year is 3rd & long, where the conversion rate has fallen from 2022’s fairly normal 43.4% outside garbage time to just 24.2% in 2023.

The key to the successful run game, despite what in my opinion are some clear offensive line issues, is the way that Riley’s offense controls the box count through a novel power-blocked RPO scheme. This is my third year of studying every one of Riley’s games and I continue to be amazed by how well he’s combined an Air Raid route tree with RPOs and then used power instead of zone blocking as the basis for it, any two of which would seem to be like oil and water. Only a few teams have figured out how to recognize and stop it, while most never solve it get utterly decimated by it all game long. Some examples:

  1. :00 – Note how it’s a WR (Branch, here) as the H-back to the fieldside, that’s a giveaway about 80% of the time. Stanford’s defensive staff ineptly treats that position as a blocker (they’re from Wisconsin) and never assigns anyone to cover him, instead sending the entire box at the run. Williams pulls it on reading the end, and then doesn’t even have to perform the second part of the read – a DB or LB choosing between attacking him or covering Branch — because nobody’s there, so he just tosses it for a big gain. This happened nine times and then the game was a blowout.
  2. :13 – Notre Dame has the formation cracked, look at how they only have four guys – three DL and a LB – actually playing the run, while it’s desultory single coverage for the three decoy receivers to the outside and slot. The three defenders who matter are going right for the three members of the triple-option: the end crashes the back to force the QB pull, the LB on the QB to make him dance, and the NB screaming down onto the “H-back” to beat the slot man’s block and take away the toss, earning the throwaway.
  3. :20 – This is the other RPO, with a flanker screen which can be identified by the offset pair right next to each other on the fieldside. The formation is meant to get the defense heavy in the box, pull seven defenders inside with boundary CB out of the play as well, and neutralize one more by getting him to come down on the QB who’s pulled the ball and run laterally. Now they’ve got a two-on-one advantage and can get positive yards.
  4. :29 – UW doesn’t repeat Stanford’s mistake and maintains coverage on both the QB and H-back. The curious thing about how central this RPO is to USC’s offense is that they don’t take optimal run reads very often – this should have been a handoff with how wide the end is, and it should have been a QB keep with how the safety is beating the blocker.

That makes for lighter work for the o-line in the run game, and the gap schemes which are the foundation of the core RPO play when they hand the ball off are simple for them to execute because by definition they’re one-on-one. It helps that they have excellent backs in #6 RB A. Jones, #0 RB Lloyd, and when he keeps it, Williams himself. Some examples:

  1. :00 – The simple threat of the RPO is enough to get five yards here – field trips pulls one LB out of the box and the LT’s pull gets the other one to take the decoy gap, so Jones gets through even with a poor seal from the RG.
  2. :07 – USC won this gap scheme run against Utah, with each OL getting a hat on hat cleanly at the line of scrimmage and then almost so at the second level, every single time Riley called it … which was why it was so odd he only did so twice in the second half. Here’s the last one.
  3. :14 – Cal’s in dime with five in the box for some reason so this one is real easy, and the dime defender just sort of wiggles seven yards deep while waiting for Branch to hit him rather than squeezing down hard which might have limited this.
  4. :22 – McRee is blocking the wrong guy here, he’s supposed to bluff the end and let Williams read him, then move up and actually hit the ILB. He accomplishes none of those goals and the end crashes the back so Williams pulls and has an unblocked backer to deal with (a former Trojan himself), which he does adroitly, then humiliates a safety for good measure.

There have been two chief causes of failure in the run game. First is simply when enough defenders defeat their one-on-one blocks in those gap schemes that the whole thing falls apart. Second is when they switch it up and call zone-blocked plays, which this line just can’t execute (I suspect this is where the lack of gelling and developmental time hits the hardest) and there is nearly a 30 percentage point falloff in success rate compared to power-blocked runs. Here’s a representative sample of failed rushing plays:

  1. :00 – The LT is supposed to come off his combo to pick up the backer here, but even two-on-one they can’t control the DT much less hand off to the LG alone so he can get in his second-level block. Lloyd tries cutting across but the C has lost control and the TE has grown tired of blocking so that’s a no-go.
  2. :09 – The overhead angle here is excellent for showing how messed up the zone assignments are – it’s not just the TE’s whiff, with the defense’s alignment the RT is supposed to be going the other way, the C is supposed to be sliding right and then up into the DT to open the gap, and the LG chips and up. It’s like the entire line has their assignments backwards, except the LT and RG. The back gets simultaneously tackled by six defenders.
  3. :30 – This is the one-inch punch problem – zone blocking requires harvesting power from the lower body and up through the strike with proper technique because you don’t get a big buildup of running like with those gap scheme pulls. As stationary targets they just get bowled over by bigger linemen because of sloppy footwork and lunging.
  4. :36 – Cal’s best defender shows how you do it – immediately get inside the RT to disrupt the pull. Now there’s a free backer coming at Lloyd, who bounces out, and the backer whom Pregnon initially hit is now on the correct side of the play to chase him down.


USC’s defense is currently ranked 88th in F+ advanced statistics, up a bit from last year’s finish of 94th. On my tally sheet, the Trojans have an aggregate defensive success rate of 48% (295 vs 320), allowing 6.6 adjusted YPP with 18.5% of opponents’ plays achieving explosive yardage. That’s about two points better in efficiency, 0.6 better YPP allowed, and half a point better in explosiveness allowed compared to 2022.

The entire improvement is limited to the first five games, however, with the most recent five games being at 2022 or worse levels. Here’s the breakout:

It’s impossible to say what changes new co-DCs Nua and Odom might make to the defense after Grinch’s dismissal, though as Alicia pointed out on the podcast their previous responsibilities were defensive line and linebackers, respectively, so it’s hard to imagine them instituting sweeping changes to the front. Schematically, this defense has gradually shifted from a 3-3-5 with an innovative rush end and frequent stems and shifts to a standard issue 4-2-5, and structural changes are unheard of week-to-week in college ball even with a coaching change. I’ve never seen the Trojans in two years play anything besides a nickel regardless of offensive formation, down & distance, or field position, and even last week when they constantly dropped eight into coverage, it wasn’t a third ILB but rather just keeping one of the DEs at depth.

Over the offseason, USC made some personnel changes and transfer portal additions to bolster their underperforming defensive front and to make up for a couple of losses of the rare high performing player. In fact, the Trojans made more moves here than than several of the other Pac-12 teams I felt were too complacent regarding their own poor defenses or facing more substantial losses. But given the size of the problem, I still don’t think USC did enough work here, and their hit rate on new players hasn’t been nearly high enough to affect the bottom line. In my opinion, only one new defensive player has performed at above replacement value, #90 DT Alexander from Georgia, with the rest being no different or even less effective compared to what they had last year in the front.

Alexander has contributed to a more effective interior pass rush — it’s certainly not coming from the additions at the edges, the only other guy who’s getting even a little pressure is returner #51 DE Byrd, a 2-star senior who’d previously transferred in from Wyoming – and that largely explains the bump in pass defense success rate, especially at the beginning of the year. Otherwise this front is absolutely dependent on the blitz to get home and gets nothing done rushing four if Alexander is contained or has rotated out.

I also think that the transfer of #17 CB Roland-Wallace from Arizona has had a steadying effect despite a bit of a talent ceiling (and possibly a long term hindrance to the development of a well recruited room), and #1 CB D. Jackson is coming into his own as a cover corner, though Alicia thinks he looks lost at times. Here’s a representative sample of successful pass defenses:

  1. :00 – Good bullrush up the middle by Alexander to force an errant throw.
  2. :08 – Nice physical PBU by Rolland-Wallace.
  3. :21 – That’s Byrd beating the RT to hit the QB, and Jackson in tight coverage.
  4. :34 – The high-scoring game last week obscured a weird stat – UW only hit three downfield passes over 15 yards from the pocket, otherwise it was all screens, runs, and scrambles as USC mixed a three-man rush with a lot of blitzing. This one clearly affected the passer as he places the ball five yards out of bounds.

There are several weird things about the pass defense. First is that I’ve never seen any rotation at corner over the last several games, with five different bluechips getting zero playing time at all except special teams, which Alicia and I thought was unhelpfully reminiscent of the previous staff. Second is that despite having a fairly large linebacker room that Alicia spent a long time parsing during our offseason podcast interview, they’ve only played four of them who all have abysmal grades on my tally sheet, without giving anybody else a shot. Third is that the safety room has somehow – reader, I wish to stress I would have thought this impossible – gotten worse compared to last year. That unit was Grinch’s responsibility as well, and Alicia told me it will now be taken over by analyst and former Trojan safety Taylor Mays, so perhaps he’ll bring some hard-headedness to the group.

In my opinion, those personnel issues are most borne out in the situational effectiveness numbers, which have collapsed even further in short-yardage while improving a bit in medium and long yardage compared to 2022. Last year, USC had situational defensive success rates between 25% and 36% against the run and the pass in 2nd, 3rd, and 4th & short – those were bad numbers, but you have to have a genuinely good defense to be better than even in short-yardage. But this year in every one of those situations the Trojans are now under 20%, and worst of all, they’re only winning 2nd & short – the deep shot down, for obvious reasons – at an 11% rate.

Here’s a representative sample of failed pass defenses:

  1. :00 – The broadcast gave half a dozen replays of this and I still can’t figure out what coverage USC is in. Cover-0 I suppose, there’s no safety over the top, but with one backer on a delayed rush, another covering the RB, and a CB is spying the QB (?). At any rate the nickel is supposed to cover the No.2 receiver on the post in man and can’t, and he doesn’t have any help.
  2. :29 – Utah’s in a jumbo package with a sixth OL in an ineligible jersey on the left side, so when the WR goes in motion the CB over him doesn’t have a whole lot left to do. But I can’t tell if they’re in zone or man – him staying put says zone, but the other CB’s leverage says man. Anyway, both the backer and the safety fire onto the back leaking out so the motion into wheel is wide open, meaning the defense got outsmarted by Andy Ludwig.
  3. :43 – Play action pulls eight defenders into the box so there’s no underneath coverage and neither Jackson nor Roland-Wallace was able to break up this slant pass all day, so it gets nine yards through the air and another eight on the ground. Cal ran this exact play a dozen times and USC never figured it out.
  4. :51 – The Trojans played three on the line for much of the game last week, and furthermore bailed out everybody from midfield coverage to take away deep passing. UW responded with a lot of running and inside screens, like this one in which a freshman OL picks up the slim LB and carries him four yards downfield.

USC’s rush defense is badly underwater for the season, with a 42.5% success rate (122 vs 164), allowing 5.6 adjusted YPC and 17% of opponents’ designed rushes to gain 10+ yards. This is the area that has the least amount of difference when comparing the first and last five games, since it’s pretty much been equally bad from the week 4 game against Arizona State onward, and during the first three games in which they performed well the opponents ran so rarely that the rush defense barely accumulated any per-play stats.

There seems to have been a deliberate effort to recruit smaller, speedier players in the front, who aggressively knife in to create havoc plays and disrupt the run before blocks can develop. And it does seem to be the case that the general category against which the rush defense is most effective I’d describe as outside runs with complex blocking schemes, where they can leverage that kind of advantage. In the alternate, sometimes Alexander just blows up the center or an interior guard and an inside run dies before the running back can bounce. Some examples:

  1. :00 – That’s Alexander just wrecking the center and ending the play before it can begin.
  2. :06 – Both backers guess the play correctly and slice in diagonally before the pullers can get their blocks set up – they’re just faster to the play than the offense is, and the DBs are right behind them on the same logic. This is the ideal for how a speed defense is imagined to work.
  3. :21 – Alexander and Byrd again.
  4. :28 – Not that there’s a ton of them on my tally sheet, but USC has a perfect success rate against option pitch plays. It’s just the structure of the defense and how the LBs trigger, both of them are going to hustle laterally which means they’ve got one defender for each potential ballcarrier, and there’s no risk of misdirection the other way from the offense to take advantage of that aggression.

But mostly, watching USC’s rush defense for the past two seasons has been a painful exercise of seeing undersized linemen getting washed down, no one ever setting the edge, and linebackers constantly missing run fits while misunderstanding the basic concept of a scrape exchange. Some examples:

  1. :00 – I think this is the most bizarre thing about watching USC’s defense, rush defenses where I’m utterly baffled as to what they think their run fits are. I don’t even know how to confidently grade this play or about 40 others on my charts – a DT, both LBs, and the free safety are all running to some crazy place that isn’t the play or anything close to it. #8 DB Zi. Branch (the WR’s older brother) makes the play because the Notre Dame receiver decides to stare at him instead of blocking; sadly he was hurt at the end of the Cal game and will miss the rest of the year.
  2. :07 – This is an unusual QB and RB combination but it’s a standard read option play otherwise, and the Trojans’ standard problem in stopping it with nobody defending the outside.
  3. :14 – Both backers stick their noses in early and get beat, and two d-linemen get thrown to the ground simultaneously by one of the worst o-lines in the league.
  4. :21 – I had a lot of options to choose from in this game to show the size mismatch between USC’s front and UW’s 4-star o-line. This one shows it pretty clearly, as well as a TE just washing down the entire line, nobody setting the edge, and the free safety taking a terrible angle. The only hit any Trojan delivers is against the umpire.