Georgia Tech Football: Pitt Advanced Stats Review – Offense

After I actively campaigned for it in this very column a few short weeks ago, Georgia Tech made a change at head coach before their game versus Pitt. Did it dramatically improve their effectiveness at playing football? On defense — well, Robert will cover that tomorrow. As for the offense, let me offer this (admittedly unrelated) opinion from some friends of the program:

Playing in Pittsburgh underneath the tendrils of a fading hurricane, Tech struggled to get anything going through the air and instead relied on explosive plays, special teams (!!!), and turnover luck to grind out a result. Some might be convinced to write off this offensive performance as part of the administrative and staff upheaval on the Flats, but I struggle to do that: the same offensive staff remains in place and there’s simply not enough time during a game week to do much more than tweak an existing system. Let’s dig into the details.

Overall

First, the charts:

The common refrain around football analytics is that “running backs don’t matter,” but if we completely ignore the context of that quote (that there is no inherent running back skill that is the greatest determinant of their success on a specific running play), it’s clear that that was simply not the case for Tech in Pittsburgh this year. This game was sloppy, it was wet, it was wild — and the only offense Tech could generate was via rush.

And what it did generate was, on average, pretty good: Tech’s rushers gained 6.6 yards per carry (259 yards on 40 rushes), boosted primarily by the efforts of Hassan Hall and Jeff Sims. Per Sports Info Solutions (SIS) via Bill Connelly, Tech had a 12.7% explosive play rate and 40% of its rushes went for five or more yards.

But averages are deceiving: Tech’s offense was buoyed by explosive plays — it was not consistently productive on the ground otherwise. As Connelly noted in his box score, “5 GT players gained 165 yards; the other 66 gained 169.” Of Sims’ 5.84 total rushing EPA, 4.58 came on an 18-yard touchdown run with 85 seconds left in the game. Hall’s exceptional day was mostly inflated on that same drive: his 63-yard rush to get Tech down to the Pitt 10 accounted for 3.38 of his 4.56 rushing EPA. Without these two plays, Sims averages 0.09 EPA/play and Hall goes for 0.06 EPA/play; neither of these marks are exceedingly poor by any means (the average rush in 2021 was for 0.00 EPA/play), but they’re down significantly from both players’ actual averages of 0.39 and 0.23, respectively.

These explosive plays recontextualize what was not a wholly effective day on the ground. Per SIS, 41% of Tech’s rushes were stopped at the line of scrimmage or prior and per ESPN, 51% of Tech rush plays went for two yards or fewer. When rushing on third down, Tech only converted 44% of those rushes into first downs. While Tech had 154.5 highlight yards (9.66 per rush opportunity), Tech only generated 16 rushing opportunities (41% of all rush plays) all game. By all accounts, this was an extremely boom-or-bust game.

Expected Points

Tech didn’t have a turnover in this game, but it did nearly have four passes picked off, which speaks to what was a rough day in the air for Jeff Sims (and technically Nate McCollum). In fact, Sims’ ~3 yards/dropback might be one of the worst marks I’ve ever seen for a consistent starting quarterback, period. That being said, his offensive line didn’t do him many favors: per SIS, Tech allowed a 48.6% pressure rate and 11.4% sack rate. To Tech’s credit, it’s starting to scheme up deeper pass plays for Sims (SIS had him at 8.9 air yards per attempt; Robert had him at 9.7), but he’s struggling to complete passes at all levels consistently (-17.9% CPOE per Robert’s charting).

I know there are a lot of Jeff Sims believers on Twitter, especially in pockets of the analytics community, but given his struggles this season and his underlying metrics (-0.21 EPA/dropback, 34% success rate, 3.96 yards/dropback), it’s hard to recommend that he continue to start at quarterback. As a thought exercise, here are the other options available to offensive coordinator Chip Long:

• In 30 total snaps (19 of them throws) for Clemson in 2021, Taisun Phommachanh posted a 0.03 EPA and 6.89 yards per dropback. On the ground, Phommachanh generated 0.09 EPA/rush in 11 carries for 69 yards.
• At Akron in 2021, Zach Gibson generated 0.19 average EPA on 110 dropbacks, completing 73% of his passes. He also rushed for 102 yards and 0.11 EPA/rush on 15 carries.

There are obvious caveats with both of these choices: we can only judge Phommachanh on a small sample size for a woeful Clemson offense, while Gibson’s numbers may be inflated given the amounts of garbage time available to his 2-10 Akron team. Nevertheless, given an interim head coach and seven games remaining in the season, it seems like Tech has an opportunity to get more tape on all of its available quarterback options so that whoever’s in charge next spring (whether parts of the offensive staff are retained or not) can make a more educated decision.

Situational

Per SIS and Robert’s charting, the Jackets ran constantly on first down (62%), early downs in general (57%), second and long (53%), and third and long (50%). I harped on this last time, so I won’t make too much of a meal of this point again, but after five games of unflinching inefficient rushing, it’s clear that this is a concept Tech is married to, for better or for worse (read: for worse). It’s possible that part of this reticence to make more optimal decisions (read: throw the ball when it best behooves you) is because of unease with performances at the quarterback position (see above). However, there have been opportunities to work out some of these kinks against talent-disadvantaged competition (Western Carolina) AND/OR try out other options at quarterback and practice optimal play-calling in garbage time of decided games (Clemson, Ole Miss), but Tech has instead remained true to its scheme and remained run-first.

Last week’s coaching turnover provides a chance for us to reflect on opportunities for improvement for this program across the board. With that in mind, I offer this quote from the last time I wrote an actual advanced stats column:

Tech is going to have to shore up some of its situational play-calling and tactical decision-making if it wants to gain every available inch and every available advantage against some of its tougher competition.

I wonder if this kind of data-driven mentality will be a part of the criteria for the next staff. As a newer fan (freshman class of 2015), my understanding of Tech’s ethos as a program begins and ends with optimizing for competitive advantage when at talent disadvantage (see: option, flexbone). When we talk about maintaining or building Tech’s program culture, that’s what I think of — fundamentally, I expect the smart school to make smart decisions to win football games. Tech might have won this past weekend while not doing so, but I remain unconvinced about the sustainability of that strategy.