clock menu more-arrow no yes

Filed under:

Advanced Stats Review: GT vs. Duke

GT’s most dominant performance of the season paired an overwhelming rushing attack with a surprisingly disruptive defense. 

Danny Karnik, GTAA

Final Score: Georgia Tech 56-33

Projected EPA Margin of Victory: GT by 30.93

GT Win Probability (Based on Success Rate, Yards Per Play, and EPA): 95%

Georgia Tech returned to competition on Saturday night and handily defeated a Duke team that has been a thorn in our side for the past 6 years. This was GT’s most dominant victory of the season, pairing an overwhelming rushing attack with a surprisingly disruptive defense.

We like to be honest around here, and this game had a completely different flow and outcome than we projected last week in our preview. Nine of Duke’s points came directly off of Georgia Tech special teams miscues; otherwise, this would have been a 30+ point victory for the Yellow Jackets. We mentioned in our preview that Duke’s defensive ends were elite rushing the passer but mediocre against the run; thankfully, Georgia Tech’s offensive staff made note of this and came in with a run heavy game plan that accentuated Tech’s strength and attacked Duke’s weaknesses.

Success Rate Comparisons and Individual Player Advanced Stats

GT Duke Success Rates

Success Rate O GT Offense D Duke Offense National Avg
Success Rate O GT Offense D Duke Offense National Avg
Down 1 43.30% 1 45.70%
2 38.10% 2 36.00%
3 64.30% 3 41.20%
4 0.00% 4 50.00%
Qtr 1 31.60% 1 31.60%
2 42.10% 2 42.90%
3 43.80% 3 60.00%
4 69.20% 4 25.00%
Pass P 39.30% P 42.90% 41%
Rush R 48.70% R 40.00% 42%
Overall 44.80% 41.80% 42%
Success rate is the baseline metric for efficiency. As a reminder, a successful play gains 50% of the needed yards on 1st down, 70% on second down, and 100% on third or fourth down.

Georgia Tech paired an elite offensive rushing attack, succeeding on 49% of its 40 called running plays, with a solid enough performance on defense. Notice that GT’s offensive success rate improved each quarter; Coach Patenaude and company did an excellent job of identifying GT’s relative strength and building on it as the game progressed. The called counter runs for Jeff Sims late in the game were brilliantly setup by run calls earlier in the game. Encouragingly, the defense locked down in the fourth quarter, holding Duke to 25% success as GT extended the lead from 2 points to 23 in the final quarter.

GT Rushing vs. Duke

Rusher Called Carries Rushing Success Rate Highlight Yards
Rusher Called Carries Rushing Success Rate Highlight Yards
Sims 7 0.71 53
Gibbs 6 0.5 72
Mason 18 0.39 39.5
Smith 5 0.4 45.5
McCollum 1 0 2.5

Looking at the individual rushing statistics, there is a lot to like. As a reminder, highlight yards measure the running back’s contribution to the rushing yardage, giving the running back partial credit for the early yards in a carry and full credit for the yardage after a gain of 8 or more. Georgia Tech had a stunning 212 highlight yards, as Jeff Sims and the running back group created explosive runs all night and powered a dominant rushing performance in spite of mediocre blocking. Jahmyr Gibbs was once again elite, pairing 72 highlight rushing yards with 49 yards after the catch on his 4 receptions. If Gibbs’s injury is serious, Georgia Tech loses most of its offensive explosiveness. His freshman campaign has exceeded any reasonable expectation, and he projects as an elite running back for the rest of his GT career.

GT Receiving vs. Duke

Receiver Receiving Success Rate Avg Target Air Yards Targets % of Team Air Yards Yards After Catch RACR WOPR
Receiver Receiving Success Rate Avg Target Air Yards Targets % of Team Air Yards Yards After Catch RACR WOPR
Camp 1 13 2 9.67% 3 1.12 0.37
Sanders 1 5 1 1.86% 19 4.8 0.26
Carter 0.5 18.25 4 27.14% 1 0.37 0.47
Harris 0.14 16.71 7 43.49% 3 0.07 0.39
Gibbs 0.25 0.33 4 0.37% 49 141 0.52
Mason 0.5 4.67 2 5.20% 7 6.36 0.08
McCollum 0 20 1 7.43% 0 0.4 0.05
*RACR compares a receiver’s total receiving yardage to the air yards on targets to them, and anything above 1 is good. WOPR tells us how heavily the receiver carries the passing game by combining the proportion of targets, air yards, and total receiving yards.

GT CPOE vs. Duke

Jeff Sims Passing Performance
Jeff Sims Passing Performance
Completion Percentage 56.50%
Completion % over Expectation -1.90%
CPOE on 10+ yard throws -20%
CPOE on 20+ yard throw -18%

Looking at the receiving and passing performance together, I’m confused. The leader in targets for the game was Peje’ Harris, who is certainly not our best receiver and had only one successful reception on 7 targets. He averaged almost 17 air yards on his targets, while Sims had by far his worst game of the year on throws more than ten yards past the line of scrimmage. I would imagine those two numbers are related. Things improved significantly when the passing game shifted focus to Carter and Gibbs in the first half before getting Camp and Sanders a little more involved in the second half. Given the strength of Duke’s pass rush, GT’s passing game plan would have been much better if it had initially targeted our more reliable receivers on shorter targets.

GT Pass Defense vs. Duke

Player in Primary Coverage Targets CPOE Allowed
Player in Primary Coverage Targets CPOE Allowed
Thomas 4 32.15%
Carpenter 4 -40.80%
Swilling 3 -0.67%
Showell 2 68.90%
Sims 2 -53.00%
Walton 4 -23.47%
K. Oliver 3 -16.40%
Thomas 1 -41.00%
Johnson 5 -30.37%
Walker 2 17.50%
Kennard 1 25.00%
Jackson 1 33.00%
Curry 7 1.91%
Allen 3 -20.20%
Domineck 1 -65.00%
*CPOE is completion percentage over expected, based on the depth of the target. Negative is better for defensive players.
  • Tariq Carpenter had by far his best game of the year; on four targets, he allowed 41% below expectation, while providing great run support.
  • Juanyeh Thomas continued his late-season struggle.
  • Kenan Johnson was picked on but repeatedly succeeded.
  • David Curry received the most targets, primarily in the short to medium passing game, and held up decently.
  • Zamari Walton continued a strong season on the outside.

Let’s turn now to the full advanced box score to see where GT got the best of the Blue Devils.

Advanced Stats Comparison

GT Duke Advanced Box Score

Adv Box Score GT Duke National Avg
Adv Box Score GT Duke National Avg
Snap Count 68 79 71.5
# Pass Plays Called 28 49 31
Avg Starting FP 67.25 69.2 70.5
YPP 7.19 4.09 5.7
YPA (incl. sacks, scrambles) 6.04 5.98 7.39
% of Passes on 1st Down 39% 51% 40.17%
% of runs on 2nd and long 73% 23% 39.80%
Avg EPA/play 0.13 -0.28 -0.01
Avg EPA/pass -0.1 -0.05 0
Total EPA 8.81 -22.13 -0.96
Avg Air Yards / Completion 4.77 6.43 6.14
Air Yards / Attempt 11.7 13.43 8.89
CP 56.52% 50.00% 62.54%
CPOE -1.88% -6.01% 0.00%
Total Line Yards 169.5 101.5 2.55
Opportunity Rate 58.97% 32.14% 42.42%
Power Success Rate 100.00% 50.00% 68.60%
Stuff Rate (Offense) 20.51% 26.67% 19.17%
Havoc Rate 27% 21% 21.00%
Pressure Rate 16% 32% 30.70%

In Georgia Tech’s most dominant win of the season, where was the game won?

  • Explosive Runs: GT was successful on 49% of its called runs and turned that into an elite 8 yards per carry. Gibbs, Mason, Smith, and Sims combined for rushing gains of 61, 48, 36, 28, 26, and 22, not to mention several more gains in the 10-14 yard range. GT consistently gashed Duke at the second level in all four quarters.
  • Managing Pressure: Duke got pressure on 32% of Jeff Sims’s QB drop backs but managed only a 21% havoc rate. Sims threw one interception and was sacked twice, but overall he managed pressure very well, avoiding rushers and tucking the ball in smart ways. He avoided the devastating plays that might have happened earlier in the year facing that kind of pressure.
  • Preventing Explosives: While the offense had a season-best explosiveness performance, Tech’s defense held Duke to only two runs of ten or more yards and three passes of twenty or more yards. Things weren’t perfect, but GT didn’t have the massive coverage lapses that had plagued the team against the weaker offenses earlier in the season.
  • Surprising Disruption: Perhaps the most encouraging development of the game was the defense recording season highs in both havoc rate and run stuff rate. I’ve been harping on these metrics since the spring as two important indicators of defensive development. As some of the young guys have increased their roles along the defensive front, we’re seeing hints of progress. I want to particularly highlight Jordan Domineck, who recorded two sacks, an additional pressure, a half tackle for loss, a pass defended on a zone blitz, and of course, a forced fumble that he recovered for a touchdown. He has been the disruptive pass rushing force I was hoping for coming into the season.

EPA Highlights

EPA calculates the expected number of points added (or lost in the case of a negative number) on a particular play based on the down and the location on they field. Akshay Easwaran did a great job priming us for EPA this summer. Check out his column here for background.

The EPA totals for this game leave us with a 31 point projected win for GT during the non garbage time portions of the game. What makes the difference between that assessment of the game and the actual 23 point margin? Two plays: the fumbled punt that gave Duke a touchdown and the kickoff downed by GT at its own 4, which directly led to a safety. Looking purely at the offense and defense, GT was more than 30 points better than Duke on Saturday night; GT’s previous best in this metric was the 16 point EPA margin over Louisville. Now, we’ll take a look at the most helpful and hurtful plays for GT from this game, and we actually have a full slate of helpful plays to examine this week!

Most Helpful Plays

  1. Tariq Carpenter’s interception following David Curry’s tipped pass, turning a Duke 1st and 10 at the GT 47 into a GT 1st and 20 at the Duke 20. 7.07 EPA.
  2. Duke’s fumbled shotgun snap, recovered by GT 31 yards behind the original line of scrimmage. 6.06 EPA.
  3. Antwan Owens’s fumble recovery on a Duke 1st and 20 from the GT 49, giving GT the ball at the 50. 5.55 EPA.
  4. Jordan Domineck’s strip sack and recovery for a touchdown from the Duke 2. 5.54 EPA.
  5. Jahmyr Gibbs’s 61 yard touchdown run on first and ten to cap off GT’s opening drive. 4.76 EPA.
  6. Jordan Mason’s 48 yard run on 3rd and 3 to the Duke 20. 4.10 EPA.

This week, our top five becomes six, as GT produced impactful turnovers and explosive offensive plays. This is more fun than the last three games.

Most Hurtful Plays

  1. (The McCollum fumbled punt would be number one, but we only include offensive and defensive plays)
  2. Jeff Sims’s interception late in the second quarter, returned to the GT 17. -6.32 EPA.
  3. Dontae Smith’s fumbled handoff exchange on 2nd and 10 from the Duke 24 in the third quarter. -5.01 EPA.
  4. Chase Brice’s 30 yard touchdown pass on 3rd and 9 in the middle of the 2nd quarter. -4.05 EPA.
  5. Immediately following that Duke passing touchdown and the Dontae Smith special teams error, Jordan Mason was swarmed and tackled for a safety. -3.73 EPA.

GT had three costly turnovers, which was somehow much better than Duke. As a testament to how the defense limited explosiveness, only one of the most hurtful plays was allowed by the defense.

Tracking Season Goals

*I set these goals for the 2020 season in some of my offseason preview work. We will be tracking them as we go this year.

GT Duke Season Goals

Metric Season Goal This Week Season Long
Metric Season Goal This Week Season Long
Offensive Passing Success Rate >40% 39% 41%
Offensive Power Success Rate >70% 100 74%
Defensive Power Success Rate <70% 50% 83%
Defensive Stuff Rate >20% 27 18%
Defensive Havoc Rate >21% 27% 15%

As you can see above, Georgia Tech hit four of its five goals during this game and missed on passing success rate by only one percentage point. The defensive showing, especially considering how many young players were deployed up front, is especially encouraging. We continue to be barely ahead of our target on the two offensive goals, while the three defensive goals improved slightly this week but are still well off target.

Takeaways

  1. This was Georgia Tech’s most dominant performance of the season. That is something to celebrate; Duke ended up as a one point betting favorite before the game, and GT played like a team who was about 30 points better than Duke.
  2. Jahmyr Gibbs is a generational talent. In less than one half, he had 72 rushing highlight yards and 49 yards after the catch, meaning he was almost solely responsible for those yards. In the short term, we will need to monitor his leg health; in the long term, a decent offensive line could unleash him in ways that garner him conference player of the year type recognition.
  3. The defense has a few promising young disrupters. Jordan Domineck had the monster game described above, Jared Ivey had two pressures, and Akelo Stone added a late sack. We need to continue to play as many young guys as possible in the final three games.

With four weeks in between games, the Georgia Tech coaching staff faced a challenge to keep the team engaged after a difficult stretch. They also had several weeks of practice time to innovate. The efforts were largely successful. Despite a strange early game passing plan, Coach Patenaude deployed his running backs and his versatile quarterback in effective ways. Coach Thacker found ways to get up disruption with a young front that has once again been depleted by injuries. This was a game that Georgia Tech had to win to demonstrate progress to close out the season. I’m proud of the staff and the players for showing up, and now we have a chance to improve against three tougher opponents to finish out a strange season.