clock menu more-arrow no yes

Filed under:

Advanced Stats Preview: GT vs. Duke

GT hasn’t played in a month. Our advanced stats model has a new way of projecting point spreads. What could possibly go wrong?

Georgia Tech v Duke Photo by Grant Halverson/Getty Images

Thanksgiving is tomorrow. It’s rivalry week. The table is set…for GT-Duke. From what I could find, this will be the first time since 1930 that GT will play someone other than UGA on the Saturday after Thanksgiving. It feels weird. But everything is weird this year.

Heading into the final four (scheduled) games of the season, this is GT’s best chance for a win. It pains me to say that Duke has won 5 of the past 6 games in this series. Seemingly every recent year, GT fans expect to win this game and find themselves gut punched by an underwhelming performance. This year, Vegas has shrugged its shoulders and labeled the game a Pick ‘Em. Optimistically, SP+ pegs GT as a 5 point favorite, giving us a 62% win probability. This week, I will be introducing a new method I am using to project the point spread for each game, which you can find in the final section. First, let’s look at the stats matchup and relative advantages each team should expect to have:

GT Duke Advanced Stats Preview

Metric GT 2020 Duke 2020 Advantage National Average
Metric GT 2020 Duke 2020 Advantage National Average
Offensive Success Rate 45% 38% GT by 7% 42%
Offensive YPP 5.12 5.1 GT by 0.02 ypp 5.7
Offensive EPA/Play -0.09 -0.11 GT by 0.02 EPA/play -0.01
Offensive YPA (including sacks and scrambles) 6.58 6.5 GT by 0.1 ypa 7.4
Offensive 3rd Down Success 36% 41% Duke by 5% 42%
Offensive Stuff Rate Allowed 26% 0.19 Duke by 7% 19%
Defensive Success Rate 46% 40% Duke by 6% 42%
Defensive YPP 5.7 5.5 Duke by 0.2 ypp 5.7
Defensive EPA/Play 0.15 0.03 Duke by 0.12 EPA/play -0.01
Defensive YPA 7.8 7.5 Duke by 0.3 ypa 7.4
Defensive 3rd Down Success 49% 40% Duke by 9% 42%
Defensive Stuff Rate 17% 15% GT by 2% 19%
Defensive Havoc Rate 14% 18% Duke by 4% 21%
*GT numbers come from my play by play charting. Opponent numbers come from @CFB_Data and cfb-graphs.com

Across the board, the match looks relatively even, as Duke holds an advantage in 8 of the categories we track, while GT has five categories of advantage. GT has been slightly more efficient on offense; Duke has been slightly more efficient on defense and given up far fewer explosive plays than GT has.

When GT Has the Ball

GT’s slightly below average offense will try to sustain drives against Duke’s slightly above average defense. One thing to watch is the performance of Duke’s defensive ends, Chris Rumph and Victor Dimukeje; they are both excellent pass rushers who struggle to defend the run. Duke is not elite at generating havoc or at stuffing runs, which bodes well for a GT offensive line that does not do well against those kinds of defenses. GT’s coaching staff has been generating lots of excitement about Jordan Mason’s improved health entering this game, and we may well see the return of lots of 2 back personnel with Mason and Jahmyr Gibbs. This could generate good advantage for GT in the run game and in the quick throw passing game. GT will need to be careful about getting into obvious passing situations and should avoid attempting a lot of deep drops that could give opportunity to Duke’s strong pass rush.

Let’s take a look at how GT has performed against the relative strength of previous defenses and see what that might tell us about what to expect against Duke.

GT Offensive Projection vs. Duke

Team Defensive Season Long Efficiency Score Defensive EPA/Play GT Offensive Efficiency Score GT Offensive EPA/Play
Team Defensive Season Long Efficiency Score Defensive EPA/Play GT Offensive Efficiency Score GT Offensive EPA/Play
FSU 3.33 0.206 2.95 0
UCF 2.67 -0.025 2.59 -0.07
Syracuse 2.41 0.026 2.42 -0.06
Louisville 2.2 0.065 3.45 0.43
Clemson 1.46 -0.167 0.47 -0.78
Boston College 2.28 0.037 2.12 -0.23
Notre Dame 1.52 -0.17 1.1 -0.35
GT Previous Opponent Average 2.27 0 2.16 -0.09
National average 2.39 0 2.39 0
Duke 2.2 0.026 2.01 -0.12
*Efficiency Score is Success Rate Allowed * Yards Per Play Allowed (From a GT offensive perspective, higher is better)
*GT’s offensive efficiency score and offensive EPA/play are projected based on opponent-adjusted performance in the previous games compared to Duke’s season long statistics.

The matchup looks like a slight advantage for the Duke defense, and I would expect a GT offensive output in the low 20s.

When Pitt Has the Ball

This is a subpar offense that GT should be able to hold in check. Unfortunately, things haven’t always gone well when I have written those words. Statistically, the Duke offense looks pretty similar to that of FSU in season long metrics, and that game produced the best GT defensive performance of the year. However, in the last three games, GT has given up Efficiency Score and EPA/play results that exceeded the season average for the opposing offenses. This is the trend that has GT fans most worried right now. Again, this is a match up that presents a great opportunity to reverse that trend. Duke’s offensive efficiency has been ok, but Chase Brice has been a turnover machine. In fact, they do not have a turnover free game yet this year. Using previous opponent adjusted performance to guide us, GT should be able to hold Duke to a slightly below average performance.

GT Defensive Projection vs. Duke

Team Offensive Efficiency Score Offensive EPA/Play GT Defensive Efficiency Score (lower is better) GT Defensive EPA/play
Team Offensive Efficiency Score Offensive EPA/Play GT Defensive Efficiency Score (lower is better) GT Defensive EPA/play
FSU 1.93 -0.12 1.56 -0.11
UCF 3.36 0.21 3.12 0.23
Syracuse 1.31 -0.18 1.99 0.1
Louisville 2.56 0.11 2.29 0.1
Clemson 2.9 0.171 5.34 0.46
Boston College 2.14 0.05 2.67 0.28
Notre Dame 2.68 0.13 2.95 0.13
GT Previous Opponent Average 2.37 0.05 2.85 0.15
National average 2.39 0 2.39 0
Duke Projection 1.938 -0.111 2.31 0.04
*Efficiency Score is Success Rate Allowed * Yards Per Play Allowed (From a GT defensive perspective, lower is better)
*GT’s defensive efficiency score and defensive EPA/play are projected based on opponent-adjusted performance in the previous games compared to Duke’s season long statistics.

Here’s the takeaway: These are two offenses that are pretty good at moving the ball but have been devastated by turnovers. We all know that it’s important to win the turnover battle. If that number is relatively even, Duke looks to have the slight advantage.

Prediction

This is the first game this year where I have used my own GT numbers, our opponent-adjusted performance from earlier in the season, and our current opponent’s season long statistics to generate my own point spread for the game. We are using the same formula incorporating projected success rates, yards per play, and EPA margin that generates our postgame win expectancies that you find in the Advanced Stats Review articles. We mentioned earlier that Vegas has this game as a Pick ‘Em, and SP+ has GT as almost a 5 point favorite. Using a 2.5 point adjustment for home field advantage, my numbers favor Duke by 5.

How did that happen? Duke’s defensive advantage against the GT offense is stronger than the advantage GT’s defense should have against Duke’s offense. As we begin this stretch run, I was hoping to see my spreadsheet spit out a confident prediction in GT’s favor. Unfortunately, that didn’t happen. My hope is that the coaching staff and team really did use this unforeseen month off to generate some significant personnel and schematic improvement. We will find out Saturday night under the lights.

27-21 Duke.