Advanced Stats Season Review Part 2: Offense

In the first part of our offseason review, we looked at the Big Picture. I came away encouraged at the leap that GT took statistically from 2019 to 2020, especially as we looked at where the team’s overall performance in key advanced stats placed us nationally. Tech rose from the bottom of the P5 to close to the middle of the pack in the very predictive metrics of net Success Rate, Yards Per Play, and EPA/play.

This week, we’re going to dive more deeply into the offensive performance. What can the advanced numbers tell us about what happened in 2020 and where the focus needs to be in preparing for the upcoming year?

Looking Back

The Georgia Tech rushing attack was led by the breakout performances of Jahmyr Gibbs and Jeff Sims and received effective complimentary play from Jordan Mason, Jamious Griffin, and Dontae Smith. On the ground, Tech posted a 48% Success Rate, which is well above the national average of 42% on rushing plays. This was a huge jump up from 2019’s 39% and helped provide a baseline of offensive efficiency and stability. To get a better understanding of the overall impact of this improvement in rushing, let’s take a look at how this efficiency on the ground tracked with Tech’s overall performance.

This result is startling. For Georgia Tech in 2020, rushing success rate had essentially no effect on the total Expected Points Added outcome in the football game. The running game provided a floor that kept the offense from hitting the depths of 2019, but it did not propel the team towards higher overall performance. On the other hand, what if we take the same look at the passing game? In 2020, GT hit the national average of a 41% Success Rate on called passing plays, which was up from the paltry 28% the year before. This did have an effect on winning football games!

Georgia Tech’s offensive passing success rate explained 47% of the overall variance in total EPA outcome. That’s massive! Out of the four components of offensive passing and rushing and defensive passing and rushing, offensive passing success explained half of the game’s outcome for GT in 2020. That tells us something significant about the importance of focusing on continued improvement in the passing game. A friend ran these numbers for all of FBS, and they showed a similar importance in the passing game (R² of 0.43). The rushing success rate had an R² of 0.29, which is much higher than it was for GT but still significantly lower than the passing success rate’s effect on total EPA outcome.

In short: passing is everything. We need to recruit, develop, and deploy our players in light of that maxim.

Areas for Growth in the Passing Game

Now, let’s take a look under the hood to see what is keeping the passing game from excelling and what that tells us about the needed focus for 2021. There’s essentially four components of the offensive passing game: protection, route runners, the quarterback, and the play calling.

I won’t spoil the surprise by telling you that Georgia Tech’s pass protection was very, very bad in 2020, and that hurt the passing game throughout the season. I want to highlight some work put out this week by Steven Plaisance, which helps to quantify the three player-based components of the passing game. First, his data shows how poor the protection was in 2021.

You can gauge performance by the distance underneath the trend line, and GT is noticeably better only than Kansas, who basically doesn’t have a football team. GT had the 4th worst pressure rate allowed of anyone in the P5 this season on passing plays. We can get a glimpse of how that affected passing performance by looking at the effect of pressure rate on EPA/pass on a game by game basis. The sample size is small, as I didn’t begin tracking pressures until the 4th game of the season, but the trend is still clear: more pressure lowers passing effectiveness.

Takeaway #1:

Improve offensive line play in 2021. I’m compelled by an idea I heard recently from Eric Eager. He identifies some positions in football as “And” positions where you need strength across the board to be effective, while some positions are “Or” positions where a singular star can make the difference. Offensive line is an “And” position. Weak links sink the ship. Georgia Tech’s offensive line needs improvement across the board. One weak link can create chaos. There should be positive change on the line in 2021: two transfers in Devin Cochran and Nick Pendley who could shore up the left side of the line, Jordan Williams taking big steps forward after a true freshman year of starting. The big question will be whether C and RG still perform like weak links that keep the OL from taking the steps forward that we need to grow in the all important passing game.

Next, what about the pass catchers? Once again, Plaisance’s work is helpful but tells a discouraging story. You can find GT in the bad quadrant once again when comparing the catch rate and number of targets per route run. Putting these two metrics together captures how effectively receiving units can get open and then catch the ball. GT comes in below average in both.

Takeaway #2:

Improve receiver play in 2021. According to Eager, WR is an “Or” position. One guy can make a lot of difference. Ricky Jeune played a version of this role from 2015-2017, but GT really hasn’t had that guy since DeAndre Smelter and Darren Waller in 2014. The passing game can’t fully takeoff until we have him. Perhaps it is James BlackStrain, Leo Blackburn, or one of the true freshmen from this year who missed most of the season with injury. If not, GT will continue to struggle to move beyond mediocre in the passing game.

Next, what about the quarterback himself? Jeff Sims did an admirable job being thrust into the starting lineup as a true freshman without a normal offseason of preparation. But how did he compare nationally in overall performance? Plaisance’s numbers help us see how a quarterback performed in relation to the supporting cast. In other words, given the strength of a team’s offensive line and receiving corps as measured above, did the QB perform better or worse than expected?

Unfortunately for GT, Jeff Sims falls below the line (see what I did there!). He underperformed in Completion Percentage above Expectation and EPA/pass even considering the weakness of the offensive line and receiving corps. Here’s another way to see that in action.

This chart looks at the relationship on a game by game basis of the average depth of target for passes thrown by Jeff Sims and the relation to Completion Percentage over Expectation. We don’t have this data publicly available for all college football quarterbacks, but we do have it in the NFL. The chart below shows the relation between ADOT and CPOE in the NFL:

As you can see, higher depth of target is correlated with improved CPOE in the NFL. For Jeff Sims, his CPOE goes down when his average depth of target goes up. The same relationship holds when we look at the effect of ADOT for Sims’s passing success rate and EPA/pass.

Takeaway #3:

We need better quarterback play. It’s exciting to start a true freshman. Jeff Sims has encouraging tools in his arm strength and elite athleticism. But he’s not a good quarterback yet. As you might imagine, in Eager’s framework, QB is the ultimate example of an “Or” position. You need one guy, and he can change everything. I hope we get to see Sims become that guy in 2021. If not, the door stays open for Chayden Peery, or perhaps a 2022 recruit to try to show that they can be the guy.

There’s one more element to the passing game that we haven’t yet fully explored: playcalling. Especially when starting a true freshman at quarterback, the staff needs to be very mindful about setting up situations where he can be effective throwing the ball. Tough situations where the defense knows you have to pass tend to produce bad outcomes. One way we can measure that is by looking at performance on obvious passing downs. Bill Connelly has defined a passing down as 2nd and 8 or longer, as well as 3rd or 4th and 5 or longer. On those plays in 2020, Georgia Tech posted an EPA/pass of -0.186, to go with an EPA/rush of -0.026. Nationally, those numbers for EPA/pass 0.00 and EPA/rush of -0.04. When GT needed to throw and the opponent knew it in 2020, it didn’t go well. Let’s look at this from one more angle.

The chart shows how GT fared in EPA/pass on first down compared to second down throughout the season. On average, GT was 0.16 EPA/pass better on called pass plays on 1st down compared to 2nd down.

Takeaway #4:

For CDP, call more pass plays on 1st down to give Jeff Sims the advantage of throwing when the defense expects run. This keeps pass rushers from pinning their ears back and often means one more player in the box instead of being in coverage. Help the young QB out.

Conclusion

For Georgia Tech to really excel in 2021 and going forward, it’s going to come down to improving the passing game. We’ve provided 4 ways to move towards that goal in 2021: eliminate offensive line weak links, find a go to receiver, improve quarterback play, and call more first down passes. Like we said, it comes down to recruitment, development, and deployment. Easy enough, right?