Georgia Tech Football: Advanced Stats Review Part 4 - The Hidden Factors

Last year, the impact of these “Hidden Factors” was not so hidden. As we detailed here, penalties, turnovers, and special teams (mostly the latter two) cost GT 1.74 expected wins in 2020. These areas don’t show up in some of the key metrics we use in evaluating offense and defense, like success rate, yards per play, and total offensive and defensive EPA (estimated points per play), so we are looking at them separately once again to figure out what kind of impact they had during the 2021 campaign.

I won’t bury the lede. It wasn’t much.

As so often happens, each of these areas regressed back towards a mean of zero, and cumulatively, they had very little impact on GT’s overall season performance.

Penalties

For each hidden factor, we will show the game by game performance and then talk about the total season impact in that area. (One aside - you won’t see the NIU game included here, as ESPN never posted the play by play data for that game). Last year, the season ended with an uproar about penalties, but this season’s performance was actually worse. After losing 5.4 Expected Points on penalties over the course of 2020, Georgia Tech lost 12 EPA in 2021. Still, this is a relatively insignificant number over the course of the season. The offseason emphasis that we heard about removing penalties didn’t have an observed effect, but these penalties amounted to losing only about 0.33 wins over the course of the season.

Special Teams

In our offseason analysis, we’ve yet to identify an area of significant improvement. That changes today! Georgia Tech’s special teams were far more effective in 2021 than in 2020. The Kickoff Unit improved dramatically, mostly driven by the high increase in touchback rate courtesy of Gavin Stewart. The field goal unit wasn’t great, but it wasn’t the worst-in-country unit of 2019 and 2020. The punt team was worse, as expected with the departure of Harvin, and the punt return team had very little impact.

There was some improvement here for sure, but these numbers also remind us just how random special teams can be. The measurable positive impact of STs for the season can be boiled down to a few things: the Gibbs kickoff return TD against BC, Duke’s missed field goals, and two onside kick recoveries at UVA. Those things simply can’t be counted on to repeat themselves in future years, but Georgia Tech fans can at least be glad to have found some baseline competence on a few more of the ST units.

Last year, in 10 games, GT lost 39.6 Expected Points on Special Teams. 2021 saw a net positive of 9.1 EPA, which translates to gaining roughly 1/4 of a win over the course of the year.

Turnovers

GT didn’t benefit from turnovers for the 2021 season, but things were a lot more under control than last year. In 2020, GT lost 29.5 Expected Points over the course of the season, which is equivalent to about 0.71 wins. We observed last year that after the Syracuse game, GT gained expected points on turnover plays for the rest of the season. Sims clearly improved his ball security over the second half of the season, and despite some frustrating plays, especially against Pittsburgh, that improvement did carry over into 2021.

The offense greatly improved its ball security, but the defense created almost no turnover impact after the UNC game. For the season, GT lost a modest 8.3 points on turnover plays. You can see the massive impact of the two defensive touchdowns surrendered to Notre Dame. Overall, the net turnover EPA was driven more by the defense’s inability to force turnovers for most of the second half of the season. Turnovers cost GT about 1/4 of a win, cancelling out the slightly positive impact from special teams.

The Big Picture

Before we wrap up, I wanted to zoom out and look at these same hidden factors at a national scale. The graph shows each team’s final rating in The Binion Index, which we use as a proxy for overall team quality, as well as the total EPA impact from penalties, special teams, and turnovers. You can think of the quadrants as follows: top right is good and lucky, bottom right is good and unlucky, top left is bad and lucky, and bottom left is bad and unlucky.

Of course, there is some skill involved in each of these factors, but as we have seen before, they tend to have lots of variance year to year, and we generally expect the impact to regress towards 0. Georgia Tech from 2020 to 2021 is a classic illustration of this. You can see GT just below and to the left of the center point here, indicating a relatively bad team that got slightly unlucky over the course of the year. Georgia Tech won 3 games and was probably about a 3.5 win team in underlying quality. That’s not a huge deviation.

On the other hand, take a look at Iowa and Florida. Florida was a good team that lost almost 3 games because of our hidden factors, while Iowa was a slightly above average team who won 3 additional games because of these factors. Those are enormous swings and provide us some evidence for expecting Iowa to win fewer games in 2022 and Florida to win more.

Conclusions

1. Last year, we offered this assessment looking towards 2021: The special teams and turnover performances are almost certain to progress towards zero. That means GT could get almost a two win boost just by playing at the same quality on a per play basis on offense and defense. We were right about the progression towards zero, but there was no win boost because the team did not play as well on a per play basis on offense and defense. Last year, Georgia Tech was a 3 win team that played almost as well as a 5 win team. This year, Georgia Tech was a 3 win team playing like a 3 win team. Penalties, Special Teams, and Turnovers cost Georgia Tech only about a third of a win in 2021.
2. People were up in arms at the coaching staff last year for the grossly poor performance in these areas, but 2021 demonstrated that these improvements are not what will drive Georgia Tech back to being a legitimate football program. As Richard Johnson, a favorite for many of us at FTRS, says, “The game is about blocking and tackling.” Special teams and turnovers can swing a game in a very visible way, but sustained success largely comes down to getting better at basic offensive and defensive football. The direction of the coaching changes this offseason provide a small indication that the staff gets that.
3. Overall, this analysis made me feel worse about the 2021 performance, because these largely random factors can’t be blamed for the poor showing of the season. The underlying quality of the football team must improve by two or three standard deviations in 2022, and that’s a tall order for any staff.