UVA is pretty clearly one of the worst teams in the ACC. All the advanced models rate GT at least 37 spots higher than UVA.
|GT Win Probability
(*) Models only predict a point spread
(**) MELLS has a quick explainer at the bottom of this post
Even on the road GT is a solid favorite. Still far from a guaranteed win, but if Georgia Tech takes care of their own issues they should have no problem closing this one out with a victory.
When UVA Has The Ball
- Wow, UVA's offense is not great. They are below average at everything my model measures. They are last in FBS in average starting field position. Interestingly S&P+ actually has their passing game as above average, while I have them rated as 98th in the country.
- Georgia Tech's rushing defense is coming off their best performance of the year after holding Dalvin Cook to under 100 yards rushing last week. This was actually the first game all season that our rushing defense held an opponent under their season average Expected Points Added Per Play:
- Our passing defense is still struggling to impact opposing offenses. We are 85th in the country in passing success rate allowed, meaning opponents have an easy time completing passes consistently against us. And we are 105th in the country in adjusted sack rate. Let's hope we can turn on the pressure against UVA.
When Georgia Tech Has The Ball
- UVA's rushing defense might be their best quality and it's still below average. Our rushing attack is rated 9th in the country according to S&P+ but we still struggle on short yardage situations, ranking 92nd in Power Success Rate. UVA has actually done a decent job on short situations and on getting in the backfield, they rank 27th in the country in stuff rate (holding opponents to no gain or worse).
- Our passing attack is decidedly average but we had been on a decent run lately considering the opponent quality we have faced. Duke, UNC, and Clemson all have passing defenses in the top 10% of the country (the dashed black line in the chart below). FSU was the first game since Duke where our passing offense played below their opponent's average Expected Points Added Per Play Allowed.
** MELLS Explanation: MELLS (Multilevel Expected points - Linear Least Squares) is my predictive model that predicts the number of points scored in a given matchup and the variance of that prediction to generate the entire range of possible points scored by an offense. It does this using a team's opponent adjusted running and passing values for their offense and defense as well as their opponent's values. The opponent adjusted values are estimated by a multilevel (hierarchical) model that tries to separate the contribution of expected points added on each play by the offense and defense. In addition preseason estimates are given the weight of one game, so after 7 games for Georgia Tech the preseason projections are worth 12.5% (well, not exactly but close enough) of our values. I'll have a full explainer up in the offseason at the latest.