Analytics post mortem

[GT33]

Did Syracuse get kicked out of the ACC?

Francine cried them to a conference all by themselves. ******* Crying Coaches. Hope that ***** is back on the schedule next year.

 

[MountainBuzzMan]

We put up those 24 points in 8 drives.
Our ninth drive was 8 play 3:40 and left them no time on the clock.
3 points per drive is currently top 25. If you want to count the final drive that 2.67 which isn’t terrible

We can do that! Perform at a top 25 level but let our eyeballs tell us we are a bottom 25 performance and then complain about the obvious bottom 25 performance.

 

[Bogey]

We can do that! Perform at a top 25 level but let our eyeballs tell us we are a bottom 25 performance and then complain about the obvious bottom 25 performance.

It is in our DNA, you don't design and build buildings, bridges, and air planes without a huge safety factor to guarantee success. In the case of football, the "safety factor" for our victories is winning every statistic and every game by a wide margin. Then we can all be happy and sleep well.

 

[swampsting]

Our “ok” offense scored the most points and gained the most yards against the 5th-ranked defense (ranked after their loss to us) in the ACC. And except for UNC, no other team Duke played came close to the yards we put up. Our special teams shoulder a good portion of the blame for our scoring not keeping pace with our yardage. I think an offense that is statistically in the upper quartile of most objective measurements should be considered “good”.

Now, if you are talking about the team as a whole, I completely agree with you. We are a key play/penalty or 2 from being either 6-0 or 2-4.

Aside from UNC and Hampton, Duke hasn't exactly been tested by high flying offenses. MTSU and Northwestern are 112 and 123 in total offense, respectively. And UConn's stats are bloated by having faced Merrimack, Buffalo and FAU.

 

[roadkill]

Aside from UNC and Hampton, Duke hasn't exactly been tested by high flying offenses. MTSU and Northwestern are 112 and 123 in total offense, respectively. And UConn's stats are bloated by having faced Merrimack, Buffalo and FAU.

True, but I think FEI ratings/rankings are opponent-adjusted. @slugboy can correct me if I'm mistaken about this.

Also, I would add that Duke was tested by a high-flying offense last Saturday.

 

[slugboy]

Aside from UNC and Hampton, Duke hasn't exactly been tested by high flying offenses. MTSU and Northwestern are 112 and 123 in total offense, respectively. And UConn's stats are bloated by having faced Merrimack, Buffalo and FAU.

I thought they looked solid. They tackled well

True, but I think FEI ratings/rankings are opponent-adjusted. @slugboy can correct me if I'm mistaken about this.

Also, I would add that Duke was tested by a high-flying offense last Saturday.

They are opponent adjusted

 

[swampsting]

I thought they looked solid. They tackled well

They are opponent adjusted

Not saying they're not properly coached. But they really hadn't faced a potent offense yet, either, with the exception of Carolina. You could play Iowa (from 2023) every week and have a top 10 defense but are you really any good? You really don't know until you play a good offense.
We won't find out much about Duke's defense in two weeks, either (FSU). But the two weeks after that, SMU and Miami, OK, let's see how good they are then.

 

[slugboy]

Not saying they're not properly coached. But they really hadn't faced a potent offense yet, either, with the exception of Carolina. You could play Iowa (from 2023) every week and have a top 10 defense but are you really any good? You really don't know until you play a good offense.
We won't find out much about Duke's defense in two weeks, either (FSU). But the two weeks after that, SMU and Miami, OK, let's see how good they are then.

I think the story of those games will be about those defenses against Duke's weak offense. After a while, their defense will fail because their offense gets shut out.

 

[roadkill]

Not saying they're not properly coached. But they really hadn't faced a potent offense yet, either, with the exception of Carolina. You could play Iowa (from 2023) every week and have a top 10 defense but are you really any good? You really don't know until you play a good offense.
We won't find out much about Duke's defense in two weeks, either (FSU). But the two weeks after that, SMU and Miami, OK, let's see how good they are then.

I’m unsure how to interpret your post. You seem to be saying Duke hasn’t played a good offense yet, thus we can’t tell if they are a good defense or not. That’s actually what advanced stats like FEI and EPA/play can tell you, since they factor in strength of opponents, and by mid-season tend to have enough data to converge on a realistic view of a team’s strengths and weaknesses.

You are also implying that GT doesn’t have a good offense. I happen to disagree, and so do advanced stats.

A single game, for instance Duke vs. Miami, can provide data but is subject to the “any given Saturday” rule which allows for improbable results (see Bama vs. Vandy). How a team performs in aggregate over the season is usually a better gauge, and is what advanced stats use.

Also, perhaps we aren’t on the same page on what constitutes a “good” offense or defense. I tend to draw the line at the upper quartile of FBS. By this metric, we have a good offense and Duke has a good defense, using objective advanced stats. If your line is good = top ten, then we are going to simply disagree regardless of metric.

 

[cpf2001]

I’m skeptical of the accuracy of adjusting for opponent across “classes” per-se. Can shutting down a bunch of bottom-half offenses tell you THAT much about how you’d do against a top-decile offense? Or are there more step-function-like levels where two defenses could do similarly against a meh offense but one actually is much better at tackling, or faster, or whatever, in a way that’s revealed only when they play top teams?

I think generally opponent-adjustments are pretty good by mid season but in the case of Duke when they hadn’t really played ANYONE I think it’s a specific situation where the verdict is still out.

 

[roadkill]

I’m skeptical of the accuracy of adjusting for opponent across “classes” per-se. Can shutting down a bunch of bottom-half offenses tell you THAT much about how you’d do against a top-decile offense? Or are there more step-function-like levels where two defenses could do similarly against a meh offense but one actually is much better at tackling, or faster, or whatever, in a way that’s revealed only when they play top teams?

I think generally opponent-adjustments are pretty good by mid season but in the case of Duke when they hadn’t really played ANYONE I think it’s a specific situation where the verdict is still out.

Agree that opponent-adjusted stats are likely not that accurate in some extreme cases. And Duke has played a relatively weak schedule. But as of today they have played three P4 opponents and only one scored more than 21 points. And that's with a weak offense that tends to keep their defense on the field.

I gagged a little bit when I realized I was defending Duke's team, sorry.

 

[cpf2001]

Agree that opponent-adjusted stats are likely not that accurate in some extreme cases. And Duke has played a relatively weak schedule. But as of today they have played three P4 opponents and only one scored more than 21 points. And that's with a weak offense that tends to keep their defense on the field.

I gagged a little bit when I realized I was defending Duke's team, sorry.

As of today Duke's played at least one team with a pretty good offense, compared to before last Saturday.

 

[ibeattetris]

True, but I think FEI ratings/rankings are opponent-adjusted. @slugboy can correct me if I'm mistaken about this.

Also, I would add that Duke was tested by a high-flying offense last Saturday.

Duke is currently ranked 33 in DFEI.

OOA is "opponent offenses faced" and that is rank 101. So really, Duke has performed well, but their opposition has been really poor. Their non-adjusted value is .43 (22nd) and strength of schedule reduces it to .25 (33rd). I have questioned in recent years whether FEI adjusts for SOS as much as I would, but I haven't looked into it enough to offer anything.

 

[Spalding Jacket]

I would say Duke passed the eye test of a good defense. They weren’t out of position, they pursued well and tackled soundly. A defense that does that is going to be at least average. We just had some faster/ shiftier players who were able to break out for some big gains. Oline also was able to get consistent push. Our offense had to work for every score of that game, no gimmes.

 

[LongforDodd]

Anyone have the PFF grades for our OL vs. Duke?

 

[slugboy]

Anyone have the PFF grades for our OL vs. Duke?

PFF Grades For Every Offensive Player On Georgia Tech In Their Win vs Duke

It was not pretty, but Georgia Tech found a way to end Duke's unbeaten season and get a win to move themselves to 4-2. The running game and, especially the defe

www.si.com

LT Corey Robinson- 73.5 (83)
LG Joe Fusile- 70.4 (83)
RG Keylan Rutledge- 69.5 (83)
RT Jordan Williams- 69.4 (83)
C Weston Franklin- 69.0 (83)
TE Jackson Hawes- 60.0 (2)
TE Ryalnd Goede- 59.9 (75)
TE Avery Boyd- 55.0 (9)
LG Harrison Moore- 51.7 (13)

 

[stinger78]

PFF Grades For Every Offensive Player On Georgia Tech In Their Win vs Duke

It was not pretty, but Georgia Tech found a way to end Duke's unbeaten season and get a win to move themselves to 4-2. The running game and, especially the defe

www.si.com

LT Corey Robinson- 73.5 (83)
LG Joe Fusile- 70.4 (83)
RG Keylan Rutledge- 69.5 (83)
RT Jordan Williams- 69.4 (83)
C Weston Franklin- 69.0 (83)
TE Jackson Hawes- 60.0 (2)
TE Ryalnd Goede- 59.9 (75)
TE Avery Boyd- 55.0 (9)
LG Harrison Moore- 51.7 (13)

60.0 is considered average, so all but one were average or above. That’s an improvement against a pretty good D. I am happy with that.

 

[leatherneckjacket]

60.0 is considered average, so all but one were average or above. That’s an improvement against a pretty good D. I am happy with that.

Also, notice that a certain LT who scored in the 20's against Louisville did not play.

 

[slugboy]

60.0 is considered average, so all but one were average or above. That’s an improvement against a pretty good D. I am happy with that.

For what it's worth, 65 is average. PFF scales their grades to a 100 point scales, with 65 as the average. As a range, you might see 60-70 as an "average" block, 50-60 as below average, below 50 as poor, 70-80 as above average, and above 80 as exceptional.

The defensive grades are at https://www.si.com/college/georgiat...rgia-tech-in-their-win-over-duke-01j9h2vq5w6m.

I'd like to see the plays and the grades that go with them.

 

[yeti92]