Today in Analytics...

roadkill

Helluva Engineer
Messages
1,081
"We played VT, and I think we're both lower after playing each other than before--guess it's not a zero-sum game."

I get that the offense showed improvement, which is obvious. But for our defensive ranking, this doesn't make sense to me. It's saying that despite the prior week's massive beatdown at FSU, our D ranked even worse after holding VT to only 2 scores. I also understand the concept of opponent-adjusted, but I have to wonder about how the calculations are done. Perhaps the pick-6 and punt return were counted against the D?
 

slugboy

Moderator
Staff member
Messages
10,748
"We played VT, and I think we're both lower after playing each other than before--guess it's not a zero-sum game."

I get that the offense showed improvement, which is obvious. But for our defensive ranking, this doesn't make sense to me. It's saying that despite the prior week's massive beatdown at FSU, our D ranked even worse after holding VT to only 2 scores. I also understand the concept of opponent-adjusted, but I have to wonder about how the calculations are done. Perhaps the pick-6 and punt return were counted against the D?
I'd have to look at the Composite after FSU. Most of our move could have happened then. It just looks like our defense isn't as far to the right as it used to be.
Also, there are 2nd and 3rd order factors--like Clemson's thrashing by ND moved them down and moved us down too.
 

leatherneckjacket

Helluva Engineer
Messages
1,764
Location
Atlanta, GA
"We played VT, and I think we're both lower after playing each other than before--guess it's not a zero-sum game."

I get that the offense showed improvement, which is obvious. But for our defensive ranking, this doesn't make sense to me. It's saying that despite the prior week's massive beatdown at FSU, our D ranked even worse after holding VT to only 2 scores. I also understand the concept of opponent-adjusted, but I have to wonder about how the calculations are done. Perhaps the pick-6 and punt return were counted against the D?
VPISU's offense is really, really bad, but they actually performed better against us with regards to EPA than against other defenses. So, our defense went down. Our offense is still really bad, so no movement there.
 

roadkill

Helluva Engineer
Messages
1,081
I'd have to look at the Composite after FSU. Most of our move could have happened then. It just looks like our defense isn't as far to the right as it used to be.
Also, there are 2nd and 3rd order factors--like Clemson's thrashing by ND moved them down and moved us down too.
VPISU's offense is really, really bad, but they actually performed better against us with regards to EPA than against other defenses. So, our defense went down. Our offense is still really bad, so no movement there.
The EPA must be the key. It can seem counterintuitive though. And the especially magnitude of the change. Eyeballing the scatter charts (whose scales don't always make sense as they repeat the same values on their axes), our D went from about -0.11 (good) to about -.03 (a bit above average) after a single game.
 

ibeattetris

Helluva Engineer
Messages
3,545
FEI Update:

DFEI #39 🔥
OFEI #111 🤮

I believe that's the highest DFEI ranking GT has had since...? Absolute insanity how much the defense has improved.

Now if only we had an offense to match...
If we end at 39, it would be the highest ranked DFEI we've had since the metric began in 2007.
1669134951021.png


It will also be far and away the worst offense we've had.
 

slugboy

Moderator
Staff member
Messages
10,748
If we end at 39, it would be the highest ranked DFEI we've had since the metric began in 2007.
View attachment 13566

It will also be far and away the worst offense we've had.
Ugggggggh, last year 🤯

The composite was already out:



Overall, not good. It's like a C- grade. We have some gruesome performances and some great ones. We're a Jekyll and Hyde team.

1669135252053.png


The ACC is the weak sister of the P5 this year. For all the smack we talked pre-season about the B12 being dead, they're good from top to bottom.
1669135377068.png
 

JacketOff

Helluva Engineer
Messages
2,818
That 2007 team’s defense started out so promising with that ND game, too… I wouldn’t have guessed 2008’s was above it.

The 17/18 rankings are also interesting.
By the numbers, 17 should’ve been one of PJ’s best years. Better on both sides of the ball than in 16, and had a better offense with basically the same defense as in 08.

If Taquon pitches that ball against Tennessee that year could’ve looked a lot different. Weird how it played out, should’ve beat Miami too. And had UCF come to Atlanta to play, their run of being the best G5 school may have never happened.
 

TheDanielHordShow

Helluva Engineer
Messages
1,224
Location
Atlanta
By the numbers, 17 should’ve been one of PJ’s best years. Better on both sides of the ball than in 16, and had a better offense with basically the same defense as in 08.

If Taquon pitches that ball against Tennessee that year could’ve looked a lot different. Weird how it played out, should’ve beat Miami too. And had UCF come to Atlanta to play, their run of being the best G5 school may have never happened.
I think we were 5-1 at home, the one loss being to Georgia, and 0-6 on the road. That team absolutely should’ve made a bowl and were one or two plays away from it happening.
 

stylee

Ramblin' Wreck
Featured Member
Messages
668
By the numbers, 17 should’ve been one of PJ’s best years. Better on both sides of the ball than in 16, and had a better offense with basically the same defense as in 08.

If Taquon pitches that ball against Tennessee that year could’ve looked a lot different. Weird how it played out, should’ve beat Miami too. And had UCF come to Atlanta to play, their run of being the best G5 school may have never happened.

Yeah, I noted the weirdness of 2017-2018 in a post a few weeks ago. The fact that our offense was top 25 in 2017 and top 20 in 2018 seems nutso on the face of it —— TM was the least capable passing QB in the CPJ era other than Oliver. We had Jeune in 2017, which helped, but the overall lack of a passing game should have doomed us.

However, TM was a fantastic, tough runner and we had a real bumper crop of B-Backs during this stretch: Mason, Howard, and Benson in 2017. I especially liked Benson, who at 10.6 in the 100 meters was the first speed burner starting at B-Back since Dwyer (Snoddy never got the footwork down and moved to A-Back).

We also still had Cottrell, Lynch, and Searcy at AB these years. Although Qua and Lynch couldn’t contribute as much in the passing game as they did in 2016, they still were solid offensive weapons.

The o-line was solid for both years: Braun, Bryan, Cooper, Devine in 2017, Lee, Marshall, Stickler, etc.

Overall, the offensive flavor in these years was very much between the tackles, heavily weighted towards BB and QB runs, and without much threat of the vertical pass. Although we hit Brad Stewart late in the 2018 UGA game, we probably would have thrown much more with Tevin, or Justin, or Vad, behind center against that defense. Notably, the 2018 rotation had Stewart and young Malachi Carter and Jalen Camp.

The fact that 2017-2018 were statically superior to a 2008 lineup featuring Bebe, Dwyer, and Nesbit is crazy, of course, but the overall more interesting fact is that we were on an offensive upswing from 2016 and might have continued improving the next year with one of the other QBs battling with Tobias.
 

ibeattetris

Helluva Engineer
Messages
3,545
Maybe deep in the analytics side of things, but here is an AWS blog post about their machine learning model for determining if a team should go for it.
 
Top