What are your measures for success with the new defense?

[slugboy]

IMO, this is spot-on ... tackles for loss, TO's and 3-and-outs.

Problem with 3rd down conversion percentage is that it doesn't measure 3-and-outs--it measures how well you do IF you get them to 3rd down in the first place. If they convert on 1st or 2nd down, they never show up there. Likewise, if they lose a fumble or throw a pick on first down, they don't show up either.

There are some stats that get close to what you want, but there's a flaw in each one. For example, you can look at how often your opponents punt, but that doesn't show turnovers or failed 4th down conversions or end of half. Success rate (one of the columns here) shows a combo of holding the opponent to 3-and-out somehow, plus getting points for doing well on 1st and 2nd down.

 

[Lavoisier]

It's going to be tough to improve a defense where you lost the entire starting secondary and where you are implementing a defense that is a polar opposite to the scheme you have been recruiting to for the past 5 years. I will consider our guys wrapping up when they tackle and being able to take the proper angles a success in year 1 and anything else will be gravy. It will show me the staff knows how to coach guys properly, something I am not convinced Ted Roof knew how to do.

I think it would be unfair to expect anything more this year. Year 2 will be a different story and I will hope to see improvements across the board in negative plays and 3-and-outs.

 

[Longestday]

Points per drive should show what you need...

The others show what you need to work on to improve points per drive.

You want a low score and normalize to the number of drives

 

[vamosjackets]

Someone (@ilovetheoption , @Longestday , @AE 87 , @slugboy , @alentrekin , Bueller?) help me understand DFEI and/or S&P. I like ppd, but the "opponent adjusted" part of those stats make me think they could be the best measure. Are they the best measure, and if so, why - how do they accomplish being the best metric?

I'm thinking for my money, since at this point I don't understand how DFEI and S&P are being calculated, I would compare our defense with our primary opponents by ranking our Defensive PPD (points per drive) versus the other teams in the Coastal (in Coastal games only). That being the case, I'd like to see us ideally in the top 3 defense in the Coastal but would be satisfied this year being in the top half of the Coastal in ppd.

 

[slugboy]

I don't know what's best. DFEI and S&P+ defense both seem to be really good in getting the noise and gunk out (garbage time, downplaying weak opponents, …).
S&P+ had us as the #66 defense last year and DFEI had us as the #63 defense last year. I'm not going to beat myself up over the difference between 63 and 66. They're both middle of the pack, your defense won't be the thing that gets you to a bowl game numbers.
In DFEI (lower is better), we had a rating of 2.25 (#63) and went 4-6. Stanford had a DFEI of 2.24 (#62) and went 9-5. Michigan State had a DFEI of 1.64 (#25). Clemson was #1 at 0.83, UGA was #5 at 0.98, and VT and Miami were both top 15.
I made a chart. When you get around the top 10 or so, defenses don't gradually separate from the pack, they dramatically do. Same thing for the bottom 10 defenses. Between that, it's gradual improvements:

S&P+ is the same thing, just a little smoother. We're right in the middle on both. Not awful, but textbook mediocre:

 

[bobongo]

It's going to be tough to improve a defense where you lost the entire starting secondary and where you are implementing a defense that is a polar opposite to the scheme you have been recruiting to for the past 5 years. I will consider our guys wrapping up when they tackle and being able to take the proper angles a success in year 1 and anything else will be gravy. It will show me the staff knows how to coach guys properly, something I am not convinced Ted Roof knew how to do.

I think it would be unfair to expect anything more this year. Year 2 will be a different story and I will hope to see improvements across the board in negative plays and 3-and-outs.

Based on that it would indeed be unfair to expect anything more, but not unrealistic to hope for a lot better. The new secondary seems talented and hopefully will show better results in a more aggressive style which provides tighter coverage. Also, the guys seem cranked to play in this new, more proactive and disruptive D.

 

[bobongo]

Someone (@ilovetheoption , @Longestday , @AE 87 , @slugboy , @alentrekin , Bueller?) help me understand DFEI and/or S&P. I like ppd, but the "opponent adjusted" part of those stats make me think they could be the best measure. Are they the best measure, and if so, why - how do they accomplish being the best metric?

I'm thinking for my money, since at this point I don't understand how DFEI and S&P are being calculated, I would compare our defense with our primary opponents by ranking our Defensive PPD (points per drive) versus the other teams in the Coastal (in Coastal games only). That being the case, I'd like to see us ideally in the top 3 defense in the Coastal but would be satisfied this year being in the top half of the Coastal in ppd.

Points per drive is only part of the story. How long did those drives take? That's the problem with the bend-but-don't break style - it keeps the ball control offense off the field. I would gladly take the same PPD this year with shorter drives in plays per drive. The results would hopefully be manifested on the offensive side of the scoreboard.

 

[a5ehren]

I'm looking for a better Havoc Rate, mainly. That's TFLs, break-ups, turnovers, etc. We were awful at this under Roof (literally DFL last year, I believe) - and our offense seems to be pretty well geared for it.

 

[ilovetheoption]

I don't know what's best. DFEI and S&P+ defense both seem to be really good in getting the noise and gunk out (garbage time, downplaying weak opponents, …).
S&P+ had us as the #66 defense last year and DFEI had us as the #63 defense last year. I'm not going to beat myself up over the difference between 63 and 66. They're both middle of the pack, your defense won't be the thing that gets you to a bowl game numbers.
In DFEI (lower is better), we had a rating of 2.25 (#63) and went 4-6. Stanford had a DFEI of 2.24 (#62) and went 9-5. Michigan State had a DFEI of 1.64 (#25). Clemson was #1 at 0.83, UGA was #5 at 0.98, and VT and Miami were both top 15.
I made a chart. When you get around the top 10 or so, defenses don't gradually separate from the pack, they dramatically do. Same thing for the bottom 10 defenses. Between that, it's gradual improvements:
View attachment 3958

S&P+ is the same thing, just a little smoother. We're right in the middle on both. Not awful, but textbook mediocre:
View attachment 3959

GT finished 40th out of 65 P5 teams. That's a little worse than textbook mediocre. TO me, 38th percentile is pretty cruddy.

 

[gtrower]

Points per drive is only part of the story. How long did those drives take? That's the problem with the bend-but-don't break style - it keeps the ball control offense off the field. I would gladly take the same PPD this year with shorter drives in plays per drive. The results would hopefully be manifested on the offensive side of the scoreboard.

Yes. Even if we keep the same ppd-against, if our TOP-against improves (which is pretty much a certainty) then the TOP disparity will start to manifest on the field late in the game. I’m fully expecting us to lead the country in TOP this year and I think 36+ minute ATOP is attainable.

I’ve always wished we could have combined Tenuta’s D with this offense. This should be something along those lines.

 

[slugboy]

GT finished 40th out of 65 P5 teams. That's a little worse than textbook mediocre. TO me, 38th percentile is pretty cruddy.

I was including all the FBS teams, so 63rd to 66th out of 130. But in the dregs of P5 by that standard.

(For those looking at the chart, the horizontal is team defense ranking on the scale and vertical is the rating. We're in the 60's on both charts)

 

[vamosjackets]

Someone (@ilovetheoption , @Longestday , @AE 87 , @slugboy , @alentrekin , Bueller?) help me understand DFEI and/or S&P. I like ppd, but the "opponent adjusted" part of those stats make me think they could be the best measure. Are they the best measure, and if so, why - how do they accomplish being the best metric?

I'm thinking for my money, since at this point I don't understand how DFEI and S&P are being calculated, I would compare our defense with our primary opponents by ranking our Defensive PPD (points per drive) versus the other teams in the Coastal (in Coastal games only). That being the case, I'd like to see us ideally in the top 3 defense in the Coastal but would be satisfied this year being in the top half of the Coastal in ppd.

Just realized my last statement was dumb. There's only 7 teams in the Coastal, so being top half is being top 3. I guess I would be satisfied being in the top 4 of the coastal this year in defense but would like to see us as top 2 in the coastal long term.

 

[AE 87]

Someone (@ilovetheoption , @Longestday , @AE 87 , @slugboy , @alentrekin , Bueller?) help me understand DFEI and/or S&P. I like ppd, but the "opponent adjusted" part of those stats make me think they could be the best measure. Are they the best measure, and if so, why - how do they accomplish being the best metric?

I'm thinking for my money, since at this point I don't understand how DFEI and S&P are being calculated, I would compare our defense with our primary opponents by ranking our Defensive PPD (points per drive) versus the other teams in the Coastal (in Coastal games only). That being the case, I'd like to see us ideally in the top 3 defense in the Coastal but would be satisfied this year being in the top half of the Coastal in ppd.

FEI and S&P+ are ultimately proprietary advanced stats, meaning that we'll never know exactly how they're calculated. IIUC, they both try to account for competition by calculating a differential with respect to the average team against the same schedule. They are each, iirc, transparent on how they calculate garbage time which is ignored.

The philosophical difference between the two stats, iiuc, is that FEI tends to put more weight on drive efficiency (points/drive) while S&P+ puts more weight on play proficiency. They combine these stats into a stat that they call F+ which they use for their "official" team rankings and predictions.

I think that the calculation of both stats has been tweaked a couple/few times over the last several years. Still, imo, they undervalue the difference between Pwr5 and Grp5 teams. I think PPD vs Pwr5 for more than a few games is the best somewhat raw stat.

 

[GTdragons]

I don't know what's best. DFEI and S&P+ defense both seem to be really good in getting the noise and gunk out (garbage time, downplaying weak opponents, …).
S&P+ had us as the #66 defense last year and DFEI had us as the #63 defense last year. I'm not going to beat myself up over the difference between 63 and 66. They're both middle of the pack, your defense won't be the thing that gets you to a bowl game numbers.
In DFEI (lower is better), we had a rating of 2.25 (#63) and went 4-6. Stanford had a DFEI of 2.24 (#62) and went 9-5. Michigan State had a DFEI of 1.64 (#25). Clemson was #1 at 0.83, UGA was #5 at 0.98, and VT and Miami were both top 15.
I made a chart. When you get around the top 10 or so, defenses don't gradually separate from the pack, they dramatically do. Same thing for the bottom 10 defenses. Between that, it's gradual improvements:
View attachment 3958

S&P+ is the same thing, just a little smoother. We're right in the middle on both. Not awful, but textbook mediocre:
View attachment 3959

Take a look at App State’s rankings. 24th by Defensive S&P+ and 23rd in DFEI. Going to be a big year for this defense!


Sent from my iPhone using Tapatalk

 

[Augusta_Jacket]

I would simply like an improvement on the the third down conversion average from last year and an improved turnover ratio. Those two stats would make me a happy camper.

I recall a saying of "live by the blitz, die by the blitz". Under Roof, blitzes did not seem effective and Tech got burned often with blitzes. Would be nice to see CNW bring new blitz packages where a defender actually gets free and is able to make a play.

I'm with you on the turnover ratio, but I think expecting improvement on a third down conversion % that was 12th best in FBS is a LOT to ask any defense.

 

[vamosjackets]

FEI and S&P+ are ultimately proprietary advanced stats, meaning that we'll never know exactly how they're calculated. IIUC, they both try to account for competition by calculating a differential with respect to the average team against the same schedule. They are each, iirc, transparent on how they calculate garbage time which is ignored.

The philosophical difference between the two stats, iiuc, is that FEI tends to put more weight on drive efficiency (points/drive) while S&P+ puts more weight on play proficiency. They combine these stats into a stat that they call F+ which they use for their "official" team rankings and predictions.

I think that the calculation of both stats has been tweaked a couple/few times over the last several years. Still, imo, they undervalue the difference between Pwr5 and Grp5 teams. I think PPD vs Pwr5 for more than a few games is the best somewhat raw stat.

Great explanation, thanks!
1. I wonder how they define "garbage time".
2. When you say "they combine these stats into ... F+", who is "they"?
3. So, are you saying that F+ is the most comprehensive of all (implying that's what we should look at if we had to choose one)?

When I think about the better defenses in the ACC, I think of Clemson, VT, Miami, and maybe FSU (depending on their coordinator). If we could get somewhere into that group with these aggregate ranking stats (like F+), something like top 4 or 5 ACC (which is where we used to be), so like top 3rd, that would be the goal. If we can get into the top half of the ACC this year, I'll be very happy.

 

[vamosjackets]

Take a look at App State’s rankings. 24th by Defensive S&P+ and 23rd in DFEI. Going to be a big year for this defense!


Sent from my iPhone using Tapatalk

The question is whether those stats do an adequate job of accounting for the different leagues/levels. Will this translate to that kind of ranking when playing a much higher level of competition, though with a higher level of talent to compete with?

 

[AE 87]

Great explanation, thanks!
1. I wonder how they define "garbage time".
2. When you say "they combine these stats into ... F+", who is "they"?
3. So, are you saying that F+ is the most comprehensive of all (implying that's what we should look at if we had to choose one)?

When I think about the better defenses in the ACC, I think of Clemson, VT, Miami, and maybe FSU (depending on their coordinator). If we could get somewhere into that group with these aggregate ranking stats (like F+), something like top 4 or 5 ACC (which is where we used to be), so like top 3rd, that would be the goal. If we can get into the top half of the ACC this year, I'll be very happy.

Football Outsiders. You can read around about the various stats starting here and following links
https://www.footballoutsiders.com/stats/fplus

 

[slugboy]

You don’t see all the math, but here’s last year’s S&P+ profile for Georgia Tech
https://www.footballstudyhall.com/pages/2017-georgia-tech-advanced-statistical-profile
You can see which games defense was better than offense, and vice versa. Special Teams is not on the chart.
Doesn’t show opponent adjustments or garbage time calculations.
Seems fair to me


Sent from my iPad using Tapatalk

 

[g0lftime]

I just want to see us increase the turnovers we create. We just never seemed to cause fumbles under Roof. Increased interceptions should mean tighter coverage and pressure on the QB. Hopefully this defense improves both.