More Pts/Drive v Pwr5 Fun

AE 87

Helluva Engineer
Messages
13,030
As a reminder, pts/drive is a "raw" scoring statistic calculated for offense and defense independently. I award 7 pts for TDs (regardless of what p.a.t. was attempted or made) and, obviously 3pts for FGs. Drives are calculated by summing (Rushing TDs, Passing TDs, FG atts, Punts, Failed 4th, Turnovers). End of half drives and safeties are ignored, and fumbles on returns are eaten as minor error.

TDs from Special Teams and Defense returns are ignored as scoring so that offensive ppd reflects only offensive drives (this differs from the scoring offense Pts/Game stat).

I also calculate a differential between a team's points/drive and points/drive allowed. So, for example, if a team's Off PPD - Def PPD is .5 and they average 12 drives a game, then they would've averaged scoring 6 pts more than they allowed.

I made the OffPPD and DefPPD calculations for Pwr5 (or BCS AQ) opponents based stats from cfbstats.com for every season under CPJ.
 

AE 87

Helluva Engineer
Messages
13,030
Where GT has been ....

Over the last 4 years
GT (vs Pwr5 or BCS AQ)
OffPPD: 2.68, 2.95, 2.55, 3.52.
DefPPD: 2.42, 2.34, 2.20, 2.46

Ave (Teams playing more than 2 Pwr 5 or BCS AQ opponents)
OffPPD: 1.93, 2.04, 1.94, 1.99
DefPPD: 2.19, 2.21, 2.32, 2.20

ACC Ave (vs PWR5 or BCS AQ)
OffPPD: 1.96, 2.15, 2.00, 1.99
DefPPD: 2.04, 2.15, 1.96, 2.02

Now, a couple things jump out from these numbers.
(1) As has been said repeatedly, our offense has been reliably efficient. The degree to which GT's offense has been better than average against P5 opponents becomes even more impressive when one notes that the ACC has had better than average Defense. Over the last 4 years we've averaged OffPPD v Pwr5 of 2.93
(2) As has been said repeatedly, our defense has been consistently a problem. Fwiw, our DefPPD v BCS AQ from 2008-2010 were 1.88, 2..28. and 2.49. Over the last four years, we've averaged a DefPPD vs Pwr5 of 2.36.
 

Carober18

Georgia Tech Fan
Messages
83
Thanks for this AE. A couple of thoughts...

I'm not sure I could extrapolate that the ACC's defenses are better than average, because the majority of the games going into your calculations are intra-ACC, so the DefPPD for the ACC is not going to stray too far from the OffPPD unless I am missing something. But, your point #1 is still spot on.

I think the other thing is that this shows how poor our defense has been, so long as you believe that a more rested defense should perform relatively better compared to a similarly rested offense.

What do folks think would be a successful season, with the defense? I would think somewhere between 2 and 2.2 would be considered a success, and anything south of 2.0 would be exceptional.
 

AE 87

Helluva Engineer
Messages
13,030
What's Does this Statistic mean for GT?

The Diff-PPD v Pwr5 stat is actually a fairly decent indicator of performance. Over the last 5 years, almost every P5 team whose Diff-PPD v Pwr5 was greater than 1 has won 11+ games.

So, based on past performance and the return of JT and the OL, I feel pretty good about predicting an OffPPD vPwr5 of 3+ for GT. That means that if we can get a DefPPD vPwr5 of about 2, then this measure predicts that we should win 11+ games.

Now, here's where it gets more fun. When Roof was at Penn State in 2012, their DefPPD vPwr5 was 1.37. By comparison, it was 1.49 in 2011 and 2.21 in 2013.

Teams whose Diff-PPD vPwr5 were greater than 1.9 typically won the national championship. An exception being 2010 when the undefeated TCU was not invited to play. In 2014, the Diff-PPD vPwr5 for Ohio State and Oregon were 1.43 and 1.49 respectively (TCU was 3rd with 1.42).

So, GT's Offense will probably be about 3.1 ppd v Pwr5. If our D improves to a little above average, say 2.1 ppd vPwr5 allowed, then we'll have a differential of 1 -- meaning 11 wins could be expected.

By comparison, Duke's D was better than 2.1 ppd vP5 allowed for the last two years. They also improved from 3.00 ppd vs pwr5 allowed in 2012 to 2.06 in 2013. So a major jump in one year isn't impossible.
 

AE 87

Helluva Engineer
Messages
13,030
Thanks for this AE. A couple of thoughts...

I'm not sure I could extrapolate that the ACC's defenses are better than average, because the majority of the games going into your calculations are intra-ACC, so the DefPPD for the ACC is not going to stray too far from the OffPPD unless I am missing something. But, your point #1 is still spot on.

I think the other thing is that this shows how poor our defense has been, so long as you believe that a more rested defense should perform relatively better compared to a similarly rested offense.

What do folks think would be a successful season, with the defense? I would think somewhere between 2 and 2.2 would be considered a success, and anything south of 2.0 would be exceptional.

Good points. On the average D, your point is right that most of our pwr5 opponents are intra-conference, but our conference Off average is on par with the national average. So, I think it works out. Ideally, I'd parse out divisions, but with cross-over it gets too complicated.
 

AE 87

Helluva Engineer
Messages
13,030
Now, my fascination with this statistic is that even though the offensive and defensive stats are statistically independent, they are of course not completely independent to the extent that they are based on actual game play.

For example, starting field position on offense and defense--which, of course, impacts the likelihood of scoring--depends on the performance of the other unit. So GT's O benefited at times from turnovers and suffered at times from the rarity of forced punts.

However, another benefit of an improved D would come from the potential way it changes opponent game plans. If we score and stop a couple of times, it will cause the opposition to become more predictable trying to score quickly.

In the past, I think our offense has forced opponents into playing more aggressively, but our D was up to the task. It will be interesting to see what happens.

Anyway, I think we have a really realistic shot at another season of Diff-PPD vPwr5 of greater than 1, and a pretty good shot at a Diff-PPD v Pwr5 of 1.5. That's a number that should put us into play-off range.

Obviously, we're playing some good teams this year, but optimism is not without some numbers to back it up.
 

Carober18

Georgia Tech Fan
Messages
83
This is great information. Thank you.

You nailed one of the key factors, field position. I think some have noted this but it is likely under appreciated overall--Butker's ability to produce touchbacks on kickoffs is very valuable, probably more for a team like GT that, in past years, probably lacked some of the depth that factories have (and the depth probably shows on ST).

There are too many factors for someone like me to calculate, but the little things matter, from field position to impacting opponents' game plans, both prior to the game and within a game. Just as you mention our offense forcing opponents into playing more aggressively, we have also seen the other side of the coin, when our offense was not producing efficiently, and the reality was we needed to throw the ball in order to come from behind. Although I would debate the cliches commentators spout off about our ability to throw, the offense is not as good at other offenses at throwing the ball when everyone in the stadium knows you have to throw the ball.

Again, thanks for the information.
 

zhavenor

Jolly Good Fellow
Messages
468
Now, my fascination with this statistic is that even though the offensive and defensive stats are statistically independent, they are of course not completely independent to the extent that they are based on actual game play.

For example, starting field position on offense and defense--which, of course, impacts the likelihood of scoring--depends on the performance of the other unit. So GT's O benefited at times from turnovers and suffered at times from the rarity of forced punts.

However, another benefit of an improved D would come from the potential way it changes opponent game plans. If we score and stop a couple of times, it will cause the opposition to become more predictable trying to score quickly.

In the past, I think our offense has forced opponents into playing more aggressively, but our D was up to the task. It will be interesting to see what happens.

Anyway, I think we have a really realistic shot at another season of Diff-PPD vPwr5 of greater than 1, and a pretty good shot at a Diff-PPD v Pwr5 of 1.5. That's a number that should put us into play-off range.

Obviously, we're playing some good teams this year, but optimism is not without some numbers to back it up.
Have you tried to somehow incorporate the teams yds per possession as well into a calculation? You already have a working number of possessions so that would be I think easy. That would give some indication of how far the offense had to go. Maybe multiply the two with some kind of weighted number to give more importance to the point per drive stat. What do you think?
 

AE 87

Helluva Engineer
Messages
13,030
Have you tried to somehow incorporate the teams yds per possession as well into a calculation? You already have a working number of possessions so that would be I think easy. That would give some indication of how far the offense had to go. Maybe multiply the two with some kind of weighted number to give more importance to the point per drive stat. What do you think?

No, especially if you have in mind some kind of starting field position indicator. I would have to go to another data source, like footballoutsiders, to get that, I think. For the data I'm using, I don't have any way to parse out yds in scoring drives from yds in drives that ended in punts or turnovers etc.

So, I can get a yds/drive, but I'm not sure what it would tell me. Typically teams that score on more drives will have more yds/drive than teams that are being forced to punt. So, for example, in 2014, Oregon averaged 3.53 pts/drive and GT averaged 3.52 pts/drive (#1 and #2 respectively). However, GT averaged 48.33 yds/drive and Oregon averaged 44.72 (#1 and #2 respectively). Does the extra 3.5 yds/drive required of GT translate into an extra .01 ppd?

I kind of like the beauty and simplicity of the raw stat right now. I'm also looking for possible correlation between ppd vs pwr5 and yds/play vs pwr5. Obviously, there doesn't have to be since slow and steady can win the race, but it's fun.
 

AE 87

Helluva Engineer
Messages
13,030
For those thinking that this year's offense may be as good as last year's, that's a tall order. GT's offense was historically off the charts last year & by a wide margin. If we're any where near as good, we'll be an excited fanbase again in January.

http://www.footballoutsiders.com/stats/feioff

Why do you post this in this thread like this? We've already discussed the data from your link, and this thread doesn't challenge it.
 

Boomergump

Helluva Engineer
Featured Member
Messages
3,284
Whatever the metric, ppd, ofei, or whatever, GT's offense was flat out dominant last year. What gives me pause is that there have been sooo many in our fanbase clamoring for a change from the spread option since CPJ arrived. That makes no sense to me at all. What could possibly be better than the offense we have enjoyed the last 7 years? Thanks AE87 for sharing the numbers that bring this to further light. I said the following on another board before moving over here and got roasted for it, but I will say it here again: CPJ's spread option attack is the best offensive scheme in the entire history of college football. That statement was made long before 2014 happened and JT was our QB and I still stand by it.

I get some people's arguments about its effect on the defense. However, my response to that has always been, if you have the chance to be the absolute best at any phase of the game, wouldn't you take it, even if it had a slight adverse effect on the opposite side of the ball? Of course you would. Being best at anything ain't easy. BTW, I don't necessarily believe it has that much of an impact on the defense, but our results on the field haven't disproven that notion either, so the jury is still out for me.
 

AE 87

Helluva Engineer
Messages
13,030
Whatever the metric, ppd, ofei, or whatever, GT's offense was flat out dominant last year. What gives me pause is that there have been sooo many in our fanbase clamoring for a change from the spread option since CPJ arrived. That makes no sense to me at all. What could possibly be better than the offense we have enjoyed the last 7 years? Thanks AE87 for sharing the numbers that bring this to further light. I said the following on another board before moving over here and got roasted for it, but I will say it here again: CPJ's spread option attack is the best offensive scheme in the entire history of college football. That statement was made long before 2014 happened and JT was our QB and I still stand by it.

I get some people's arguments about its effect on the defense. However, my response to that has always been, if you have the chance to be the absolute best at any phase of the game, wouldn't you take it, even if it had a slight adverse effect on the opposite side of the ball? Of course you would. Being best at anything ain't easy. BTW, I don't necessarily believe it has that much of an impact on the defense, but our results on the field haven't disproven that notion either, so the jury is still out for me.

Agree. I think that this year should tell the tale on the impact of our O on our D.

As I look at our personnel, talent and experience, and coaching on D, I think we should be above average. In ppd v pwr5, I'd guess 1.9-2.1 or better. If we're worse than 2.2 again, then you'd have to wonder.
 
Top