Correction: 6.5 yards per play generally puts you in elite company as an offense (the top ten in any given year). You’ll notice that on 1st downs, we’ve been “elite” in 2 of the last four years under Bobo. And last year (2010), we were “whiz bang Hawaii” elite. In 2009, we were only just below 6.5 on 1st downs. And, finally, in 2007, which was our worst year on first down, I’ll bet during the infamous second half of that season we were right up there as well.

It’s hard to see how first down playcalling is our problem.

I appreciate your kind words about the sort of discourse we try to have here. I am grateful for your faith, and I’m promoting this fanshot to the front page.

These stats also don’t show any effects to 2nd and 3rd down play calls, and the efficiencies of those. It’s great we “averaged” 7 yds on first down in 2010. So why did we ever have to punt then?

Also, more directly to Tankertoad’s point, a quality opponent means, by defnition, we run less O plays than against the cupcakes (an assumption based on the D will 3-and-out inferior teams, providing our O more TOP against their inferior defense). This drives the statistical average up, and gives Bobo’s offenses inflated numbers.

Does anyone know where I can get detailed (like statistician detailed) data breakdown on football games? It’s got to be out there. I’d like to crunch some numbers myself.

the cause of that problem is the question about which we (might, potentially) disagree. We can’t refer to the fact that we are losing the big games as evidence for any particular theory that purports to explain why we are losing those games. Right?

I’m not sure what I’ll be able to do to split these stats out further, especially as far as 3 and outs are concerned. But random variation around the mean by itself indicates that we’re not doing awfully in that category. If you average 7 yards on first down, then you’re almost certainly not having a lot of three and outs. It’s not like we’re getting one 70 yard play every ten first downs and then nine plays of zero yards, putting us in 2nd and 10 90% of the time when we throw on 1st. Again, looking at the number of 15+ and 25+ plays we hit shows that we also hit plenty of medium plays, too.

Last season, our average result when we threw the ball on 1st down was 9.9 yards. To put it mildly, that’s insanely good. That’s not our average per completion, it’s our average per pass attempt. Of course, there’s variation. So, sometimes we completed for 20 yards, and sometimes we threw incomplete. But on average we virtually got the first down on a single play. That is a very, very potent passing attack on first down.

Furthermore, we didn’t throw most of the time. Look at the run/pass splits. We threw more on first down last season than any of Bobo’s previous three seasons, but then again we should have since we were so successful at it (again, 9.9 yards per attempt). (Our interception rate was very low last year as well, don’t forget.) Still, even at this highpoint of passing frequence on first down, we ran the ball 57% of the time on 1st down in 2010. So how often are we “first and bombing” it, really? On the 43% of times that we actually did pass, we got the first down immediately almost 1/3 of the time. About 1/8 of the time, we completed a pass of 25 yards or more (23 times). Even if every single one of our incompletions on 1st down last year was on a failed “first and bomb” attempt, which is certainly not the case, that means we had a success rate on “first and bomb” of 27.3% (23/84). This, notice, is much higher than the frustrated estimates we’ve been throwing around in these threads here on Dawgsports, which have been in the 5-10% range.

All that said, I would like to split these out to distinguish our performance vs. good teams from all the others. However, we should also be in agreement before I do that that all teams will expect to see significant drop-offs in their performance against their better opponents when compared to their overall performance against their entire schedule. Agreed? How much worse will be the question, but in the last two weeks I’ve already done a go-around with a crazy person over at the Senator’s blog regarding our scoring offense in games against ranked opponents, and in that category UGA does much better relative to the rest of the conference in that category than they do overall. In other words, we drop off less than other SEC teams do when we play ranked opponents, at least as far as points on the scoreboard are concerned. I won’t be surprised if we find something similar for our first down performance (it will be worse against ranked opponents, but will still rank about the same or even better in the SEC as it ranked against all our opponents put together).

But let me try to break the numbers out and take a look.

Finally (for now), let me say that I like your hypothesis as a hypothesis. It isn’t crazy or unreasonable. Both the pacing issue and the issue of putting the D in a hole if a high-risk/high-reward offense sputters are very important. Your hypothesis (if I may call it that) is a very reasonable one on its face, and if it turns out to be true then Saturday’s game will be the quintessential example of its veracity for years to come (over 350 yards of offense, but never even ran a single play in the red zone! That. Is. Astounding.) Furthermore, I admit that I’m not offering an alternative hypothesis at the moment, which kinda sucks of me. But what can I do? If I figure one out, I’ll put it out there for everyone to pick apart, I promise. In the meantime, though, I don’t think the facts bear out the “1st and bomb” hypothesis of our struggles. I don’t have to have an alternative ready to go to point out the failings with the current regnant theory.

but again, if you average 6.5-7 yards per play on 1st down (not to mention last year’s 9.9 yard average when we threw on 1st down), and the standard deviation is not something crazy, then we can be reasonably certain that we are not producing “3 and out on our next 4 possessions” after completing a long bomb. Of course, at some point in the season, you will have some bad streaks with several 3 and outs in a row. But that is true of every offense. If you flip a coin 1,000 times, you will get some “ugly” streaks in there. You’ll nail tails eight times in a row, or get heads seven times in a row, etc. Purely random variance is going to produce some streaks, even if you are 1995 Nebraska.

Instead of flipping a coin, think of it as a ten-sided die (watch out for Tech fans coming into this thread now). You are such a potent offense that you always have a 70% chance of getting a first down on any given play. (This is an impossibly good rate of getting first downs, used as an illustration). You’re just that good. So, every play you run is like rolling the ten-sided die, and if you get 4-10 you get a first down.

Over the course of a season, you will roll that die well over 700 times (over 700 plays per year, assuming you aren’t doing the Oregon/Auburn/West Virginia/Okie State hurry-up thing, in which case it will be closer to 1000). In those 700 “rolls of the die,” you will, despite how awesome you are as an offense, have three and outs. You will even have some three and outs in a row. (i.e., 6 or more consecutive rolls getting 1-3 on the die). And, of course, real offenses are always much less potent than this.

So streaks happen. But, you can use your average yards per play, and your situational stats for the various downs (1st, 2nd, and 3rd), and the variety of outcomes you get on those plays, and you can see pretty clearly that teams with higher average yards per play are not going to go 3 and out as often as teams with lower average yards per play. It would be a statistical fluke if we gained 7 yards on first down, on average, and also did at least average or better on 2nd and 3rd down, and yet ended up with a “lot” of 3 and outs. The truth is, we don’t have a lot of 3 and outs.

Now, the ones we do get might come at the worst possible times, etc. I do suspect that our big problem is situational, dependent on the in-game context. I don’t think we have some general problem of ‘calling too many first and bombs", or “not running enough,” or “not enough/too much I formation,” etc. I suspect that, if there is a weakness in playcalling, it’s all about timing of calling the wrong thing at the wrong time, and doing so in a way that isn’t just dumb luck (everybody sometimes calls “paper” when the defense has called “scissors,” and that’s just the way it goes).

I went through this on blutarsky’s site a few weeks ago with a crazy dude. (Not that you’re crazy). It appears the Senator took down the post, b/c now i can’t find my (admittedly too lengthy comment) where I discussed all of this. Let me try to sum it up here, piggy-backing off of what you just did to get us started.

Every team has a drop in their scoring offense when they play ranked teams. So the real issue is, how far do we drop vs. the drop of other teams in the SEC? And the answer there is that we have, except for last season, done very well in this regard.

SEC rank in scoring offense, all games
2007: 5 (32.5)
2008: 3 (31.5)
2009: 7 (28.9)
2010: 4 (32.1)

SEC rank in scoring offense, vs. ranked opponents
2007: 2 (35.5)
2008: 3 (26.5)
2009: 5 (20.0)
2010: 8 (15.8)

Put 2007 aside for a moment, and we’ll come back to what an amazing thing the Dawgs did that year in a moment. Notice the general pattern: in three of the four years of Bobo being the full-time OC, we ranked as well or better in the SEC vs. ranked teams as we did in all our games combined. In other words, yes our scoring dropped off when we played the best teams on our schedule, but so does everybody’s scoring when they play the best teams on their schedule. And our drop-offs were not as steep, over all, as the rest of the conference’s drop-offs. Why? Because we ended up ranked more highly (or equally) once everyone’s drop-offs are taken into account.

Now, what about last year, the one year this isn’t true? The “nice” thing about last year’s miserable showing (haha), is that it is very easy to explain. We played the majority of our ranked opponents early in the year, when AJ Green was suspended, the coaches were holding the playbook back to compensate (rightly or wrongly) for a freshman QB, and when offenses usually struggle anyway. As a result we scored 16.3 fewer points per game against ranked opponents than against our combined schedule, and our SEC ranking fell from 4th overall to 8th vs. ranked opponents. This terrible showing was terrible (just as redundancies are redundant), but it’s pretty easy to understand how the numbers came out that way.

For reference, I mentioned every team drops off in scoring when they play ranked opponents. How much so? (What follows is my very close but not exact memory of the SS I had put together on this a couple of weeks ago, but which I have lost for the time being. Full disclosure)

The average drop-off is 7 points, and the standard deviation is about 4. For non statistics-inclined folks, that means that about 67% of SEC teams drop off anywhere from 3 to 11 points per game in scoring when they play ranked opponents. Of the 48 SEC offenses since 2007, only 7 have done better than a 3 point dropoff and only 6 have done worse than an 11 point drop off.

For fun, two of the worst drop-offs were Arkansas, in 2007 and 2009. Another was UGA last year, yes. (But remember that we did pretty well 07-09). And the second-worst drop-off of all was last year’s Florida squad, which plummet from 29.8 ppg to 12.7 ppg against ranked opponents, and from an 8th ranking to an 11th ranking in the SEC. This kind of thing happens, folks.

Much more fun to talk about are the handful of teams that actually did better than a full-standard deviation-below-the-mean drop-off when they played ranked teams (which, again, means you had a drop-off of less than 3 points or so). Of the six teams that had a “dominant” drop-off number from 2007-2010, two actually went up when they played ranked teams. In other words, their “drop-off” was not actually a drop-off at all, but a rise-up. This should be pretty shocking any time it happens, for obvious reasons. How do you put up more points against your good opponents than you do against your schedule-at-large? Something funky is going on when that happens. The two teams that pulled this off were 2008 Ole Miss and 2007 UGA. (2007 Bama only had a drop-off of 0.1, and 2007 UK only had a drop-off of 0.6; both very strong showings but still drop-offs). In 2008, Ole Miss actually went up 0.6 ppg when they played ranked opponents vs. what they scored against all of their opponents put together. An interesting occurrence, to be sure.

But 2007 Georgia? The Dawg went up by 2.9 ppg when they played ranked opponents (from 32.6 to 35.5)! This puts them over 2 full standard deviations from the mean, making this basically a 1-in-100 event. In the next four seasons of SEC play, we probably won’t see a team do this again.

Of course, the explanation for how Georgia did this in 2007 is pretty clear. 2007 was the mirror image of 2010, in that we played our ranked opponents at the end of the year instead of the beginning. Not only that, but 2007 was just one of those years where a team “wakes up” in the middle of the year and fundamentally changes their play on the field. Moreno became the feature back, Stafford started hitting lots of long passes, and the Dawgs rocked Florida (Tebow not 100%), blacked out Auburn, toyed with Tech, and humiliated the entire state of Hawaii. Earlier in that season, the Dawgs had struggled against what were in hindsight mediocre South Carolina (12 points) and Tennessee (14 points) teams, and they only got into the low 20s against Vanderbilt. So, the Dawgs struggled, then woke up and dominated, and happend to play the best teams when they were dominant. 2007, what a strange year it was.

OK, so summing up: against ranked teams, Bobo’s offenses generally do better relative to the rest of the SEC than they do against their combined schedule. So it is hard to see our performance against ranked opponents as the sign of our decline, seeing as how we outperform our conference in that regard, even as we’ve struggled to win games.

We threw for 1601 yards on 160 first-down pass attempts last year. I know how amazingly good a stat this is, which is why I kinda made a thing out of it. :-) If Bobo is killing our offensive production, it’s not because he calls too many long passes on first down.

Quickly, I wanted to respond. That may be correct, though I want to say for College ball it doesn’t apply. I have no numbers to validate that, and I could be talking out of my delusional mind. I’ll test that observation tonight with recent MNC winners.

especially as you describe them in the second paragraph. I’m not sure they are indicative of our “normal” Bobo-era problems, though, or just what happened to go down against Boise. The insane “350+ yards of offense but zero plays run in the red zone” thing, for example, is definitely not normal for us. Similarly, I don’t think we usually have so many drives go 5 plays are fewer.

A lot like 3rd down conversion, these kinds of numbers tend to jump around from game to game. At the end of the year (or midyear as a few games pile up), we can aggregate and draw a conclusion. But we might be on our way to a 45% 3rd down conversion rate, for all we know, and this was just our one stupid game along the road. I don’t mean that flippantly, but in all sincerity. These bad things all tend to be bad in bunches. We were explosive but not methodical on Saturday, which means we had a lot of 3rd and longs. When you have a lot of 3rd and longs, you tend to have a low 3rd down completion percentage. In any event, even a very bad team over the course of a year is going to break 30% on 3rd down conversions. So, it is definitely safe to say that Saturday will turn out to be an unusually bad night for us in that regard. Just how much better we actually are at a thing like that over the course of the season, remains to be seen.

I am a believer in Boise, though. If that had been the 2003 SECC game, we would just grudgingly accept that LSU became something better than us as the year went on, and whipped us. No embarrassment over who administered the whipping. But because it’s Boise, we might be over-reacting just a bit. (This includes the Sunday morning version of myself, too).

A very good team that needs some time to find it’s footing could have looked like we looked Saturday night. Of course, a pretty bad team that ain’t gonna gel at any point could also have looked like we looked Saturday night. I’m not putting on happy glasses, because we don’t know which is closer to the truth at this point. I’m just asking us not to put on the sad glasses yet, either. Season openers are as misleading as bowl games (any single game is misleading as an overall indicator). At least we got beat by someone who is good, is all I’m saying. This is not “omg we just fell to Colorado.” Not yet.

Small consolations, these. But let’s give the boys a chance against Carolina (I know we will).

In the Boise State Game, and Xon, I realize your point is about a much larger set of data, Georgia actually had a higher yards per play than Boise did:

Total Yards 381 379
Total Plays 60 68
Mean 6.35 5.57
Std 13.76 5.96

Number of plays better than five yards 20 30
Percent of plays better than five yards 33% 44%

Xon mentioned that you really should not expect high yards per play and many three and outs unless you have a very high standard deviation. On Saturday that is exactly what happened. On average Georgia got as many yards as Boise. However that similar average masks a huge difference in volatility.

When we talk about a “first and bomb, Carlton Thomas up the middle,” stereotype, it might not be accurate over the course of a full season, or many seasons. However, I think it is quite accurate that the difference in Georgia and Boise’s offensive production was consistency.

Xon, you obviously know your stuff where the stats are concerned; I’d be interested in your thoughts on whether the volatility on Saturday was the exception or the rule.

My take is that rather than rolling ten sided dice where each outcome is equally likely, Georgia’s offense is more like russian roulette, where the bad outcomes (from the defense’s perspective) are rare, but really bad when they happen. I’ll try and unpack this in a full post before too long.

The non stats. If you get a TD off of first and bomb, it looks great on paper.

But what I saw on the field in the first half was our D taking BSU three and out repeatedly. Where is the “flip the field” stat. I dont care what any stat says, find one person that says “we didnt have rely on our D over and over”.

IF our O gets 20 yards, our punter has the ability to pin them in. Instead, we beg our punter to save our O. Starting field position along with TOP – we are slaughtered. If you put me on the headset with Staffiord and MoMass, I would have great numbers for 1st and bomb. That’s not the problem. The problem is the chess match, which is something you can see more than stat.

You’ll see that 2009 and 2010 that is exactly what we did. Compare that to 2007 or Florida of 2008, Alabama of 2009: the ratio (1st down, 2nd down, 3rd down) of run plays to pass plays is approximately: 2-2-.75. In laymans, on first and second down, the 2007 Dawgs, and the 2008/2009 SEC/Natl champs ran the ball twice as much as they threw. Auburn last year was the anomaly, being a significantly run-first offense, whose ratio’s were actually outside the standard deviation.

For some reason, the Dawgs went to a 1.34-1.85-.65 ratio in 2009, and it appeared the same in this season’s opener. Why is this significant? It validates Tankertoad’s (and others’) observation of 1-and-bomb, run-up-the-gut, desperation pass on 3rd.

Xon has done great work on the success of 1st down, and what I’d like to challenge him on is our subsequent success on 2nd and 3rd, particularly against top 25 or even 50 defenses. I haven’t found that data yet.

Realistically, these are good numbers, they give us a glimpse, but we have only just started studying this. They are by no means comprehensive or even revealing of any one particular issue. Except, IMO, the points per game versus ranked opponents. Xon notes the SEC as a whole has declined with the Dawgs since 2007, but if we look (again) at the SEC champs—thus Natl Champs—of 2008, 2009, and 2010, their points per game against ranked opponents are 43.6, 32.1, and 41.2 respectively. FAR from Georgia’s 15.8. I think that illustrates the problem…but why? We could turn this into a master’s thesis, easily!

There’s got to be an accounting method we can work in this mix. On offense, positive yards would be a debit and negative a credit; on defense the converse…then any advantage in turnovers and TOP would be retained earnings! ;)

at least from the Boise game say exactly what happened. Georgia’s offense got a lot of yards on a few plays and not a lot of yards on many plays. I think this had many consequences, including, keeping the defense on the field for too long, not creating drives that got into field goal range even if they didn’t score touchdowns, and creating bad field position for the defense.

That field position disadvantage, was exactly what you don’t want against an offense like Boise’s that doesn’t make many mistakes – you need, to make them repeat more sets of downs for their mistakes to come through.

The statistics and the results on the field sometimes tell different stories, but in this case, I don’t think they do.

I agree that the eyeball test is a real thing, and it matters. I’m not a “stats rule the world” guy, actually. Also, remember, I’m not using these stats to defend Bobo. I’m all for us figuring out that Bobo is doing X, Y, and Z wrong, and he needs to reverse course. So, it’s not as though I’m saying “the stats say we’re good, therefore Bobo is a great OC.” That’s not my point at all. My point is that, when you offer a particular explanation of how Bobo is messing up, that particular explanation might be something that we would expect to show up in the stats. In the case of “1st and bomb,” for instance, it is clearly something we should be able to see in the stats. And we don’t.

I am inclined to think, with you and several others, that the real problem lies in context-specific in-game decision making, with the way our fairly potent offense ends up being utilized in this or that situation. On that sort of thing too, though, there will theoretically be some stats that might help us verify of cast doubt on whatever theories are offered. But those stats are going to take some more work to gather, because in-game play-by-play statistics are something that isn’t ready to hand the way things like total yards or points are.

Haven’t seen you post in a while? Are you back home?

Dr. Saturday sometimes uses the stat “wasted yards.” When examining box scores of games whose yardage and time of possession totals don’t match the scoreboard, he breaks out total offense into drives that scored points and drives that didn’t, to show which yards were “wasted.”

I haven’t done that for last Saturday’s game, but, given how close the yardage totals were for the two teams, it’s pretty clear that Boise State wasted fewer yards than Georgia.

Of course, the problem with that metric is that not every drive that fails to end in points is useless. As tankertoad rightly notes, a time-consuming drive ending in a punt can be useful if it alters field position, and, sometimes, scoring isn’t in your best interests (Washaun Ealey in the fourth quarter of last year’s Georgia Tech game; the Maryland defensive back who intercepted the pass at the end of last night’s game). As noted in the multiple recent references to the final drive of the 2002 Ole Miss game (the final drive of the 1978 Florida game applies equally as well), a time-consuming drive that chews up clock and allows you to kneel it out to get the win in a close game is a good drive, because it did what it was designed to do.

That, I think, is the crux of the point: An offense must be judged by (i) what it is designed to do, (ii) whether it does what it is designed to do effectively, and (iii) whether effectively doing what it is designed to do leads to victory. When Mark Richt was calling the plays, I knew what our offense was designed to do, and I was able to judge whether it did so effectively (it did, even though it did not score as many points as Mike Bobo’s offenses, because it was designed to serve multiple functions, of which scoring points was an important one, but not the only one) and whether it led to victory (which it also did, to an extent not seen since the early ’80s).

I don’t know what Mike Bobo is trying to do, so I have a hard time telling you whether he is doing it effectively (which is why the stats and the eyeball test appear to tell us dramatically different things; if he doesn’t know what he’s trying to accomplish, we have a hard time coming up with a rubric with which to grade his efforts), and I know it isn’t leading to wins. What worries me is that I’m not sure Mike Bobo knows what he’s trying to do, either.

I am reminded of the scene in “Patton” when George C. Scott wakes up a soldier who’s supposed to be on duty after touring the barracks and finding it a shambles. Scott asks him what he’s doing, and he honestly replies, “Trying to get some sleep, sir.” “Get back down there, son,” Scott answers. “You’re the only son of a bitch in this outfit who knows what he’s trying to do.” If you don’t know what you’re trying to do, you’re almost certainly not succeeding at it . . . and, if you did, how would you know it?

That’s why Mike Bobo can make a spectacular play call to get the ball to Brandon Boykin, have it succeed fantastically, and never call it again. He didn’t know what he was trying to do, so how was he supposed to know when he did it?

Because we dont have a "WTH" stat.

That’s a damn good line. I am not trying to give myself credit. It’;s just freaking funny, true, and a good one.

BSU is a big play offense, they thrive on the big play. But they realized we had the secondary that wasn’t going to give up the big play and adjusted. Their OC took advantage of what was given. And exploited it. And embarrassed us.

includes too much conventional wisdom and common sense. Therefore Bobo will not do it. If it makes sense to do it, do the opposite, then the opposing defensive coordinator will never expect it!

Hopefully I can get to it tonight.