My blog has moved!

You should be automatically redirected in 6 seconds. If not, visit
http://www.ledger-enquirer.com/bulldogs-blog/
and update your bookmarks.

Wednesday, September 30, 2009

Bobo & Willie: More Fun With Numbers

Yesterday I posted some interesting statistical work done by reader Jim F. regarding Willie Martinez's ability to make in-game adjustments. The reaction from readers was mixed, which is about what I expected. First, as I readily admitted in the post, stats don't prove everything, they simply provide context. If that context is being provided to an already fully-formed opinion, well, it's not very useful. For everyone else, my hope was that it might shed more light on the questions of why Martinez's defenses were struggling. In either case, it was not meant to say that Willie wasn't struggling.

But, the real beauty of yesterday's post wasn't the grand illumination of Georgia's defensive misfortunes. It was that it ignited a bit of creative and mathematical ingenuity on the parts of some other readers. So, let's take a look at what they found...

First, from David E., who wasn't sold on the numbers that supported Willie's ability to adjust in game. One of the common criticisms of the analysis was that teams are often less aggressive in the fourth quarter when they have a lead, so numbers across the board would be down.

There's some merit to that, I'd imagine, but that hasn't really been the case for Georgia's opponents this season. Either way, numbers only make sense in context, and David wanted to see how the Bulldogs' competition fared from quarter to quarter, too. Here are the results:

Looking at yards allowed by Georgia, Georgia Tech, Florida and Vandy this year....

Time
UGA
Ga Tech
Florida
Vandy
All 972 861 480 425
1st half
554 378 269 264
2nd half
418 483 211 161
1st qtr
305 137 74 74
2nd qtr
249 241 195 190
3rd qtr
302 281 87 63
4th qtr
116 202 124 98

Now by QB rating...

Time
UGA
Ga Tech
Florida
Vandy
All 125.79 133.09 75.29 80.19
1st half
136.57
112.62 88.55 85.54
2nd half
115.29 155.3 61.85 72.94
1st qtr
165.19 96.02
67.46 67.46
2nd qtr
106.33 127.72
104.8 99.88
3rd qtr
147 215.22 61.54 66.48
4th qtr
75.89 115.92 61.98 79.7

And lastly, by first downs...

Time
UGA
Ga Tech
Florida
Vandy
All 41 34
24 18
1st half
23 16
14 13
2nd half
18 18 10 5
1st qtr
13
5
5 3
2nd qtr
10
11
9 10
3rd qtr
13
10 3 2
4th qtr
5
8 7 3

So, what does this tell us? I'll let David give his thoughts:

"As you can see, we are not that great at "making adjustments." Between the first and second half, only tech did not improve in their defense. Also, while it seems that our fourth quarter numbers are low relative to our first quarter numbers, this isn't unique to us, and, I think most disturbingly, speaks more to how terrible our first quarter numbers are. This little exercise has not helped my confidence. Cripes!!"

Two points I'll agree with David on: First, "Cripes" is a great thing to yell in situations like this. Second, Georgia's first quarter numbers (and really, the numbers overall) are pretty bad. Again, there have been no cupcakes on the Bulldogs' schedule like Tech, Florida and Vandy have enjoyed, but Georgia's numbers across the board are almost double what Florida and Vandy are allowing.

On the other side of the coin, I will offer this: In virtually every category for all three teams David studied, the numbers went up in the second and fourth quarters from where they were in the first and third quarters. For Georgia, however, it's just the opposite.

I'm not sure what that means, but here's one thought: Football is a game of adjustments. You come in with a game plan, and one team's is better than the others. Then coaches adjust, and you might expect a swing. Then coaches adjust again, and you might expect a swing back. Then again, and so on. These numbers tell me that for teams like Tech, Florida and Vandy, they've come into the game with a good plan, then adjusted at the half to the adjustments made by the opposition for the second and fourth quarters.

Georgia, on the other hand, has come in with a game plan that hasn't quite worked, been forced to adjust midway through each half.

Obviously that is far from fact, but it's one way of looking at the numbers, and it's a different opinion than I think most fans have of Willie.

But let's get back to the originator of this little exercise, our fine reader Jim F.

Rather than hide behind his first bit of analysis, Jim went back to find some context. To be fair, he included some delightful graphs of these numbers that I am simply not smart enough to get to format properly in this blog post. If any of you would really like to see the full standings, email me and I'll be happy to forward along the file in Excel format.

Anyway, here's Jim's explanations of his work...

What I got: The Overall QB Rating, 1st Half QB Rating, 2nd Half QB Rating and 4th Qtr QB Rating for all SEC teams. I want to get a SEC Rank of where UGA pass D adjustments stood.



Theory: Regardless of opponent, a "good DC" will make adjustments at the Half to nullify what the opponent is doing in its game plan, and within the game or late-game performance (either 'going for the throat', a trade mark of a "good defense" or coming up big in pressure situations)



Limitations: QB Rating itself -not the prefect stat, as stated the other day. If you asked, I doubt seriously not even 90% of even hard core fans, without Google, could accurately tell you what the formula is. I believe it is: ((QB age/2) + sq root of QB IQ) + (on a 10 point scale the Avg. score of cheerleader rooting him on) + TD)) - minus interceptions. :>). Suffice it to say the lower the "better" a Defense is playing.

(*Say that might be worth another prize pack: What is the formula?*)



I used variance as the indicator, again attempting to limit the argument - weak opponent vs. stronger opponent. What happened during the course of the game was what I was trying to get to. Variance is the change that happened, supposedly credited to decisions made during the game on the sidelines - adjustments. You could have used raw numbers, but I didn't go that route for reasons stated.



By using variance, it does legitimately leave some room for criticism. That is it is really measuring "Most Improved D/C during the game Award". I am taking two snapshots: One at the half and one at the end - giving out two pieces of hardware. For the halftime, I used the drop, hopefully, seen during the second half as compared to the first half. For the late-game performance, I used the variance between fourth quarter and overall.

But the truth is again that it is like "test score improvement". Well a 25 to 50 is a 100 percent improvement, but that potential "misleading" possibility is still out there. Being totally fair though three wins out of 4 is not a total failure.



Also the other potential 'misleading' variable is a) the number of games played and b) the number of BCS opponents that they played. Can other teams D stats be misleading because they played FLA Atlantic instead of a Big 12 team? After I got the QB Ratings to help minimizes that distortion or "noise" I went back and added the number of BCS opponents each team has faced on top of the graphs (on right hand side axis), and will let you decide if a team's D stat is puffed up by cupcakes or legit. But I didn't add in total games to graph, too confusing. But getting rest and playing with rested legs after 2-a-days is an important factor.



Results: UGA D Ranked fourth in halftime adjustments in making QB Rating drop, and first (best!) with fourth quarter adjustments. Auburn actually ranked No. 1 in halftime adjustments. With all the attention on the other side, I guess Chizik is taking on his D/C, umm HC, responsibilities.

OK, so Jim's additional research doesn't really vary greatly from David's, despite studying some different teams. So I think it's fair to say that, comparatively, Georgia -- i.e. Willie -- is doing a good job in the second and fourth quarters of games. But again, what does that mean?

That's the part that's open for debate, and I would guess that your explanation will likely be defined by how you view Willie as a coach to begin with. Indeed, the numbers are only a part of the story. You have to make the ending up for yourself.

Of course, we've spent all this time talking about the defense, and that's hardly fair to Mike Bobo, Georgia's offensive coordinator who has earned a share of criticism himself.

I wrote in my practice notes yesterday how Georgia was hoping to jump start its running game. Well, reader John B. decided to do some homework on the issue and thinks the problems may not be with Richard Samuel or Caleb King, but rather with Mike Bobo.

I'll let John explain his premise:

"I wanted to see how UGA's play selection and success/failure compared to other top teams on first down. My criteria was the team must be top rated and can't have more than one loss. Plus, I took the games that those teams played against their best competition. This was done because UGA has played no cupcakes yet.

"Games Selected: Bama vs VT, Boise vs Oregon, Miami vs GT, LSU vs Washington, USC vs OSU, Jean Shorts vs Coonskin Caps, TCU vs Clemson, South Carolina vs Ole Miss.

"The summary of the above winning team's first down play calling was charted to compare to UGA, then averaged to see how UGA's average over four games compared to these top programs' eight games."

Interesting premise, so let's see the results...

Category
UGA
Others
1st down passes 47 69
Completions 28 41
Incompletions 19 28
Comp %
59.57 59.42
Pass Yards
512 522
Avg Yd/attempt
10.89 7.57
Rushes 57 160
Rush yards 287 836
Rush yds/attempt 5.04 5.23
Pass to Run Ratio 45%/55% 30%/70%

(*Remember, this is play calling on FIRST DOWN only, not overall. So while Bobo appears "balanced" you have to factor in that first is traditionally more of a running down, while third down, for example, tends to skew in the other direction.)

This is pretty fascinating, but let me allow John his thoughts first...

* UGA is passing on first down 15 percent more than other top programs
* UGA's avg yards per pass attempt is significantly higher than other top programs. More than 3 yards per attempt. Thanks AJ and Orson!
* UGA has 10 fewer yards passing on first down than the 8 top teams selected combined
* Joe Cox's completion % on first down is nearly identical to the other 8 top teams
* UGA's avg yards per rush is on par with other top teams. Only lagging by less than .25 yards per attempt

John's conclusions:

"I believe UGA's struggles with the running game is a direct result to the lack of commitment to running the ball on 1st down. If Mike Bobo doesn't make a concerned effort to establish the run from the very beginning, how can you expect to be able to run it in crucial game winning or clock eating drives? I believe if this trend continues it will be tough to compete in the SEC East.

"Success in the running game comes from a team mind-set and commitment to be a physical team. You have to practice it in order to create the mind-set and team attitude that "no one is going to stop us." If we need two yards, we WILL get 3.

"I understand you have to get the ball to AJ. He's the best player on the field (PERIOD). BUT look at the Arkansas game. UGA ran the ball on every first-down play in the first 5 drives of the game. This opened up first down passing plays later in the game of 25, 20, 50, 18, 44 and 23 yards. Cox was 6 for 8 for 180 yards after we commited the first five drives to establishing the run.

"YOU RUN TO PASS NOT PASS TO RUN!"

OK, back to my thoughts...

First, you have to wonder about the simple paradox of cause and effect here. Is Georgia's running game struggling because they aren't running enough on first down or are they not running enough on first down because the running game is struggling?

I think the answer to that has to be the former, because as John points out, Georgia's success on the ground on first down is virtually identical to the success achieved by the victorious team in those other eight games.

Still, I'm not entirely sure I agree with John's conclusion that by not running on first down, Georgia is failing to set up the proper mind-set for success. There may be some truth to that, but I think the bigger issue is a matter of situational dynamics.

I spoke with Mike Bobo yesterday about the running game's problems, and he said he thought a good bit of that had to do with turnovers and penalties. It makes some sense.

Georgia ranks 115th nationally in penalty yards (314), with a majority of those coming on offense. That means Georgia is facing more first-and-15s or second- and third-and-longs as a result. Those aren't running situations, obviously.

Also look at the Bulldogs' turnovers. They have 12 already this year. Two occured on special teams, which cost the team a drive. Add in the fake punt by South Carolina, and that's another drive that Georgia lost. Then look at when the other 10 turnovers occurred:

Against Oklahoma State, Georgia had three turnovers -- one on the second play of a drive, one on the third and one on the fifth. That last one occurred on Georgia's last drive of the game when the Bulldogs were in pass-only mode as they tried desperately to play catch-up.

Against South Carolina, Georgia had two more offensive turnovers -- one on the second play of a drive, one on the first.

Against Arkansas, Georgia also had two offensive turnovers -- one the first play of a drive, one on the third.

And last week against Arizona State, Georgia had three offensive turnovers -- two on the third play of a drive, one on the first.

So that's essentially nine more drives that last three plays or fewer before ending with a turnover.

Then go back to John's numbers again. Georgia is throwing on first down 45 percent of the time. Those throws fall incomplete 40 percent of the time. That means that 18 percent of the time, Georgia is facing a second-and-10, which again is hardly an optimal running situation.

The bottom line is that Georgia's running game hasn't been given a chance to succceed yet. If 18 percent of its drives result in a second-and-long, 12 drives have been cut short by turnovers in three plays or less, and the team ranks among the worst in the country in penalties, there is simply no way for the running backs to get in a groove because the situations rarely dictate that being a possibility. Georgia has run the ball as much as they have because Bobo is insisting on it, not because the game has allowed it.

So while I think John has some serious merit in his analysis, I also think that if the Bulldogs can hold on to the ball and cut down on the flags, there's a good chance we'll see some improvement in the running game without any other factors being considered. Add to that improved play by Samuel and King, some more efficient play calls by Bobo and better blocking by the disappointing offensive line, and Georgia may really have a nice ground game going by the end of the season.

ADDENDUM: There's really nothing like tying two unrelated thoughts together, right? And while the analysis of Willie and the analysis of Bobo may not seem related that closely, here's some food for thought:

Georgia has 53 offensive drives so far this season (*not including drives to run out the clock before the half or to end the game).

A poor running game or poor play calling can result in short drives and lots of three-and-outs.

Twelve of Georgia's drives have ended in turnovers in 5 plays or fewer.

Eight have been scoring drives that lasted 5 plays or fewer.

Nine have been three-and-outs.

Seven more have been four or five plays before Georgia was forced to punt.

So far this season, Georgia has had possession of the football for a total of 113 minutes, 45 seconds out of a possible 240 minutes (i.e. 47 percent of the time).

But, add in South Carolina's fake punt, and on 37 of 54 possible possessions (69 percent) Georgia has run five plays or fewer before either scoring or giving the ball back to the opposition.

Since the defense cannot give up any yardage when the offense is on the field, and since the offense has not been on the field nearly as much as it could have been, it's fair to say that there's been a bit of an extra burden put on the D this season.

If Georgia was even moderately more successful in avoiding penalties and short drives, that time of possession number could swing wildly in the other direction, which by definition, would reduce some of the ugly numbers going against the D right now.

21 comments:

Anonymous said...

It would be nice to know how sucessful Samuels was on sweeps to the short side of the field vs King's success? We know the freshman can not run up the middle but how does Samuel compare up the middle vs either of the other tail backs.

David Hale said...

I might be able to find the numbers for toss sweeps, but at this point, the totals for King and Samuel probably would be such a small sample size that I'm not sure how much real value they would have. Until we get a little further into the season, that type of thing is probably one of the issues where scouting outweighs stats heavily. It simply LOOKS like King is better, and that's probably the case.

JRL said...

"Cripes" my head is spinning now. Bottom line we will not have many answers (numbers be damned) until we play a few more games and Florida plays someone (anyone really) worth a damn.

In the interim I have very little faith in BoBo or Martinez.

Top 10 recruiting for 5 or 6 years running and we are looking up at Bama and Florida - something is amiss.

NCDawg said...

I love this. Stats lie, but it is so much more respectable way to lie than just having an opinion.

The defensive numbers are really interesting. First, I can conclude John is either unemployed or a government employee, he has WAY to much time. Second, UGA seems to be relatively better at in-game adjustments (1st Qtr to 2nd Qtr and 3rd Qtr to 4th Qtr). These tend to be minor adjustments in reaction to what they are seeing on the field. These could be as much our talented defenders reacting as it is coaches making adjustments.

Where Georgia seems to be week is in scheming - setting the plan based on what the other team will do. This is shown by the comparative terrible performance in the 1st Qtr and to a lesser extent in hte 3rd Qtr after half-time scheme adjustments.

One contributing factor to scheming is preparation time. The defense seemed to work pretty well against a high octane Okie St. The rest of our opponents with early exhibition games could prepare for the games for weeks. If preparation time is the key, the importance of scheming should start to balance out - beginning with LSU this week.

Dawg4Life said...

I agree that CWM has a really hard time setting up a scheme for the opposing team's offense. With the trend and the numbers that you have provided we will be playng catchup all game. CWM does make adjustments during the game but again once half time comes the opposing coach makes adjustments and we are right back where we were when the game started.
Interesting part is due to the lack of running the ball on first down the offenese is playing catchup every series by not gaining yards on first down. I understand that we have a great WR in Green, but why would we not try to establish a running game to pull the defense up and setup the play action passes.

Anonymous said...

You can argue the stats all day but I would argue that the "improvement" in 4th quarter adjustments is not relevant. If you watch the games for the last three years, teams always seem to jump on us early offensively. It seems to me that we play a "read and react" kind of defense then adjust. Unfortunately, we find ourselves playing from behind way too much (see this year). Yes, turnovers are part of the problem but this goes back to the last two years when we didn't turn it over as much. We rarely seem to put "mental" pressure on the opposing teams' offense.

Liam said...

Statistics help one understand what is going on but do nothing to help fix the problem. As the last poster said, we play a soft defense that does not create pressure from the outset. This allows other teams to gain confidence, even in our stadium. CWM's initial plan is always the same- "Bend, don't break." The top teams of late, Florida, Alabama & LSU (with Pellini), attack from the first play on. By the 2nd &
4th quarter their opponents are dazed, confused and frustrated. We, on the other hand, are always trying to hang on and win one for Willie. Don't be confused by a bunch of stats- our record vs top teams for the past couple of years shows the most meaningful statistic of all- INCREASING NUMBERS OF LOSSES IN BIG GAMES. CWM may know how to coach but until he changes his philosophy we will always come second in the big games.

Todd said...

I want to say, how nice was it to actually see a db (Rambo) make a break on a ball! Didn't think I would ever see a db at Georgia do that again.

I wonder how many points have been scored on Willie's Defense in the first 2 series of the game.

My biggest problem with Bobo is play selection with the wrong personel. I know you can't become predictable, but come on. Samuel on a toss sweep to the short side of the field does not work.

I believe talent and coaching is the problem on D. On offense, I don't think Georgia has a really good rb and Joe Cox will continue to get raped by dbs if keeps eyeing the wrs down.

David said...

Numbers. Blah. All I know is that our defense looks lost most of the time and our offense is streaky. That is youth or coaching. I think it is coaching. I don't think that UGA will ever win the MNC with the coaching staff in it's current configuration. What needs to be done? Hell I don't know. UGA just looks bad/average to me. With the talent in Athens right now that is unacceptable.

JRL said...

I have to agree with the last David to post. How many of our coaches would be hired by top 15 programs or even in the SEC.

Does anybody think Bobo or WM could get the same position at LSU, Fla, Bama or even Tenn?

I really like CMR but after him what do we have...............?

Anonymous said...

Caleb King can make the most beautiful 4 foot run

Anonymous said...

Personally I don't think any of the stats mean anything at this point and this is why:

We haven't done a dam thing consistantly, meaning the stats are all out of place.

With that said, I think there is waaay too much talent on this team to be so bad. All of the excuses we have, guys can't learn the system, this guy is hurt, etc. Our recruiting has been outstanding since 1999. Where are all of the studs at?

JRL - Florida HAS played a team worth a damn - Tennessee. They hung in with Florida and will probabaly take joe cox hostage and steal 4-6 turnovers from us with that defense.

Zach said...

David,

Some great stuff for sure, and I'm a numbers guy. I'm glad you posted the comparisons of certain metrics for UGA vs other teams, without a benchmark metrics and statistics are worthless.

it would be interesting to compare the defensive productivity (yards given up, points allowed) based on the time (clock used) of the most previous offensive drive.

The idea behind this is a well rested defense should perform better, and this is more heavily correlated with execution as opposed to defensive game-planning.

if there is enough data, we could further seperate these numbers by quarter and analyze.

So a table would possibly look like this:

Time of Previous Off Drive - Yds Allowed - Pts Allowed

What do you think? I think if something like this were possible to put together, It might provide some valuable insight. Say the table shows that productivity is the same across times then that might lead us to believe that gameplanning is more of a culprit then execution.

Xon said...

I'm having trouble seeing how these offensive numbers show that Bobo is doing something wrong.

We throw more on first down on other teams. Okay, and we also are averaging far more yards per first down play. !! But we want to complain that this doesn't enable us to "establish the running game." It seems like a cliche is displacing actual productivity at this point.

JRL said...
This comment has been removed by the author.
JRL said...

Zach & Xon I think I can address both issues with one comment - our offense doesn't keep the ball long enough to allow the defense to rest. Either a turnover - a quick strike (TD) - a blown special teams play - or 3 & out puts them back on the field - usually on the short side of the field.

Xon said...

JRL,

Thanks for the interaction. I don't think you can blend, and I imagine no coaches would want to blend, all those things together into one "problem". Quick strike TDs are great. They put fear in the opposing defense and undermine their morale, they reduce the chances for mistakes by your offense b/c you run less plays, and so many things can go wrong with your execution on each play you run (as we too painfully know as Georgia fans this year). And, of course, they give you POINTS. Using "win probability" kinds of analysis, you always have a better chance of winning the game with seven more points and one minute off the clock than you did before you scored. That is obvious, but it bears repeating because it illustrates that quick strike TDs are not problems, at all. They are not like turnovers and kick coverage problems. They do not put the defense on a short field, etc.

As to the defense getting rested, sure that is nice but I think any decent cost-benefit analysis would have to conclude that it is worth it to have seven more points after a quick-strike drive, even if it means the D is a little more tired than they would have otherwise been when they start their next series.

Also, getting specific with these 1st down numbers. These other great teams that had great performances against serious competition averaged 5.8 yards on first down. Not too shabby!

Georgia is averaging 7.3 yards per play on first down this year. That is juggernaut, Big-12-eat-your-heart-out production. That has to be #1 in the country. It certainly would be #1 in the land simply on a per play basis over the course of a season, so it must surely be tops for first down plays only.

We are averaging 1.5 yards more PER PLAY on first down this year than those other teams. So Bobo should run the ball more why, exactly?

I promise I am not a WAC fan who got lost on one of America's highways and ended up rooting for an SEC team. I am not claiming that all proverbial coaching wisdom is wrong. I am not saying that D is more important than O, or that running the ball well doesn't matter. I'm just saying that, when it comes to proper "balance" and offensive production, 7.3 yards per play means you are doing something right, not wrong (and that is actually the understatement of the year).

What is an extra 1.5 yards per play worth? (And remember this is a comparison against other good teams who are doing well on 1st down). I am skeptical that "getting in a groove" is nearly as important in the running game as people often make it out to be, but even if it is, we are averaging about 5 yards on the ground so again what is the problem? But even if we are somehow hamstringing our running backs with out playcalling, maybe it is worth it since we are, as an offense, moving the ball at a highly efficient rate.

Highly efficient rate, that is, when we don't turn it over. But again, that number boggles my mind. For an extra 1.5 yards per play on first down, maybe that's why we have more turnovers. But maybe our aggression is worth it, too. If our D could just start getting more takeaways, then I'll live with a higher-than-average turnover problem on our O if it keeps moving the ball like it has been.

JRL said...

Xon your argument is persuasive to a point :-) However on the other side of the argument your quick strike / put the defense back on the field approach results in shoot outs like we have experienced. When that happens frequently the last team with the ball wins.

Twice this year we have nearly lost games because we couldn't stop our opponent late in the fourth quarter with our gassed defense on the field.

I think we will know exactly what we have after the LSU and Tenn games.

Charles said...

I'm not quite ready to give credit to Willie Martinez for making adjustments. Another possible reason Georgia's defensive performance has increased as the game moves on (particularly in the 4th) is superior talent and depth. As players tire through the course of the game, that talent and depth becomes a major advantage over teams with less depth (like south carolina, Ok state, Ark) and thus is likely one reason our D does better in the 4th

David Jones said...

JRL said: "Twice this year we have nearly lost games because we couldn't stop our opponent late in the fourth quarter with our gassed defense on the field."

Honestly, the opposite is true. In our shootouts, our better conditioning and depth has resulted in better production in the 4th quarter.

This was especially obvious in the Arkansas game. As they wore down, our D allowed Arkansas nothing (not even the FG!).

Xon said...

JRL, thanks for the courteous response. I think we have to be more careful about mixing cause and effect. Quick strike TDs do not cause shootouts. Your D giving up quick strike TDs of its own causes shootouts. There really are lots of times where a prolific offense destroys the other team with lots of quick strike TDs, and at the same time the defense plays well and prevents the other team from scoring nearly as much. This is how many 52-17 scores happen, isn't it? Those scores do happen, afterall.