Guernica in Maize

Submitted by MCalibur on

Tales from the Dorkside: Guernica in Maize

[Editor's note: bumped. At this rate I'm going to be a spectator around these parts soon.]

Herein lies data. For those readers who prefer to skip my right brained musings in a tenacious fit to resist all culture and proceed directly to the left-brained portion of the show proceed to the So, How Goes It? section. Ahem…

The fallout from Michigan’s catastrophic failure against Illinois has left in its wake a fan base wretched in suffering. And anger. And chaos. And despair. A veritable Guernica in Maize. Pablo Picasso’s renowned painting might as well have been painted in the aftermath of last Saturday’s loss. The centerpiece of the painting features Michigan’s Defense (the horse) in the throes of death complete with Juice Williams as javelin gashing it right up the middle, exposing the gaping wound that is Michigan’s defensive barracks. 

All of the major players are shown:

  • Terrorized souls engulfed in the inferno of buyer’s remorse (far right).
  • Horrified and confused onlookers (center right).
  • Dismembered soldiers , also known as The Legend of Tate Forcier: Heisman Freshman ;complete with shattered sword (bottom).
  • Grieving mother clutching the lifeless corpse of her child (read.: hope; far left).

Even the Eye of Mordor (read: FreeP) is represented (top). Not to mention that weird looking bull thing with fire coming out of it’s butt (left). I guess that’s Brian?

Guernica

Anyway, Such a scene makes the reasonable observer wonder—what is up the suck? Misopogon has thoroughly sifted through the immediately obvious symptoms of poor defensive play and walk-on starters to provide tremendous insight into the plight of the defense. He has emphatically demonstrated the task Rich Rodriguez and his man Greg Robinson have in front of them if they are to their save their jobs and save Ann Arbor from burning: fix the defense. Accomplishing this will not be easy and it will test Rodriguez’s mettle as a head coach. And it will take time.

So how goes it?

I think reasonable people would agree that it’s not yet time to render a final verdict…at least as far as the defense is concerned. So let’s focus on what is reasonable to evaluate Rodriguez on at this point in time: offensive production.  He’s had ample time to demonstrate core competencies in his area of expertise. He’s recruited his guys, has a reasonable amount of talent depth (inexperienced or not), and has had a reasonable amount of time to install his system.

Benchmark


The prototype I’m using as the model of what the performance of what a good offense should be will be the unit RR replaced, 2007 Michigan. That team had the requisite talent and experience at every single position:  an offensive line that featured two three time lettermen (Jake Long- RS Sr.  and Adam Kraus-RS Sr.), a three time letterman at QB (Chad Henne, Sr), a three time letterman at RB (Mike Hart, Sr), and three 2-time letterman at WR (Mario Manningham, Jr; Adrian Arrington, Sr; Greg Mathews, So). That’s as good a squad that a coach can ask for.

While the schemes employed by that offense are drastically different from what is currently being used at Michigan, the differences are irrelevant. Either is suitable for executing the mission: move the ball down the field and score points.

For the sake of thoroughness, I’ll stack them up against 2006 Michigan as well. Largely the same cast of characters but with fewer injuries. Reasonable or not, this level of production is what all Michigan fans desire or expect.

Performance Metrics


To evaluate the units I’m turning to very basic and universal categories.

Plays per Drive

This is a tempo-neutral possession metric. Evaluating Rich Rodriguez’s offense by time of possession is misleading since his philosophy is explicitly unconcerned with that metric. However, all offensive schemes seek to run as many plays as they can until they score.  So, this metric also allows us to evaluate execution at a base level as well. Plays-per-drive allows us to compare different schemes to each other.

The calculation of average and standard deviation for this metric omits the highest (yellow) and lowest (red) game averages since yards per drive are highly correlated with the strength of the opposing defense. The presumption here is that one good or bad game is a fluke. Games against markedly inferior competition (blue) have been omitted regardless of game outcome. Ahem.

What we see here is that Michigan 2009 has in fact improved over 2008 in this particular metric both in average plays per drive as well as in the standard deviation of this metric. However, 2009 lags 2006 and 2007 a little in regards to average but matches the 2006 campaign in terms of consistency. The average part is not very surprising.

The benchmarks have significant advantages over 2009 in terms of personnel and experience. However, the consistency part is a bit of a surprise. This year’s team, freshmen QBs, botched snaps, and miscellaneous turnovers included is as consistent as the 2006 unit and more consistent than the 2007 unit. Anyone who has had to improve a process knows that you get rid of deviation first, and then you shift the mean. In this case, there is the good fortune of the mean shifting on its own via player maturity.

Yards per Play

This is a category of raw production.  This is more in line with offensive strategic objectives such as controlling field position, getting into scoring position, and so on. Again, the high, low, and inapplicable data points have been omitted from the calculations of average and standard deviation.



Through the games played so far, the 2009 offense has improved significantly over the 2008 team and matches the production of the 2007 team. It is also the most consistent offense captured.

Points per Drive


The bottom line. Is the offense pulling its weight in the “outscore your opponent” equation? Again, the high, low, and inapplicable data points have been omitted from the calculations of average and standard deviation.



Once again, through the games played so far, the 2009 offense has improved significantly over the 2008 team, which was consistently bad, and beats the production of the 2007 team in terms of drive average and consistency. 2009 lags 2006 in terms of average but again, 2006 is a stout benchmark.

The Takeaway

Despite its glaring and soul dong punching deficiencies, the 2009 offense stacks up surprisingly well to arguably the best offensive unit Michigan has seen in approximately two decades, probably more like four, and maybe even six. DECADES(!).  And significant low hanging fruit remains (turnovers).

Regardless, after games like last Saturday’s we are right to break out the compasses and maps and graphing calculators to reevaluate just where the heck are we, exactly?

Here's where we are:
  • Tate Forcier is a FRESHMAN who has played in EIGHT games.
  • The rest of the offense are de facto true sophomores who have only shown signs of effectiveness in about 14 games.
  • The defense does not have the breadth or depth of personnel necessary to meet the Michigan standard.

Recognizing that we have a major vulnerability in defensive personnel is in no way a slight against the Lloyd Carr stewardship. It is simple root cause diagnosis.  And, maybe RichRod can tweak a thing or two or three, here and there and over there. But, to suggest that the team has made no progress is simple ignorance at best and dubious ignorance at worst.

There is a big difference between excusing and explaining…that difference is responsibility. RichRod is responsible for his record, but its only fair to give him more time to hold him accountable as well. Forging the program into a consistent winner requires Rodriguez to demonstrate the full gamut of the requisite core competencies needed to be a successful chief executive in an elite college football program:  excellent recruiting, excellent motivating, and excellent personnel evaluation(coaches and players), and excellent focus.  If he succeeds, he will have vindicated Bill Martin decision and earned the respect of many.  If he wins it all, he will be the next Bo Schembechler.

Godspeed, RichRod. Godspeed.

[Editorial take: I don't think things are quite as sunny as the numbers suggest; in the comments it's noted that adjustments were not made for outliers like turnovers and special teams items. Michigan's gotten great production out of Olesnavage and Stonum this year. Also, Michigan has yet to face the #65, #21, and #6 defenses so far this year and will likely see their to-date respectable metrics continue to dip below the okay production of the 2006 and 2007 teams. The 2006 team was pretty good but only 38th in total offense and 26th in scoring. It may have been arguably the best collection of talent at Michigan, but it wasn't exactly set free to roam the plains, its majestic rippling muscles trampling over mascots that dare oppose it. Michigan is approaching the mediocre numbers put up by Mike DeBord.

Even considering that the progress made from year one to year two is obvious.]

Tales from the Dorkside
Illinois 2009

Comments

SFBlue

November 4th, 2009 at 1:33 AM ^

I think an interesting concept to try to quantify would be offensive efficiency. Part of what was so maddening about last week's game is that Michigan had 13 points to show for amassing 377 yards of total offense--which does not account for the fruitless red zone possession after the blocked punt. The Penn State game had some similarities; if Michigan scores a touchdown before halftime, when it's at the goal line, it's a new ball game. In some of last year's losses, this was a big issue (primarily because of turnovers, of course). Notre Dame, we put up 388 yards of total offense, but had only 17 points. Toledo, 10 points and 290 yards. Northwestern, 264 yards and 14 points (7 from a blocked punt). Surely, this could be done by using some combination of red zone efficiency and comparison of total yards of offense to points scored. My theory is that Rodriguez's teams, due primarily to turnovers but there may be other factors, have not (YET?) been as offensively efficient as, say, 2006/7 Michigan.

Joined: 11/02/2009

MGoPoints: 15699

koolaid

November 5th, 2009 at 10:46 PM ^

If we could only HOLD ON TO THE #$%$#% BALL, we may have a chance in some of those games. I sum up both this year and last year as: Freshmen QBs (only last year the QBs were even worse). Seriously... there is no position more important than QB in football. Last year we had massive problems at the QB position. This year we have two freshmen QBs that probably both needed a redshirt year.

Joined: 10/31/2009

MGoPoints: 1047

jokewood

November 4th, 2009 at 10:01 AM ^

I think you're taking total points scored and then dividing by offensive drives. However, not all of the points scored can be attributed to the offense. For example, you're crediting the Michigan offense for the Burgess pick-6 and the Woodley fumble return TD in the '06 Notre Dame game. The offense actually scored 2.4 ppd, not 3.4 ppd.

Joined: 07/30/2008

MGoPoints: 11

MCalibur

November 4th, 2009 at 11:17 AM ^

Yep, I thought about that. I went with the assumption that defensive and special teams scoring contribution get overwhelmed by offensive scoring contribution over the course of an entire season (really just a few games). All teams were treated the same way in this regard. I'll check this out to see if it makes meaningful difference.

Joined: 07/16/2009

MGoPoints: 2361

jokewood

November 4th, 2009 at 2:08 PM ^

I've calculated an adjusted offensive points per drive for Michigan as well as several other conference and national peers. My adjustments are... - only off. scores are counted (no def. or ST scores) - pick-6s / fumble return TDs are counted *against* the off. - overtime scoring is excluded (OT inflates off. stats) Also, I kept in the highs and lows. '09 -- 2.03 ppd (excludes DSU since they were a JV team) '08 -- 1.42 '07 -- 2.07 (includes ASU since they were not a JV team) '06 -- 2.15 '05 -- 2.12 '04 -- 2.16 '03 -- 2.65 (Navarre!) '02 -- 2.08 '01 -- 1.78 I'm guessing that after facing the Wisconsin and Ohio State defenses, our overall offensive efficiency will drop below 2 ppd.

Joined: 07/30/2008

MGoPoints: 11

blueloosh

November 4th, 2009 at 10:09 AM ^

The stats you use are great. And I understand removing Delaware St. Why remove any other data points? In yards per play and per possession you have removed performances against Iowa, Ohio State (twice) and Penn State. Do you not think our offensive performance against these teams is relevant? I would at least keep in worst performances, and would consider keeping in best as well. Those games matter, they count, they also represent the quality of the offense. Great work presenting though--thanks!

Joined: 10/26/2008

MGoPoints: 831

MCalibur

November 4th, 2009 at 11:13 AM ^

I did this in an effort to shed outliers both high and low without crushing the already small sample sizes. One great or terrible game can be caused by a variety of things. So I gave each team a freebie for bad games at the cost of their best performance. Also, If you factor in this year's EMU performance, you might make the conclusion that this year's squad is better than 2006 and 2007 in terms of yards per play. We all know better than that.

Joined: 07/16/2009

MGoPoints: 2361

blueloosh

November 4th, 2009 at 8:38 PM ^

I see your aim, and understand removing games where we obliterated a team so bad they are not competitive, but why are performances against strong defenses not useful data points? The 2007 OSU game is very relevant to me in determining whether we are headed in a better or worse direction on offense. I don't see how giving each year's team a mulligan renders purer statistics. But it is a minor quibble. Your work was excellent--thanks.

Joined: 10/26/2008

MGoPoints: 831

LesMilesismyhero

November 5th, 2009 at 2:28 PM ^

An outlier is a point that is not typical of the rest of the data. The reason an outlier would be removed from the data set prior to analysis is because mean and standard deviation are statistics that can be leveraged by atypical observations. Look at 2006 points per drive, 3.4 is removed because it is the high observation. If you look at the complete season, you will notice, this observation occurs 3 times in a 13 game season, meaning it was not unusual at all. In 2007 YPP UM's lowest value is 2.1 YPP below the second lowest observation. This is 2.25 standard deviations below the mean (if a new mean and sd are calculated excluding this point). Typically outliers should be investigated to determine if there is reason to believe that they came from a different process than the rest of the data. 2007 OSU was unusual, Henne was playing injured and was inaccurate, the weather didn't help, and Hart had nowhere to go against a defense loaded to stop the run.

Joined: 11/04/2009

MGoPoints: 273

LesMilesismyhero

November 5th, 2009 at 12:44 PM ^

Consider the use of a probability plot in your search for outliers as opposed to throwing out the high and the low sight unseen. Example 2007 Yards per play (YPP) Opp YPP j (j-0.5)/n OSU 1.2 1 0.04 PSU 3.3 2 0.12 MSU 3.5 3 0.19 ORE 3.6 4 0.27 NW 4.1 5 0.35 ILL 4.4 6 0.42 WIS 4.4 7 0.50 UND 4.9 8 0.58 UF 5.9 9 0.65 EMU 6.3 10 0.73 PUR 7.5 11 0.81 ASU 8 12 0.88 MINN 8.7 13 0.96 Then plot YPP as x variable and (j-0.5)/n as the y variable. Outliers can be identified on the chart by a large gap in the x direction from the nearest neighbor. This is a subjective call but when a chart is seen the OSU game will stand out like a sore thumb while all other data points fall in line.

Joined: 11/04/2009

MGoPoints: 273

Blue 8198

November 4th, 2009 at 12:44 PM ^

Thanks for the interesting analysis and giving us something hopeful to think about (much needed these days). I just hope the current year offensive stats don't take too much of a beating over the rest of the season - we do have coming up a couple of the better defenses M will see all year (buzzkill).

Joined: 01/22/2009

MGoPoints: 275

oc michigan fan

November 4th, 2009 at 1:25 PM ^

Great stuff! I would be interested to see how these 4 years of Michigan offense stack up to a Pat White led, WV offense. Any chance of picking one of White's seasons and running the numbers?

Joined: 12/06/2008

MGoPoints: 295

jokewood

November 4th, 2009 at 9:30 PM ^

West Virginia '09 -- 2.27 (likely to fall w/ Pitt, Cinci) '08 -- 1.90 '07 -- 2.96 '06 -- 3.27 '05 -- 2.33 '04 -- 2.13 '03 -- 1.98 '02 -- 2.03 Michigan '09 -- 2.03 (likely to fall w/ Wisc, OSU) '08 -- 1.42 '07 -- 2.07 '06 -- 2.15 '05 -- 2.12 '04 -- 2.16 '03 -- 2.65 '02 -- 2.08 '01 -- 1.78 I use a slightly different methodology... - all FBS games included - only offensive scores (no def or ST) - no over-time - pick-6s and off fumbles returned for TDs are scored negatively against the off

Joined: 07/30/2008

MGoPoints: 11

MileHighWolverine

November 4th, 2009 at 4:40 PM ^

- Tate has a bum shoulder and suffered a concussion mid season - Tate is from San Diego and needs time to adjust to cold weather football (i.e. Odoms last year, IIRC) Both of these issues should be mitigated next year.

Joined: 09/02/2008

MGoPoints: 7749

turbo cool

November 4th, 2009 at 8:09 PM ^

brian, you gotta love your job. If thousands of people followed my work everyday and eventually started doing it for me i'd be thrilled to say the least. That is awesome. +1 to everyone (except bouje, obviously).

Joined: 07/10/2008

MGoPoints: 977

Big Boutros

November 4th, 2009 at 8:14 PM ^

A small part of me wants to craft a diary so piss poor in its execution that I form a metafictional opposing regime to the recent surge in charttastic professionalism around here "Ass: Why Michigan is the Shit Fo Realz" a Treatise by Big Boutros

Joined: 07/08/2008

MGoPoints: 23338

G Money

November 4th, 2009 at 9:12 PM ^

While you stated that it's Time of Possession neutral, different offenses have different philosophies. Lloydball (and I don't use that negatively) would mandate that if you have a good defense, play more ball control, NOT GIVE UP the ball, and score). You may not try to get the ball downfield as fast as you can. Also, when you are playing with the lead, many coaches (not just LC) place emphasis on not doing something stupid... like GIVING UP THE BALL. Conversely, when you are playing from behind (like the past 2 years), you are cutting tooth and nail to get up field. You may also turnover the ball more. Agree with looking at who you are racking up the numbers on. Obviously, the EMU's and Delaware State's of the world pad the stat book, but in the end, you really aren't in any risk of losing to these teams. Good review.

Joined: 11/27/2008

MGoPoints: 73

Ryan

November 4th, 2009 at 9:45 PM ^

Great diary, though I did feel a bit ashamed in understanding the nerd talk (ie, process improvement, root-cause analysis, core competencies.)

Joined: 06/30/2008

MGoPoints: 147

Senator Bluetarsky

November 4th, 2009 at 10:47 PM ^

Well at least you did manage to capture our angst with Picasso's Guernica. Bill Martin's departure though is the most immediately frightening yet under-analyzed aspect of the football canvas. Rodriguez ultimately should bear unto us Penelope Cruz pre-Vicky Christina yadda yadda given, as you mention, enough time. But what is it we shall experience with Martin's successor? Franco, Torquemada, Javier Bardem . . . this thought is what gives me the nightmares.

Joined: 11/04/2009

MGoPoints: -580

Njia

November 4th, 2009 at 11:00 PM ^

Of just how much more cultured UM fandom is than its other chief rivals, (with the possible exception of ND). I daresay that there is not another sports fan blog in the entire known universe that boasts a reference to Guernica, much less turns the monochromatic pastel from hues of navy blue, black and white to Maize. Nicely done.

Joined: 09/15/2009

MGoPoints: 35773

SeattleChris

November 4th, 2009 at 11:07 PM ^

First of all great work in putting this together, however as any good process improvement practitioner would do you have to examine the outliers to determine what produced them; you might want to reference opposing defense's ypg and/or point yield for those that you removed and do the second layer of analysis to determine if RRod's o provided more big plays vs. better defenses with less experience. That analysis might give some insight into how the M offense will evolve next year with more experience/competence at key positions. If you believe, as I do, in the hypothesis that "schematic advantage" + "experience" = more wins when isolating uncontrollable variables such as turnover margin and that "schematic advantage" can be expressed by delivering more big plays against stingier defenses. For example, good coaching creates greater variability versus defenses that keep overall variability in their process down(e.g. yard per play of PSU/OSU). My point is that despite getting our asses handed to us by good defensive teams this year, the fact that we can schematically gash these teams once or twice this year is a harbinger of good things to come in 2010/2011 because the null hypothesis only applies to Juice Willams. Is this really the case? I don't know, I don't have time to look at the data, but I suggest You, the Mathlete and Misopogon should join forces and analyze the shit out of this with real-time dashboards, scorecards and the like and become the Ernie Adams to Coach Rod and regail us all with tales of your victory.... Well done!

Joined: 09/06/2008

MGoPoints: 281

funkywolve

November 4th, 2009 at 11:22 PM ^

I wouldn't even consider the 2007 offense the best offense of this decade. Henne, Hart, Long, etc. sounds good but how many games did Henne miss? At least 2 off the top of my head - ND and Wisky, and he probably shouldn't have played the whole game against OSU. How many games did Hart miss? Missed the Minny game, started against Wisky but then didn't even really play after the first series or two. I'd probably rate both the 2000 and 2003 offenses higher then I would the 2007 offense. A couple questions: when you are looking at drives, what did you do about drives right before the end of the half where a team is just running out the clock? I'm not talking taking a knee, but rather first and 10 on your own 25 with say 40 seconds left and they run the ball twice? I don't really get the plays/drive? You could start at midfield run 5 plays for 12 yds and punt - a total waste of field position. Conversely, you could start on your 1, run 5 plays for 37 yds and punt. The latter is actually a solid possesstion that with a decent punt is going to change the field position a lot. Also, with plays/drive last year they had 4.8 plays/drive against OSU but only 4.6 plays/drive against Purdue. Similiar numbers but yet they scored 35 more points against Purdue.

Joined: 10/08/2008

MGoPoints: 20184

tomhagan

November 5th, 2009 at 12:24 AM ^

Wow... just...wow... Original poster really needs to be called out on this statement: "Despite its glaring and soul dong punching deficiencies, the 2009 offense stacks up surprisingly well to arguably the best offensive unit Michigan has seen in approximately two decades, probably more like four, and maybe even six. DECADES(!). " Uh....yeah. Well...that holds water until you look at what has happened vs. Big Ten opponents... where all numbers show this offense to be PATHETIC at best. So...yeah...they can roll vs. MAC teams, a weak ND defense and a non FCS team. Wow!. Oh boy.

Joined: 08/30/2008

MGoPoints: 105

Medic

November 5th, 2009 at 10:47 AM ^

The metric shows what most people have witnessed. The offense *can really* move the ball well most games, even against B10 opponents. The shackles come in the form of turnovers, dropped snaps, and dropped passes. Did you actually look at the numbers vs B10 opponents? They actually looked better than I thought they would.

Joined: 10/24/2008

MGoPoints: 2163

RagingBean

November 5th, 2009 at 12:42 AM ^

The thing I most took away from this? That photo of Long, Henne, and Hart. I can't believe it's only been 2 years since those 3 were out on the field in Ann Arbor. It seems like ages. But, yeah, good diary, and I am still thoroughly convinced that Michigan will be a behemoth in the college football world within 3 or 4 years. That will be more than worth the struggles of now.

Joined: 10/29/2008

MGoPoints: 1757

Blue in Seattle

November 5th, 2009 at 8:53 AM ^

The amount of analysis that has been pouring out after the Illinois game has been amazing, this is another great example. I'm trying to locate Brian's post that mentioned something about the birth of a new era, because that does have special meaning for me. Each Saturday I have to balance watching my DVR'd game between the needs of a 2 year old son and a 6 week old son. I had to wait an especially long time to watch the Illinois game, since of course this was my toddler's first Halloween. Oh the agony of anticipation. While watching that game I switched from drinking beer to rum to make it all the way through, on Sunday I woke up to find that MGoBlog was shut down to prevent an angst explosion, and felt a lot like this video. Unlike Brian's Morrissey metaphor, I'm not explaining mine, it just helps me getting through the next 3 weeks. It's my hope that it can help others through the birth pangs, infant feedings, and toddler arguments of the new Michigan Football Era. thank you MGoBlog for being here.

Joined: 07/02/2008

MGoPoints: 1716

marathon95

November 5th, 2009 at 9:29 AM ^

This is good analysis, but it doesn't take into account the fact the prior years were complete and current year is incomplete. Schedules are not equally weighted 1st to last, i.e., out of conference vs in conference. You should either compare the 1st 8 games of the prior years to the this year or wait until the end of year.

Joined: 11/05/2008

MGoPoints: 169

jlvanals

November 5th, 2009 at 10:15 AM ^

is that Chad Henne, Mike Hart, et al were injured for a significant portion of that season. Ergo, instead of a 4 year starting quarterback you had a half-cocked Texan (Mallett) running the offense. Other than that, the comparison is interesting, but the difference between Henne and Mallett is considerable and, I would assume, statistically significant.

Joined: 09/21/2009

MGoPoints: 707

stubob

November 5th, 2009 at 10:29 AM ^

-1 for questionable comparison of college football to the Spanish Civil war. I know I'll get modded "Lighten up, Francis" for this, but it's in questionable taste. For a slightly less insulting analogy, I present Lord of the Rings. The Good Guys: RichRod as Gandalf. We've already seen him in a wizard hat. He's a natural for it. Lloyd Carr as Tom Bombadil, or Sir Not-Appearing-in-this-Story. Tate Forcier as Frodo. Perfect casting. Which means Martavious Odoms as Samwise Gamgee, Frodo's right-hand man. Or left hand, if Tate's a lefty. Zoltan Mesko as Legolas. Immortal, and capable of killing an Olyphant with a well-placed arrow. Which means Brandon Minor as Gimli. Why use an arrow when you've got a big-ass axe? And lastly, Denard Robinson as Faramir. May be a great leader, may get you killed. The Bad Guys: Steve Tressel as Sauron. All-seeing, all-knowing, all-evil. Justin Boren/Ryan Mallett et.al as the Ringwraiths. Drawn by the lure of power into an evil trap. The Detroit Free Press as Sarumon the White. Works against our heroes until overthrown by the White Wizard. Mark Dantonio as Gollum. Not really a bad person, but obsessed with the Precious. Charlie Weis as the Balrog. Turned Gandalf the Grey into Gandalf the White, so that's good. Ohio State as the Uruk-hai. Anybody I missed?

Joined: 08/20/2008

MGoPoints: 1553

Ernis

November 5th, 2009 at 10:34 AM ^

The quality of an analysis is dependent upon the quality of the experimental design The fact that our current offense stacks up fairly well against 2006 & 2007 using these metrics is less an indication of the quality of our current offense, and more an indication of the limitations of these metrics Aside from the score at the end of the game, the most meaningful offensive metric is 3rd-down conversions. The two metrics with which I am most skeptical are "Yards per Drive" and "Points per Drive" .... these are conceptually HUGE metrics, under which a vast array of confounders reside. Also, the outliers for these two metrics appear to be causally linked with the outcome of interest. They do not appear to be "fluke" games; rather, they are poor or solid performances against solid or poor teams, respectively. I like the first metric, though, that is pretty interesting. Finally, I think a major limitation of this study is that the current season is not yet over. As we have seen, the offense in the first 4 games is functionally distinct from the offense of all games thereafter (DSU rightfully excluded). I am skeptical that our offense will be able to "turn it around" and reach their prior output for the final three games, given the difficulty of the opponents faced.

Joined: 09/23/2008

MGoPoints: 11437

Gene

November 5th, 2009 at 10:45 AM ^

Oh, the 2007 offense was AWESOME on paper. But between persistent injuries, brain-dead play calling, and the general bloody-mindedness of the universe, it mostly meandered between competent and out-right horrible (see OSU game,) only once reaching it's true potential (against Florida in the bowl game.) So I wouldn't consider being statistically better than that offense to be much of an accomplishment.

Joined: 07/01/2008

MGoPoints: 443