The Michigan Difference - Wisconsin Edition
Here's this week's update to The Michigan Difference, updated with stats from this week's games.
Another bipolar game against Wisconsin. The final offensive output was pretty good, but the defense couldn't stand up to their rushing attack. We remain #5 in Total Offense (TO) and are now #112 in Total Defense (TD).
Disclaimer: The NCAA stats are not linear, of course, and a difference of 1 yd/gm can be a large or small difference in rankings depending on how closely spaced everyone is. So as I cautioned, this isn't a hard-core statistical exercise. This analysis is pretty one-dimensional because it's long and complicated enough as it is.
I think the greatest value in this is to look back at the early games and see how well we did in comparison to what other teams ended up doing against them - what seemed like a good or bad performance at the time may look different in retrospect.
Part the First: Offense
We know our offense is great, but what kind of damage has it done to the Total Defense (TD) ratings of our opponents? Here they are thus far:
Opponent | Games | Yards Yielded | Yds/gm | NCAA Rank |
Connecticut | 10 | 3512 | 351.20 | 44 |
Notre Dame | 11 | 3977 | 361.55 | 55 |
Bowling Green | 11 | 4742 | 431.09 | 104 |
Indiana | 11 | 4612 | 419.27 | 94 |
Michigan State | 11 | 3657 | 332.45 | 28 |
Iowa | 11 | 3423 | 311.18 | 14 |
Penn State | 11 | 3899 | 354.45 | 48 |
Illinois | 11 | 3766 | 342.36 | 35 |
Purdue | 11 | 4025 | 365.91 | 58 |
Wisconsin | 11 | 3598 | 327.09 | 26 |
What would these guys' defensive stats look like if they hadn't played Michigan?
Opponent | Total Offense, M |
Opp. Avg - M, |
M Total Offense, |
NCAA Rank |
Connecticut | 473 | 337.67 | 140% | 32 |
Notre Dame | 532 | 344.50 | 154% | 40 |
Bowling Green | 721 | 402.10 | 179% | 84 |
Indiana | 574 | 403.80 | 142% | 86 |
Michigan State | 377 | 328.00 | 115% | 27 |
Iowa | 522 | 290.10 | 180% | 6 |
Penn State | 423 | 347.60 | 122% | 44 |
Illinois | 676 | 309.00 | 219% | 13 |
Purdue | 395 | 363.00 | 109% | 57 |
Wisconsin | 442 | 315.60 | 140% | 16 |
*Opponents' average Total Defense yards per game, minus the Michigan game
**Michigan's Total Offense in game as a % of the opponent's average TD minus the Michigan game
Iowa, Illinois, and Wisconsin's defenses really wish they hadn't played us. They'd be in the top 20 nationally but for one game. Michigan has gained above our opponents' average yardage yielded in every game thus far, and their Total Defense rankings have suffered as a result. What's the damage?
Opponent | TD Rank With M | TD Rank Without M | Difference |
Connecticut | 44 | 32 | -12 |
Notre Dame | 55 | 40 | -15 |
Bowling Green | 104 | 84 | -20 |
Indiana | 94 | 86 | -8 |
Michigan State | 28 | 27 | -1 |
Iowa | 14 | 6 | -8 |
Penn State | 48 | 44 | -4 |
Illinois | 35 | 13 | -22 |
Purdue | 58 | 57 | -1 |
Wisconsin | 26 | 16 | -10 |
Average change in Total Defense ranking for all opponents: -10.1 places.
Looking at the offensive performance versus the quality of the defense:
There is little correlation between Michigan's Total Offense for a game and their opponent's average Total Defense (minus M). Whatever is limiting our offense's output in a game, it is not directly related to the number of yards the opponent usually gives up. This would suggest that the offense tends to be limited by itself, rather than the opponent.
Part the Second, Defense
So the flipside of this, then, is how much has our defensive suckitude helped out our opponents stat sheet? Where would they rank in Total Offense without having played us? We'll run the same tables again, but from the opposite tack:
Opponent | Games | Yards Gained | Yds/gm | NCAA Rank |
Connecticut | 10 | 3311 | 331.10 | 95 |
Notre Dame | 11 | 4243 | 385.73 | 58 |
Bowling Green | 11 | 3137 | 285.18 | 114 |
Indiana | 11 | 4247 | 386.09 | 57 |
Michigan State | 11 | 4549 | 413.55 | 37 |
Iowa | 11 | 4335 | 394.09 | 51 |
Penn State | 11 | 4093 | 372.09 | 66 |
Illinois | 11 | 4230 | 384.55 | 61 |
Purdue | 11 | 3429 | 311.73 | 102 |
Wisconsin | 11 | 4843 | 440.27 | 23 |
Wisconsin is easily the strongest offensive team we've faced thus far. The results of the game show that. MSU was pretty good, the rest varying degrees of average to bad.
Opponent |
Total Defense, M |
Opp. Avg - M, Yds/gm* |
Opp Total Offense, % of Opp Avg - M** |
NCAA Rank Without M |
Connecticut | 343 | 329.78 | 104% | 96 |
Notre Dame | 535 | 370.80 | 144% | 69 |
Bowling Green | 283 | 285.40 | 99% | 114 |
Indiana | 568 | 367.90 | 154% | 71 |
Michigan State | 536 | 401.30 | 134% | 47 |
Iowa | 383 | 395.20 | 97% | 51 |
Penn State | 435 | 365.80 | 119% | 72 |
Illinois | 561 | 366.90 | 152% | 70 |
Purdue | 256 | 317.30 | 81% | 101 |
Wisconsin | 558 | 428.50 | 130% | 32 |
* Opponents' average Total Offensive performance, minus the Michigan game
** Opponents' Total Offense as a percentage of their average offensive performance, minus the Michigan game
Here's a nifty graph of our opponents' Total Offense against Michigan, versus their average Total Offense per game without the Michigan game:
In this case, we do have a reasonably good correlation. Our defense does worse against better offenses. That would suggest that we're talent-limited somewhere (either coaches or players) and the opponents' offenses tend to have their way with us. In other words, our defense doesn't shut anybody down. The more yards our opponents average per game, the better they'll do against us.
To summarize:
Opponent | TO Rank With M | TO Rank Without M | Difference |
Connecticut | 95 | 96 | +1 |
Notre Dame | 58 | 69 | +11 |
Bowling Green | 114 | 114 | 0 |
Indiana | 57 | 71 | +14 |
Michigan State | 37 | 47 | +10 |
Iowa | 51 | 51 | 0 |
Penn State | 66 | 72 | +6 |
Illinois | 61 | 70 | +9 |
Purdue | 102 | 101 | -1 |
Wisconsin | 23 | 32 | +9 |
Average boost to opponents' Total Offense NCAA ranking: +5.9 places
From this perspective, the Wisconsin game was our 4th worst defensive performance of the year. As bad as we looked, three other games were worse. We were up against a very good offense, and it showed.
Part the Third: Summary
Opponent |
Michigan's O Difference on Opp TD Ranking |
Michigan's D Difference on Opp TO Ranking |
|
Connecticut | -12 | +1 | W: Good O, OK D |
Notre Dame | -15 | +11 | W: Good O, Terrible D |
Bowling Green | -20 | 0 | W: Awesome O, OK D |
Indiana | -8 | +14 | W: Good O, Terrible D |
Michigan State | -1 | +10 | L: OK O, Terrible D |
Iowa | -8 | 0 | L: Good O, OK D |
Penn State | -4 | +6 | L:Good O, Bad D |
Illinois | -22 | +9 | W:, Awesome O, Terrible D |
Purdue | -1 | --1 | W: OK O, OK D |
Wisconsin | -10 | +9 | L: Good O, Terrible D |
In subtly maize-and-blue graphical form:
New observations for this week:
-
Many of our previous opponents had good weeks offensively, making our defense look a bit
betterless bad in those previous games. - Wisconsin is easily the best team we've faced yet. Offensive and defensive performances were close to mid-pack, but we got our butts kicked.
- Our offense remains impressive and will keep getting better.
- Our defense is terrible and had better get a lot better.
- Winning is still a lot more fun than losing.
November 23rd, 2010 at 7:59 PM ^
Wow, this is very interesting and very telling. Thank you for doing this.
It's crazy to think how much Michigan's performance hurt Illinois' ranking... I know ther was OT involved, but still interesting nonetheless.
I'm going to assume we will give up somewhere between 110-120% of Ohio State's normal offensive performance. :(
November 23rd, 2010 at 8:10 PM ^
Ohio State averages 446.18 ypg. They should get 2.19(446.18) - 347.22 yards = 630 yards of TO. Which is 142% of their average. :-(
If we hold them to 110-120% of average, it will be a good day, comparatively.
November 23rd, 2010 at 11:34 PM ^
But never underestimate the boring, slow-paced and clock-eating stylings of Tresselball. If it is employed, I could see OSU yardage be closer to 100% their average.
November 24th, 2010 at 7:22 AM ^
Agreed. If osu gets up early by a bunch, the clock will be eaten up by the sweater.
November 24th, 2010 at 9:23 AM ^
Sample size is a bit small to be trying to draw conclusions from a regression analysis, wouldn't you say?
Interesting observations, though
November 24th, 2010 at 10:51 AM ^
10 points is twice as many as I usually use. I'm not pretending this is mathematically rigorous, anyway.
November 24th, 2010 at 1:30 PM ^
Fair enough
Comments