mistersuits

This week is an abbreviated session due to being out of town for a wedding.

MSU Post-mortem

Blargh! That is all.

The numbers predicted 35.1 points and 536 yards for Michigan State vs Michigan.

MSU had 34 points and... 536 yards.

Yikes. Not enough weeks of predictions to conclude a meaningful predictive trend, but still frightening.

Notes:

• As predicted, turnovers were the critical difference as Michigan left anywhere from 18 to 21 points on the field in losing the turnover battle.
• Special teams didn't cost us the game, because the game was never close enough to be lost by special teams.
• MSU rolled on us exactly as we would have expected.
• Michigan's offense slowed down exactly as we feared.
• Prediction wise, Michigan should have an offensive multiplier less than 100% against all remaining opponents until they can prove they can play otherwise vs. Big Ten opponents.
• Defensively speaking, there's no need to modify how we are predicting our opponents' scoring/yardage at this time.
• Without a way to predict turnovers, it appears these scoring predictions are only as good as the game remains turnover-neutral.

How about the rest of our opponents?

Chart of Offensive Expectations (through 6 weeks)

Rank	Opponent	N-PPG	N-YPG	SoS
1	osu	36.8	397	63.45
2	Michigan	34.6	495	68.98
3	Wisconsin	31.8	385	63.59
4	MSU	31.3	414	65.15
5	Connecticut	29.2	330	66.17
6	Iowa	28.9	367	64.08
7	Indiana	27.3	328	60.07
8	BGSU	24.9	289	68.10
9	Notre Dame	24.3	392	74.42
10	Illinois	23.9	346	72.90
11	UMass	23.1	349	58.02
12	Purdue	19.9	330	69.08
13	Penn State	17.3	319	70.54

Metrics

N-PPG or Normalized Points-per-game is taken from the teams average PPG with a SoS multiplier factored in to deflate numbers from playing bad competition and inflate numbers based on playing good competition.

N-YPG or Normalized Yards-per-game is calculated using the same SoS multiplier as N-PPG but using this metric will help us determine a less variant guess as to how offenses will perform (PPG is subject to wild variance based on turnovers and special teams).

Strength of Schedule is taken from Sagarin rankings.

Usage: The chart doesn't predict that #3 would beat #5. Instead it tries to predict with the most accuracy how many points/yards on average each of these teams would score against a common opponent.

Results

The Big Ten's collective strength of schedule takes a leap up and tightens. A lot of the numbers on this chart shift deceptively as a result.

• Penn St. continues to struggle mightily [Ed-M: Penn St. does everything mightily /PSU fan]. Purdue also did not impress.
• Wisconsin and MSU have separated themselves from the middle of the pack and established themselves.
• Iowa didn't play and suffered from some SoS modifications, but really we don't find out about them until after this week. Awesome.
• Illinois got a big boost this week and goes from a team we should definitely beat to a team we have to be concerned about.
• Our out-of-conference slate continues to weaken as the weeks go on. This trends us towards a 6-8 win season rather than 8-10 wins.

Conclusions Based on Almost Enough Data

Like last week, I am giving our opponents 125% of their N-PPG and 150% of their N-YPG for predictions vs Michigan.

This is what I said a week ago:

Last year, this is where Michigan's offense fell off a cliff. The last seven Big10 games they averaged 20.1 PPG. They did not outgain any of their opponents and they lost the turnover battle nearly every time.

Ugh. I wanted so badly to say that this was going to be a very different year than 2009, but now through six games I can't ignore the parallels.

Also, I need to define a scoring range based on turnover margins.

Best Case:

Michigan's new best-case scenario (Michigan offense operates at or near 100% N-PPG and opponents score 125%) in the Big Ten is 5-3. Purdue, Illinois, Penn St. and... Iowa. Statistically, we must win this weekend in order to go above .500 in conference play.

Worst Case

In a worst-case scenario, we fall behind our two remaining @Purdue and @Penn St. These are two road games against opponents whose offenses are struggling and for whom Michigan's defense might be the perfect medicine. Illinois looks scarier than a week ago. Iowa and Wisconsin are both very intimidating. Until our offense proves it can execute at a high level in a Big Ten game not named Indiana, this is the direction we are trending.

The Truth!

Our new outlook ranges between 7-5 and 9-3!

Prediction for Iowa:

Everything that we did not want to happen last weekend, happened.

Based strictly on the numbers:

Dear god, help us.

It's about time for our defense to turn in a performance that exceeds expectations, and when these are your expectations one can only pray that they are exceeded.

If they are, Michigan might have a shot in this game - but only if our offense actually puts together a real game when it counts in the Big Ten.

Michigan 28
Iowa 38

GO BLUE!

Each week I trundle off to my favorite sports bar in Brooklyn, pumped up with expectations that far exceed what could ever possibly happen, promising myself I will behave in front of the other patrons, and that I'll remain calm when inevitably the other team scores or when Michigan goofs up.

And yet I was compelled to blurt out "Don't fumble it! Don't fumble it!" when Cam Gordon picked off Ben Chappell. The singular defensive highlight of the day and I respond as if Michigan had been the team committing a turnover on that play. I got some strange looks for sure.

That's how you know you're traumatized.

We still live under the spectre of the 2009 season and the reality that our defense is likely not going to stop any team we play.

The statistics tell me (Mathlete) that Michigan is absolutely going to win at least another couple Big10 games this year. I am resolutely impatient, however, and cannot wait until Illinois week in November to finally claim "improvement" from 2009. In fact if we lose the next three games - games we had already written off (Brian) no less - we'll be 5-3 and in crisis mode*.

A record of 7-5 was always the most likely outcome. But at 5-0, none of us could stand finishing the season 2-5, for so many reasons. So we reach a tipping point.

Win, and no one can ever claim again that this is 2009 all over again. Reclaim bowl eligibility, set the stage for a run at the Big10 title, and silence one of our most loathed rivals in one fell swoop.

Lose, and face the reality that our defense is going to limit us from getting over .500 in conference play, no matter how amazing and awesome Denard Robinson is.

*It's not really crisis mode when that's what we had as a baseline expectation, but it is the undeniable flaw of reaching 5-3 from 5-0 instead of 2-3.

But what do the numbers say?

Indiana Post-mortem

Last week I laid out a chart of our opponents and what kind of offensive output we can expect from each.

The numbers predicted a 36.3 (15.7% under) to 32.2 (8.7% under) Michigan victory - the margin of victory (4.1) was exactly correct. I extrapolated those considering likely real football scores and came up with a 42-31 prediction.

We had a turnover neutral game and special teams played no special role, so that levelled out any scoring variance, making these prediction about as accurate as they could be. Not bad for a first time, by the numbers prediction, all things considered.

Notes:

• Michigan's offense exceeded expectations, netting 80 yards over predicted.
• Michigan's turnover was crucial. It's safe to say that we will lose every game in which we lose the turnover battle.
• I predicted Indiana would kick a field goal. Bill Lynch, however, after losing by three points while kicking four field goals in 2009, decided he was never ever going to only go for three. That attitude was the difference between Indiana's 31 and actual total of 35.
• Michigan's defense lived up to its bad expectations, yielding 175% of expected yardage.
• While Michigan gave up almost double the expected yardage, it yielded precisely 100% of the expected points. This, my friends, is how you would define a bend-not-break defense.
• Prediction wise, Michigan should have had an offensive multiplier greater than 100% against a defense as bad as Indiana.
• Prediction wise, Indiana's multiplier was slightly too low at 125% (actual was 136%). Hard to determine if it was our defense or Ben Chappell that made up that difference. I will assume it was quality play by a senior QB until he proves me very wrong this week @osu.

How about the rest of our opponents?

Chart of Offensive Expectations (through 5 weeks)

Rank	Opponent	N-PPG	N-YPG	SoS
1	Michigan	37.1	506	67.89
2	osu	36.8	386	63.18
3	Connecticut	30.1	337	65.94
4	Iowa	29.8	378	67.32
5	Wisconsin	29.6	363	61.70
6	MSU	28.1	357	58.82
7	Indiana	27.3	312	51.95
8	BGSU	25.2	271	69.16
9	Notre Dame	24.6	404	75.82
10	UMass	21.7	327	55.48
11	Illinois	20.9	311	71.12
12	Penn State	18.1	337	71.62
13	Purdue	18.1	311	62.92

Metrics

N-PPG or Normalized Points-per-game is taken from the teams average PPG with a SoS multiplier factored in to deflate numbers from playing bad competition and inflate numbers based on playing good competition.

N-YPG or Normalized Yards-per-game is calculated using the same SoS multiplier as N-PPG but using this metric will help us determine a less variant guess as to how offenses will perform (PPG is subject to wild variance based on turnovers and special teams).

Strength of Schedule is taken from Sagarin rankings.

Usage: The chart doesn't predict that #3 would beat #5. Instead it tries to predict with the most accuracy how many points/yards on average each of these teams would score against a common opponent.

Results

Michigan's N-PPG jumps into lead this week after a suspect outing by osu versus Illinois, and further expands their staggering lead in N-YPG, eclipsing 500 yards expected even after it has strength of schedule reducing it to normalized amounts, a full 100 yards more than anyone else on the schedule (120 yards more than anyone else on the Big10 slate).

• There is a full touchdown gap of production between the top two teams and the next five on the list, suggesting a competitive plateau of Iowa-Wisconsin-MSU-Indiana, all shadowed by The Denard Show.
• Indiana makes a leap with their outburst against Michigan. If they can even put up half of those numbers against osu expect their rank to continue to climb upwards as their SoS will jump way up after this week.
• Iowa made modest gains this week after a fairly conservative game against Penn State, which they were in control of the whole way.
• Wisconsin struggled big time against MSU. They are at best the 4th place team in the Big10 behind osu, Iowa and MSU.
• Illinois had as good an opportunity as they were going to get to make a run at an upset (at home, injured opposing QB), but couldn't produce.
• Penn State has been absolutely shut down now by two really good defensive teams (Iowa/Alabama).
• UConn continues to perform decently after two letdowns in their first three games.
• BGSU and UMass fall with their strength of schedule. The rest of the Big10 saw their SoS jump higher this week (duh!).

Conclusions Based on Almost Enough Data

Until given reason to expect otherwise, I am giving our opponents 125% of their N-PPG and 150% of their N-YPG for predictions vs Michigan.

However!!! The elephant in the room is not Michigan's defense. Our defense remains a constant, an ugly constant. The biggest factor remaining is whether or not Michigan can sustain it's offensive play into the Big10 schedule.

Last year, this is where Michigan's offense fell off a cliff. The last seven Big10 games they averaged 20.1 PPG. They did not outgain any of their opponents and they lost the turnover battle nearly every time. Michigan's 2010 unit, however, is light years ahead of where they were last year and, more importantly, healthy (knock on wood).

Best Case

A week ago had a Big10 best case scenario of 6-2. That remains the outlook this week except instead of our second loss coming from Wisconsin it comes from Iowa (We will beat Wisconsin 37.1 to 37.0!).

Worst Case

In a worst case scenario, wherein our offense drops off to 75-80% of current production and we still yield 125% to our opponents, Michigan will go 3-4 the rest of the way with wins over Purdue, Penn St, and Illinois. This is the same from last week (3-5) except we scratch off Indiana from the possible loss column.

The Truth!

Our new outlook ranges between 8-4 and 10-2!

Bottom line: our record improves with a sustainable, explosive offense. Even with a loss saturday, if our offense still shows up to expectation, we have much to be happy about. If our offense takes a dive, however, run for the hills.

Prediction for Michigan State:

Michigan lost a close game at East Lansing due to primarily yakety sax, snapping issues, and botched fake punts. This year sets up much more favorably for Michigan despite having serious defensive issues.

Here are a couple of statistics that might give us hope:

MSU is 10th in the Big10 in sacks allowed, turnovers coughed up and last in the Big10 in 3rd Down conversions, all of which will play a part in getting our defense off the field. Sparty is also the most penalized team in the Big10 (41 penalties for 362 yards).

In addition MSU, unlike Indiana, will kick field goals - they are 7 for 7 on the year.

NSFMF! MSU has a more experienced QB and a better rushing game than last year. What would you call a Chappellbomb that happens mostly on the ground? A Bakerrush? A Bellringing? A Capernickledandy? Whatever it's called, that's the likeliest of outcomes.

But based strictly on the numbers:

It's sobering to see 35 points and 500+ yards as an expected value. Yet there is reasonable hope we will maintain yardage parity with such ridiculous numbers.

I sincerely doubt MSU will hold Michigan to their defense's season average of 101 yards rushing. If they do, it will be a blow-out for Sparty. Last year Michigan gained 28 rushing yards on 28 rushing attempts. You can bet the house that won't happen again.

Even if you believe the assertion that "Michigan hasn't played any real defense yet", you can't argue with the fact that all five opposing defenses have yielded their largest yardage totals on the season (tpilews).

The numbers say 37.1 - 35.1 in favor of Michigan but I can't help but feeling this is a game where special teams is finally going to cost us. Yet after all of this analysis, everything is evenly divided, so I'm not going to pick against Dilithium at home.

Michigan 42
Michigan State 38

/By Saturday at 3:30 I will have convinced myself Michigan is going to win 49-14.

GO BLUE!

"I don't mean much to you, but you mean everything to me."

- Nearly Every Michigan Opponent, 2010.

After reading the excellent diaries prescribing proper fourth down etiquette according to the numbers, odds, and expected values, it makes sense to me that when you have the nation's leading rusher on a 4th and 1, you go for it. Period. But this isn't about our offense's decisions.

Most defenses must put together three consecutive good plays. Michigan's defense has a hard enough time doing that as is, but for whatever reason I've noticed this year that our defense has faced an usually large number of aggressive opponents making the (tactically correct) decision to go for it on 4th down. Part of me thought this was just me being paranoid. But tonight after the game I wanted to see the actual numbers, and boy was this hunch right.

Here is a chart with Michigan and our Big10 opponents in 2010 and the number of 4th downs each team has had to face:

Chart of Opponents Going For It On 4th Down

Does that jump of the page at you? It should.

Michigan, through five games, is facing almost twice as many fourth down conversion attempts as the next team (MSU) and three times or more as many as the rest of our Big10 opponents. Our defense is facing an enormous task of shutting down these hyper aggressive teams.

Does that mean our defense is just bad?

Contrary to that, I assert that it is an underlying trend in the games that Michigan plays, one that reconciles nicely another statistical blip coming from my stats on Normalized PPG and YPG, wherein our opponents typically far exceed their season standards when playing against Michigan.

Our opponents most of the time play their best game of the year against us.

Michigan is by far and away the red-letter, circled-twice, highlighted, make-or-break game of the year for every single team we have played. This will likely continue through Ohio State. Our opponents each and every week have thrown (and will throw) the whole playbook at us, and take risks when they normally would not - for a chance at knocking off Michigan.

• UConn wanted ever so badly to bust open it's season as a Big East Title contender.
• UMass wanted to be The Horror II.
• BGSU wanted to be Toledo.
• Indiana was absolutely out for blood big time.

Only Notre Dame, with their new head coach and coming off of a win and playing us at home, (despite us being rivals neither ND nor Michigan believes the other to be THE big rival), doesn't fit the bill of someone willing to sell their own mother in order to beat Michigan....and Notre Dame was 0 for 0 on 4th downs this year.*

There is playing to win and then there is playing as if the season ends today, and that is what we often times find ourselves facing on defense.

Can anyone really argue against the notion that the four teams listed above weren't playing their lights out when they played Michigan this year? Indiana's season, for all intents and purposes, is now over. They had hopes for an eight win season, now it's likely they will struggle to reach six. I have a hard time believing Indiana is going to come out anything but flat next week @osu.

Looking ahead, can we take some positive away from this?

Yes.

As a direct result of our opponents being hyper aggressive against us this far into the season, it inflates our opponents' PPG, YPG, and TOP. Don't get me wrong, what UConn, UMass, and Indiana did was absolutely the correct strategy - but from a Michigan perspective we don't want our opponents to play correctly by the math. We would much rather them settle for 3 or punt the ball back to Denard. All of these things result in less of our defense on the field, less points for the opposition.

And if our big remaining opponents do that we will allow fewer PPG, and this gives our offense a better chance to equalize for the win. PSU, Iowa, Wisconsin, osu - all of these teams could fit the bill as a more "stodgy" and "conservative" Big10 school. (MSU has already shown a preponderance for trick plays and going for it on fourth down).

Ok. Sounds good to me, but I'm still pissed off about our defense!

Fine. Do yourself a favor and only read the offensive UFR and only watch the youtube highlights of Every Offensive Snap. It does wonders for the blood pressure. Understand that our defense performed precisely to expectations today, but so did our offense!

But seriously, in the meantime, take a deep breath. We now have three road victories in as many years. Road games in the Big10 are brutal (PSU lost, Wisconsin lost, OSU/UM/NW all nearly were upset). And for godssakes get excited! It's MSU week!

GO BLUE!

*In other years, the UM/ND game builds up differently and everyone lets loose, but this particular year it did not set up that way.

[Ed.: Bump. As the OP notes, this data is still very shaky four games in, but the amount of improvement in the offense is so great it can hardly be a mirage.]

In my post the other day, Why should 2010 not be another 2009?, I looked at what our offense has accomplished in 2010 relative to what it had accomplished at this point in the season in 2009. It had two meaningful results:

1) This years' offense draws its potency from highly reproduceable, base set offensive plays, unlike the high variance scrambles and special teams play of 2009.

2) This year's offense is putting up far superior numbers to what they did a year ago (up 28%!!) against as-good or slightly-better competition (77th strength-of-schedule in 2010 vs 114th in 2009).

The Conclusion From the Former:

Our offense will come back to earth from meteoric numbers in out-of-conference play, BUT we have statistically significant evidence to believe that our offense will be far more reliable than last year due to depth, experience, and dilithium.

The Worry:

Our defense cannot stop any team that is executing, whether it's UMass or that-team-down-south. In other words, our wins and losses are going to be determined by how good an offense we face each week, and how well they execute.

Examples: UConn played bad (dropped passes, poor throws) and we stopped them. On the flip side UMass played well (good schemes, good execution) and they had their way with us.

Each and every Big10 offense we play is going to put up at least or slightly better numbers than their normalized offensive output.

So let's find out how bad it's going to be against us with a--

Chart of Infinite Defensive Gloom (after 4 weeks)

Metrics

Normalized Offensive Output - The important thing we're doing here is not looking at the raw PPG and YPG of these teams because it does not account for how good of competition they have played. Four weeks in, the SoS data is far from reliable, but it is at least forming.

Our opponent with the strongest SoS serves as the baseline (Notre Dame with 3 Big10 teams and Stanford). In other words, these numbers estimate what all of these teams' offenses would have generated if they had all played Notre Dame's schedule thus far (Purdue, Michigan, MSU, and Stanford).

Strength of Schedule is taken from Sagarin rankings. (BGSU and UMass are going to have way-inflated numbers at this time, but I included them on the chart anyway as a reminder this is not a perfect analysis and as an interesting couple of data points to track as the season progresses.)

N-PPG or Normalized Points-per-game is taken from the teams average PPG with a SoS multiplier factored in to deflate numbers from playing bad competition and inflate numbers based on playing good competition.

N-YPG or Normalized Yards-per-game is calculated using the same SoS multiplier as N-PPG but using this metric will help us determine a less variant guess as to how offenses will perform (PPG is subject to wild variance based on turnovers and special teams).

I am only tracking our 12 opponents because the only thing that matters is the twelve games Michigan plays and I don't want to get depressed that we are playing Wisconsin and Iowa instead of NW and Minnesota.

Results

This chart pans out as expected. That-team-down-south is the clearcut leader. (Michigan is actually second in N-PPG with 36.3 but FIRST in N-YPG with a staggering 494.5).

We see a clearly defined pecking order in the Big10 that matches very closely the general consensus: clear-cut leaders in OSU-Wisconsin, a muddled middle of Iowa-MSU-Indiana, and a struggling bottom of offenses PSU-Illinois-Purdue.

The exceptions are Indiana, which is trending higher up the rankings due to its offense, and Penn St, which was generally considered a top-4 team in the Big10 going into the season (but is clearly not the case with their offense).

UMass and BGSU will continue to fall down this chart as their SoS gets watered down with conference and 1-AA play.

Conclusions Based on Not Enough Data

NSFMF! Teams always seem to play their lights out when they play Michigan. Michigan's defense has a way of making teams look better than they are. Notre Dame for instance had their highest offensive output of the year against Michigan, operating at 125% of their average YPG.

If we take the MOST pessimistic view and give our opponents 125% of their offensive AND scoring outputs against us and only give ourselves 80% (assumption our offense slows down entering league play) of our average going into the Big10, Michigan ends the season 7-5 with wins over PSU, Illinois, and Purdue.

But remember:

If instead we give ourselves just our average offensive production going into this weekend - our Big10 expected record jumps to 6-2... 10-2 overall!! - with losses coming from Wisconsin and that-team-down-south.

Where does the truth lie? Probably somewhere in between 6-2 and 3-5. Would you take that outcome at the start of the season? In a heartbeat? I know I would.

It is going to be tremendous to watch this Michigan team storm into the Big10 season knowing that our offense only needs to hold serve and our defense can surrender season-best performances from every single opponent and we still have a fighting chance in all of those games!  And lest we forget... DILITIHIUM!

For now, I think we can look at this and add one more reason to the growing pile of why 2010 is NOT 2009! Get excited! Indiana here we come!

Prediction for Indiana:

Michigan's ground game operates at MINIMUM of 100% our normalized average and puts up above-average PPG, but since we only score touchdowns we go to the next closest number after 36! Indiana plays their lights out and operates at 125% of their normalized efficiency, mostly through the air.

Michigan 42
Indiana 31

GO BLUE!