Preview 2016: Heuristics And Stupid Prediction Comment Count

Brian September 2nd, 2016 at 3:06 PM

Previously: Podcast 8.0. The Story. Quarterback. Running Back. Wide Receiver. Tight End And Friends. Offensive Line. Defensive End. Defensive Tackle. Linebacker. Cornerback. Safety. Special Teams. 5Q5A: Offense. 5Q5A: Defense.

Heuristicland

Turnover Margin

NotreDame-Rees-fumble-vs.-Michigan[1]

The theory of turnover margin: it is pretty random. Teams that find themselves at one end or the other at the end of the year are likely to rebound towards the average. So teams towards the top will tend to be overrated and vice versa. Nonrandom factors to evaluate: quarterback experience, quarterback pressure applied and received, and odd running backs like Mike Hart who just don't fumble.

Year Margin Int + Fumb + Sacks + Int - Fumb - Sacks -
2007 0.15 (41st) 14 15 2.46(33rd) 14 13 2.17 (67th)
2008 -.83 (104th) 9 11 2.42(33rd) 12 18 1.83 (57th)
2009 -1.00 (115th) 11 5 1.83(68th) 15 13 2.33 (83rd)
2010 -0.77(109th) 12 7 1.38(98th) 15 14 0.85(10th)
2011 +0.54 (25th) 9 20 2.31 (29th) 16 6 1.38 (33rd)
2012 -0.69 (99th) 7 11 1.69 (69th) 19 8 1.38 (28th)
2013 +0.38(33rd) 17 9 1.9 (64th) 13 8 2.77 (109th)
2014 -1.33 (124th) 5 5 2.4 (49th) 18 8 2.2 (63rd)
2015 -0.31 (92nd) 10 2 2.5 (32nd) 10 6 1.4 (28th)

2015 was nothing like the other new-coach uptick in recent history. Michigan recovered and absurdly low 2 fumbles in 2015; Brady Hoke's first team hopped on 20. With very similar pass rush numbers that's just damned bad luck.

With De'Veon Smith back fumbles lost should remain low. Interceptions are an open question with a new QB and new left tackle. Takeaways should increase as Michigan moves to more zone, specifically sneaky disguised zones, and because of dumb luck on the fumbles. I'd be surprised if this doesn't end up in the top 30, except I'm never surprised by turnover numbers since they're so low sample.

Position Switch Starters

Jibreel Black Ohio State v Michigan 8THB4vo8SwAl[1]

Theory of position switches: if you are starting or considering starting a guy who was playing somewhere else a year ago, that position is in trouble. There are degrees of this. When Notre Dame moved Travis Thomas, a useful backup at tailback, to linebacker and then declared him a starter, there was no way that could end well. Wisconsin's flip of LB Travis Beckum to tight end was less ominous because Wisconsin had a solid linebacking corps and Beckum hadn't established himself on that side of the ball. Michigan flipping Prescott Burgess from SLB to WLB or PSU moving Dan Connor inside don't register here: we're talking major moves that indicate a serious lack somewhere.

The dossier:

Chris Wormley to three-tech. Half the time he'll be playing SDE and he's already displayed an ability to play the spot. Concern level: zero.

Maurice Hurst to three-tech. His best spot. Concern level: zero.

Mason Cole to center. Also his best spot. Concern level: zero.

Jabrill Peppers to SAM. This is mostly a relabeling of his previous position and an acknowledgement of modern football. Concern level: zero.

Taco Charlton and Chase Winovich to WDE. Charlton actually moved their last year and got a few starts in. I'm not concerned about that. Winovich as a position-switch top backup who will see time is less than ideal. There have been some reports that he gets edged fairly routinely. But he is the backup. Concern level: slight.

And that's it. This is the fewest number of significant moves in the history of this preview series. It's basically Winovich, the end.

An Embarrassing Prediction, No Doubt

Worst Case Barring Extreme Injury Scenarios

There aren't many games on the schedule that will be single digit spreads for Michigan. They're favored at MSU and Iowa and only a significant dog at Ohio State; while they could stub their toe against PSU or Wisconsin both of those teams are taking new QBs on the road behind shaky offensive lines. Indiana looms because #CHAOSTEAM, but after seeing last night I'm guessing their offense takes a big step back.

They could lose the three big road games. 9-3.

Best Case

Michigan has more talent than anyone they face. Only Ohio State can argue otherwise, and they've got to replace a zillion starters. If they slip up 12-0 feels way more likely than it should.

Final Verdict

This team will be a national championship contender. Michigan is stacked everywhere except OL, LB, and QB. Those slots project to be average-ish, not season-ending debacles. Meanwhile, the schedule…

OOC
9/3 Hawaii Must win
9/10 Central Florida Must win
9/17 Colorado Must win
Conference
9/24 Penn State Must win
10/1 Wisconsin Must win
10/8 @ Rutgers Must win
10/22 Illinois Must win
10/29 @ MSU Lean to win
11/5 Maryland Must win
11/12 @ Iowa Lean to win
11/19 Indiana Must win
11/26 @ Ohio State Tossup
Absent:

Northwestern, Minnesota, Iowa, Nebraska

…is super pliable.

Here's where I talk about combinatorial mathematics and how it's impossible to predict 12-0 seriously. Let's assume 3-0 in the nonconference. If Michigan has a 90% chance to win every conference game, the chance they go undefeated is 39%. Michigan does not have a 90% chance to win every conference game. There's literally no team in the country you should ever predict goes 12-0. Connelly's S&P rankings think that no team is even 50% likely to go 11-1.

This is why it's stupid to predict 12-0. I've never done it and never will, probably. Predicting 12-0 is an act of bravado not rooted in facts.

12-0.

Comments

WestQuad

September 2nd, 2016 at 3:17 PM ^

12-0

 

I had a stats professor who said that if there is a 97% chance to not go 12-0 you can always take the 3% chance.

 

(Iowa is listed as not being on the schedule and also as being on the schedule.)

ST3

September 2nd, 2016 at 3:59 PM ^

Whenever I go traveling, I worry non-stop until I figure out what 1 item I forgot to pack. Once it was my belt, once it was my tie. At my first ever conference presentation, I had to buy a tie from the hotel gift shop. It sort of matched my suit. Anyway, once I identify what I forgot, I calm down immediately because I never forget more than one thing. I noticed Brian's Iowa/not-Iowa mistake right away. That must mean that his 12-0 prediction is accurate.