bronxblue

November 26th, 2019 at 1:58 PM ^

I'm equally dubious; the fact that the only description/analysis of their model is a 24-minute youtube clip that seems to rely on some really small sample sizes to predict how one team would do against another based, it seems, on the idea that a standardized value against an "average" opponent (like SP+ and FPI do) is more misleading.  I'm not a statistician but their approach seems to be equally flawed in that it looks at somewhat-subjective (at least early on) buckets of team quality and then selects only relevant data from opponents who fit into said buckets.  So it seems to ignore/undersell how badly you may beat slightly worse teams and may over-inflate a one-off performance against a slightly better one.

Joined: 11/22/2008

MGoPoints: 172173

MGoStrength

November 26th, 2019 at 11:37 AM ^

If we hold them to under 30 points I'll call that a win.  Only PSU did that and it took two TOs.  That's within striking distance of the offense.  That being said I'd be surprised if that happens unless Fields is in fact more injured than Day eluded to.

Joined: 07/29/2010

MGoPoints: 20002

MGoStrength

November 26th, 2019 at 12:36 PM ^

Fields is really good, but completely different than Haskins.  He won't make quick decisions and reads with laser accuracy the way Haskins did.  But, he will take off and run for 70 yards in a way that Haskins couldn't.  I don't like our chances against either, but they are different so I'm not sure last year was a good litmus test for this year. 

We also need to find a way to block Young because Shea's resurgence coincided with getting good protection and Young is pretty much unblockable.  If Young is getting instant pressure Shea's numbers and ability to make reads and throw accurately will diminish significantly.

Joined: 07/29/2010

MGoPoints: 20002

lhglrkwg

November 26th, 2019 at 12:09 PM ^

I don't want to appear to be dismissive of Fields, but I think people treat him like he's far better than he is. Yes, he's great, but his passing won't absolutely pick us apart like Haskins did. He'll drop some dimes in there, but sometimes he'll miss and sometimes he does seem to hold the ball too long

A game where Fields has to pass to win is one I think we can win. Unfortunately, I expect their running game to do the bulk of the work

Joined: 07/03/2008

MGoPoints: 88119

bronxblue

November 26th, 2019 at 5:06 PM ^

He's a much less willing runner than Barrett, I guess would be more appropriate.  Fields seems like he breaks out one long run a game and otherwise scrambles more.  I haven't watched all their games but I think PSU was the first time he ran more than 13 times in a game all year; Barrett averaged that for his career, and in the latter half was probably closer to 15-16.

Joined: 11/22/2008

MGoPoints: 172173

kurpit

November 26th, 2019 at 11:47 AM ^

Aren't these the guys that predicted Michigan to win the B1G in their season preview of Michigan and picked OSU to win the B1G in the season preview of OSU?

Joined: 02/25/2009

MGoPoints: 9237

Hab

November 26th, 2019 at 12:28 PM ^

The screaming criticism of their model as it applies to Michigan is its failure to account for the team's significant and consistent uptick in passing production.  I would be interested to see the predictions based on Michigan's numbers from PSU forward. 

Joined: 01/02/2015

MGoPoints: 39683

Cmknepfl

November 26th, 2019 at 12:56 PM ^

Honestly I like these guys, I think they do good work.  Its hard to strongly disagree with the actual score predictions they put at the end which both have OSU winning.  I think despite that they are fair and objective about Michigan.  

I do take objection however to the way they categorize our run game.  I think they overly simplify our run game as one static unit that we are simply trying against different teams to see how it does.  They don't seem to consider at all the schematic changes and evolution throughout the year, and maybe they don't have the bandwidth to do so?  But this is a similar criticism to what I would call them out for overall which is that their model can't really account for a team that is trending up or down.  The percentages of what we allow an opponent through the air and ground are averaged across the year.  Michigan's totals themselves are averaged across the year.  As a result they have us predicted for only 5.8 PPA.  I think this is low.  We have been at 11 for the past two games and I think if we can be 7-8 we have a great shot to win.  

Joined: 01/15/2015

MGoPoints: 1135

WalterWhite_88

November 26th, 2019 at 1:07 PM ^

I watched the first 5 minutes and then couldn't watch anymore. The guy is just spouting off stats about yards per game and saying some strange stuff. He says 2 or 3 times "I *think* the passing game has improved." How can you be a college football analyst and say that you "think" Michigan's passing game has improved?? Anybody with half a brain can look at the past 2 games and know definitively that Michigan's passing game hasn't just improved, but it has improved dramatically! This is especially painful to watch after just having listened to the MGOBlog podcast, which is much, MUCH better at analyzing what Michigan and their opponents are doing well and not so well. 

Joined: 06/14/2019

MGoPoints: 4642