SECFans Podcast covering UM/PSU
Their model is actually saying a 1 point PSU win but it doesn't take home field advantage into consideration. The guys themselves predict a close UM win. Always interesting to hear people who have no dog in the fight cover it. Just shy of 28 minutes. Worth a listen.
October 29th, 2018 at 8:03 PM ^
I'm really digging these. Admittedly, I cannot comment much about the mathematics / modeling side of things, but I do like their analysis and commentary. I was also intrigued the model predicted a 1-point PSU lead sans home-field advantage. I haven't seen PSU play this year, but that feels too close for comfort!
October 30th, 2018 at 12:19 AM ^
I have watched Penn State on multiple occasions, though I haven't watched a full game. I am not impressed. They have an average run defense and their OL does not seem particularly consistent.
They haven't been very good on the road.
October 30th, 2018 at 8:54 AM ^
I agree. I appreciate their attempt at objectivity but I don't feel the facts support their theory that Michigan's D will struggle against the PSU offense. UM held MSU (#14 S&P+) to less than 100 yards of total offense and UW (#10) to 280 yards. PSU is ranked #22 and I think the Michigan D will impose their will on them.
Edit: MSU is ranked #14 in defensive S&P+ and #102 in offensive S&P+
October 30th, 2018 at 9:25 AM ^
Michigan St's offense is objectively terrible. They're ranked 108th in scoring and managed 13 against Arizona St. Probably not the best example to give.
We didn't say UM would 'struggle' as much as we said they are better equipped to test the UM defense than Ohio St. I picked 20 points for PSU Josh picked 21. That's half their average of 41. I don't think that's a theory unsupported by facts.
UM has only faced two top 50 offenses this year: Notre Dame (35th) and Maryland (48th). They're giving up 22.5 PPG. If you take away one of MD's 4th quarter scores, UM is giving up 19 PPG to top 50 offenses. PSU is ranked 11th. 20 points isn't in any way unreasonable.
October 30th, 2018 at 11:29 AM ^
At about the 9:30 mark of the linked video in the OP you say Don Brown's defense is so aggressive they have trouble bringing the blitz home against top 50 offenses. I paraphrased that as struggle and don't think that's a stretch from what you said.
Not sure what ranking system you're using but here's what I'm looking at:
UW: OFEI = #17, OS&P+ = #10 (held to 280 yards, 2 sacks-2 TFL's)
Nebraska: OFEI = #37, OS&P+ = #45 (held to 132 yards, 4 sacks-14 TFLs)
WMU: OFEI = #39, OS&P+ = #28 (held to 208 yard, 2 sacks-8 TFL's)
ND: OFEI = #25, OS&P+ = #34 (held to 307 yards, 3 sacks-7 TFL's)
PSU: OFEI = #32, OS&P+ = 22 (held to ? yards, sacks-TFL's?)
Adrian Martinez's stats for the season look a lot like McSorley's:
AM: Passing: 136 comp 205 att 66.3% 1656 yds 8.1 avg 11 TD's 75 long 6 int 146.0 rat
Rushing: 83/408 4.9 avg 53 long 4 TD's
TM: Passing: 122 comp 231 att 52.8% 1628 yds 7.0 avg 12 TD's 93 long 4 int 125.7 rat
Rushing: 110/617 5.6 avg 51 long 9 TD's
October 31st, 2018 at 12:04 AM ^
Technically, Maryland's offense didn't score 21 points against Michigan's defense. One TD was a kick return.
October 31st, 2018 at 12:08 AM ^
I think Wisconsin is a top 50 offense.
October 29th, 2018 at 8:03 PM ^
Hmm, that's interesting. They have us destroying Ohio State but losing to Penn State. I'll have to watch it later.
October 29th, 2018 at 8:07 PM ^
That's a "your model is probably bad" moment
October 29th, 2018 at 10:25 PM ^
I can't remember perfectly, but didn't it call the MSU and Wisconsin games almost spot on?
October 30th, 2018 at 6:45 AM ^
They said they don't like the matchups. They think Michigan's running game is pedestrian, and they're not convinced that the passing game will be able to make up for it. On the other hand, they think the defense is the truth, though they think the result against ND showed that it can't perform against good-quality offenses.
October 30th, 2018 at 12:23 PM ^
Not to mention they both pick against their model, so what does that say about their faith in it?
October 29th, 2018 at 8:20 PM ^
This smells a bad model, period.
October 29th, 2018 at 8:40 PM ^
All models are imperfect, but some are useful.
I actually work in the statistical modeling field, so I've certainly seen (and built) plenty of useless models. You just try to learn from them.
October 29th, 2018 at 8:42 PM ^
Yes but bad ones are not useful, like this one.
October 29th, 2018 at 8:48 PM ^
To play devil's advocate, how do you know it's "bad"?
I'm not arguing it's a "good" model. I think it's an unvalidated model at this point. And I tend to be suspicious of models where they aren't transparent on methodology.
One thing I like about Bill Connelly - he puts his methodology out there, along with his weekly predictions. It can all be validated. He'll also tweak his model continuously.
I play around with my own model (looking at football games on a drive-by-drive basis) but it definitely needs some work. Not doing as well in 2018 as previous years. It is super high on Army (9th in the country?!?), of all teams.
October 29th, 2018 at 8:53 PM ^
Well I'm glad you invoke Bill Connelly, because I like his modeling a lot.
I don't usually watch these guys when they're posted here, but I watched the one where they plugged our matchup with OSU into their model. They lost me as soon as I saw their main inputs are raw stats. I don't know their actual formula or weighting, but I know that raw stats lie and aren't good for predicting college football games.
October 29th, 2018 at 8:57 PM ^
Raw stats also do not factor in injuries. How much is an injured McSorely worth? How about a healthy Tarik Black?
October 29th, 2018 at 8:59 PM ^
Yeah in a vacuum raw stats really tell you nothing.
October 30th, 2018 at 1:19 PM ^
The only thing raw stats do is tell you history they aren't good at predicting the future.
October 29th, 2018 at 9:55 PM ^
What does your model tell you about Saturday’s game? Curious about your thoughts.
October 29th, 2018 at 11:27 PM ^
Fighting Franklins: 0
Hurricane Harbaugh: 167
October 30th, 2018 at 11:46 AM ^
Their model doesn't even factor in home field advantage to their own admission.
October 29th, 2018 at 8:10 PM ^
Math in southern accents is an oxymoron isn't it?
October 29th, 2018 at 9:25 PM ^
Condescension in northern accents isn't...
October 29th, 2018 at 9:38 PM ^
Love your stuff. I would point out that the three games you feel the M offense underperformed were the three road games. I felt Coach Harbaugh was especially risk-averse away from Ann Arbor. The last time Coach Franklin visited, the outcome was not pretty for him.
October 29th, 2018 at 9:42 PM ^
We did comment that being at home changes the dynamics a good bit. But road games against average or below average teams where you underperform is still going to draw our criticism. But be thankful, it's a sign of respect. Our SEC brethren hate us for it, but we're consistent. If you want contender treatment/respect from us, it comes with a good bit of picking nits. :)
I'll flip this around, if Alabama went on the road two weeks ago and beat Tennessee 20-17, I'm guessing you'd kill them for it, and rightfully so. We sure as hell would have.
October 29th, 2018 at 11:04 PM ^
Here's what your analytics don't show.. Penn State on the ROAD
Gave up 550 yards of offense to a shitty Indiana team.. 220+ yards rushing..
On the ROAD to Illinois they gave up 250+ yards rushing..
That is two bottom feeder big ten teams.. What do you think Higdon and company are gonna do to their mediocre to awful run defense? I think last week they were ranked 72nd in the country against the run..
If i had to guess they either load the box or try to, or they get steamrolled for 300 yards on the ground. Michigan controls the clock and wins 41-13
October 29th, 2018 at 11:08 PM ^
I don't think I would kill Alabama for a 3 point road win when taken in context of all of their games. Most great teams tend to have one stinker against a less than great opponent that they win.
October 29th, 2018 at 11:09 PM ^
True, but Tennessee has been destroyed by every ranked team they have played, and has one win over a team with a pulse (and considering that's Auburn, "a pulse" may be generous). Northwestern had a bad stretch early including the fluky loss to Akron that included 3 defensive TDs, but they crushed Wisconsin and had a solid road win over Michigan State. That's two wins that are way better than anything that Tennessee has. Also, Michigan statistically dominated Northwestern, and was by far the better team for the last 40 minutes.
In terms of going on the road, Penn State's best road opponent to date might be Indiana, and they only won that by 5, so feel free to kill them for that. And for other reasons.
October 30th, 2018 at 12:16 PM ^
You did but you also admitted that your model doesn't account for that. Home field is huge and I have a hard time putting much faith in any model that doesn't account for that.
October 30th, 2018 at 1:17 PM ^
Northwestern is leading their division.
October 30th, 2018 at 2:47 PM ^
And got blown out by Duke, and lost to Akron.
October 30th, 2018 at 3:43 PM ^
Did you account for Don Brown losing sleep over this game every day since last year?
October 29th, 2018 at 11:59 PM ^
Trying to make a joke but I totally deserved that one. Touche my man!
October 30th, 2018 at 9:12 AM ^
Whatever. The south sucks.
October 30th, 2018 at 9:26 AM ^
I was raised in San Diego, have lived in L.A., Chicago, A2, North Carolina and TN. There are certainly parts of the South that suck, as there are parts of the North. I dare you to spend a weekend in Nashville, Chapel Hill, Miami or N'Orleans and tell me you had a bad time. If you can do that, you have no soul.
October 30th, 2018 at 9:47 AM ^
I've been to all 4 and had a good time (not so much in Nashville though). Miami is more North Havana than old CSA, Chapel Hill is a college town, and New Orleans is Disney World for alcoholics.
When I think of the south it's east Texas, AL and MS, Macon GA, and other places with a population below say 200k. Places where the stars and bars are as common as the stars and stripes.
October 30th, 2018 at 10:20 AM ^
Oh, so we get to select which stereotypes are acceptable huh?
Hating white peoplez... so hawt right now.
October 30th, 2018 at 10:51 AM ^
Yeah, I don't get the stupid comments from the supposedly enlightened and educated people from Ann Arbor. I've lived all over this country, the south quite frequently, and the only overtly racist/prejudiced thing I can recall occurred when I lived in the Seattle area.
This to my wife, who has a very southern accent, and was the her peers assumption was she would not get along with black people.
October 30th, 2018 at 11:12 AM ^
Lol. I'm usually the one saying these comments. In fact Ace almost banned me a few months back because I called someone out for making a dumbass joke about southerners.
It's called sarcasm.
October 29th, 2018 at 8:13 PM ^
Their Wisky model was accurate...how about the other games they modeled? Didn't they do the MSU UM game? Can't remember
October 29th, 2018 at 8:50 PM ^
They didn't do an MSU game but they did say we should've won 35-7. Which if you take into account the weather and on field mistakes isn't that far off from how the game felt at times
October 29th, 2018 at 9:08 PM ^
How many times did Michigan start with great field position but fail to capitalize?
Their model has been damn accurate this season.
October 29th, 2018 at 9:09 PM ^
Anyone can say post hoc what the score should have been. If they didn't have predictive numbers, they didn't predict it.
October 30th, 2018 at 12:19 PM ^
I think their model may be missing a factor (hate level) for rivalry games as well as the missing home field factor.
October 29th, 2018 at 8:15 PM ^
I wish they would make their models' predictions public to see how accurate it is across the board. IMO they give PSU's offense too much credit; I doubt they get more than a couple big plays off against Don Browns defense.
And I disagree that the Big House won't be amped up. We won't be as loud as PSU for white-out games, but if you sit lower in the bowl you feel the energy as much as at any other stadium. Also, our home-field advantage is historically high.
Overall I definitely enjoy listening to their analysis videos; hopefully they keep making them for Michigan games
October 29th, 2018 at 8:27 PM ^
Does it factor in that Don Brown has been preparing for this game, every day since last year? If not, it’s wrong.
October 29th, 2018 at 8:56 PM ^
They definiately need to add the "Dude" factor.
October 29th, 2018 at 9:44 PM ^
Definitely + defiantly = definiately