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 9:15 PM ^
It honestly doesn't sound like they've actually watched Penn St play football since the OSU game or even then.
October 29th, 2018 at 9:37 PM ^
We've both seen every PSU game this year except Indiana (one of us saw that game, the other missed it)
October 29th, 2018 at 9:43 PM ^
Cool. Thanks. I guess we're seeing different things, but that's part if the fun.
October 29th, 2018 at 9:57 PM ^
What was your Iowa - PSU prediction?
Using that one as an example of multiple variables, how do you account for two safeties?
I'll hang up and listen now, thanks,
October 29th, 2018 at 10:06 PM ^
We didn't do the Iowa/PSU game. We focus mostly on SEC games and venture outside the SEC for one game per week. Safeties, like turnovers, are highly variable. It's not a predictive thing you can really model for, until you can.
A good example is 2016 Alabama who was so consistent in scoring non-offensive touchdowns, that it became a function of their /offensive/ yards per play. But most teams don't consistently force turnovers and translate them consistently to points.
I actually like our model better for determining potential margin of victory more than the actual score prediction, because that's the real basis for conversation, imo. And in that way, edge case situations don't really impact the game that much.
A key to our model, and college football in general, is that the lower the score, the more variability will impact the outcome. A 17-13 type game, and one play can change everything. But as the number of offensive possessions, and subsequently points, increase, the 'average' game tends to play out and variability's impact is lessened. This game was really no different. PSU wins by 6 and they're probably about 6 points better than Iowa. But if they get those two safeties and the game is under 20 points for each team, those safeties have a much larger impact on the game, potentially.
October 29th, 2018 at 10:19 PM ^
Thanks. Even crazier - at least one of the safeties was an own goal.
Not to keep everyone up tonight, but I keep sifting variables that must be hard to incorporate in models - significant injuries to a key player like a QB where the drop off to #2 is steep (single points of failure as influencers), Franklin's propensity for dumb decisions (admittedly a kind of a constant variable), and wild cards, like when refs form a third team on the field and influence the action/result.
October 30th, 2018 at 10:02 AM ^
Thank you guys for what you do and we are honestly glad to be back on your radar as you cover "big games" outside your preferred conference.
I plan on watching your whole episode and am really curious about your matchup analysis. Personally, this QB is the best we will see all year *and* he is mobile. We shut down Lewerke who has an atypical mobility (he runs once a play is truly broken it seems) but Wimbush got some runs off of us. Yes I know Don Brown has this team prepared but the schemes on paper do not match up well.
As to offense, I think fans are hoping the home turf helps Shea and our coach take some shots. PSU is a really good defense. Their DE (I think) got defensive player of the week and is averaging 3.5 TFLs. Stiff test.
We love our team as you can tell. Personally, I hope you might continue to cover us in light of some of our arrogance and opinions of Southern people.
Imo the model is solid and predicted Wisconsin spot on here. If this game is going our way I don't see it happening until the second half when Frames Janklin starts making decisions. Your model likely cannot account for a coach that huffs paint thinner in the 4th quarter, does it?
October 29th, 2018 at 9:16 PM ^
It's a good point that a big play focused offense is the way to beat Michigan's defense instead of one that tries to methodically drive the field. That was the way ND scored on Michigan, and could be an issue against PSU.. But I don't think they're looking at how Michigan has changed/improved as the season has gone on. They are mixing in more zone and trap coverage on defense while allowing the front 4 to bring pressure, and the running game has taken a big step forward due to OL improvements and QB keepers. PSU has improved as well, their DEs really seem to be coming on, but their DTs are still undersized and depth is thin. The Michigan interior OL could have a very good day and put up a very Wisconin-like game against them and just wear them out.
October 29th, 2018 at 9:36 PM ^
Hi everyone, and thanks for the post! A few things of note:
- Yes, our model is predicting a UM blowout over OSU. But even in that hypothetical model where we shared the video, we both said that we felt PSU had a better chance of the two to actually beat Michigan. A hypothetical is just that, but match ups and schemes matter in football.
- Suggesting we predicted OSU to lose big to UM but PSU to play UM close even though OSU beat PSU indicates our model is bad is an overly simplistic way of looking at college football match ups and should be avoided :)
- Ohio St runs a sideline to sideline offense that relies on getting playmakers the ball in space and having them out athlete their opponents. This absolutely would not work against Michigan (really didn't work against PSU either). Ohio St also is a poor defensive team where PSU is above average.
- Our model is a linear regression model that focuses on yards per play statistics relative to opponent average and not really total yards statistics. This is another reason it would poorly weight Ohio St as we mentioned in the hypothetical, Ohio St is allowing an atrocious 99% of opponent avg rushing and 109% of opponent average passing (PSU is 88/80)
- We both predicted UM will win, but not cover
- We said home field will be an advantage, but not the same advantage as if PSU were coming in as a top 5 team. This wasn't a slight on the Big House, it's the same everywhere, including the SEC
October 29th, 2018 at 9:38 PM ^
If this is really you, give a shout out to mgoblog next podcast :)
and keep up the great work!
October 29th, 2018 at 9:43 PM ^
Ha we mentioned in our last video that we had an account on MGoBlog and would carry on the discussion where you are.
October 30th, 2018 at 8:18 AM ^
what kind of reception do you get over on 11W or RCMB when you predict their games?
October 30th, 2018 at 2:48 PM ^
We've had a lot of great conversation on 11W. Obligatory southern wise cracks about intelligence and a few people whining that we're biased SEC homers, but otherwise really good.
October 29th, 2018 at 9:41 PM ^
Good points. Have you considered Mcsorley is actually injured and not 100%?
October 29th, 2018 at 9:46 PM ^
Everyone says this, but he was doing designed runs in the last offensive series for PSU in that game. We do consider injury for sure, but tend to not put too much weight on it when the player comes back in to play in the game where they sustained the injury.
And his 51 yarder came after the injury in a series in the 3rd.
October 29th, 2018 at 9:55 PM ^
Designed runs that went north and south. Watch the game film again, his lateral mobility was zilch with that giant brace. Its possible that he heals, but I think if he can't "wiggle" around enough, he's about to get eaten by this D.
October 29th, 2018 at 9:58 PM ^
Don Brown aggression variable will apply here.
October 29th, 2018 at 9:59 PM ^
It was so obvious that Iowa was gambling on that play that even the analysts noted that the Iowa defense was aligned with the assumption that McSorley couldn't run. McSorley made them pay for that assumption. Iowa had no one on McSorley and no one back as a safety. Brown won't make that mistake.
October 29th, 2018 at 10:09 PM ^
This may be true, but not the original premise. The assertion was that we need to take into consideration that McSorley is seriously injured. But McSorley came in after sitting two series, and played the 2nd half. And that included a 51 yard run.
Nobody knows to what extent his injuries are. But to suggest he's /so/ injuried that it definitely needs to be factored into modeling and score predicting is to do so in the face of the fact that he 1) returned in the Iowa game and 2) ran multiple designed run plays, including one for 51 yards and another on their last drive.
So, yes, he may be injured - but that'd be a wild guess that nobody knows about based on the information we currently have.
October 29th, 2018 at 10:15 PM ^
Sure, he was playing and running. Did you see the video of the injury? The mechanism of injury portends a Lateral Collateral Ligament sprain. That is the least likely knee injury to cause problems with the meniscus or ACL so good for him. However, when you continue to play on a sprain - I don't care whether it is the MCL or the LCL that is affected, it is going to swell post injury and more than it would have if you immediately rested and iced the injury. That is going to be a problem this week for him in practice. Good care can produce good results and he could be 100% but I've seen and treated plenty of these kinds of injuries. He's not going to be 100% after 7d of recovery; he'll need a brace to stabilize and protect the joint from further injury - that alone will limit his mobility. If there is pain it is going to affect his passing. Injected steroids are contraindicated in an acute injury like this. I expect his game is going to be less effective than if he were at 100%. Just my opinion based on what I saw and what I know about knee injuries from experience treating them.
October 29th, 2018 at 11:06 PM ^
Are you an orthopedic surgeon/resident/fellow?
I agree with you: my quick rewatch from youtube showed significant lateral mobility limitation. Anyone can find that full game video and see what he was doing. His designed run went north/south. Not the vintage McSorley we have seen that is a gittery bug. Interesting to see where this goes.
October 30th, 2018 at 11:38 AM ^
I'm a retired, Physician Assistant with experience in Orthopedics and Emergency Medicine.
October 29th, 2018 at 11:07 PM ^
I think it’s pretty cool that you guys have joined the conversation. I’ve watched your videos this year on Michigan when you’ve done them and think that they are excellent. I’m a Yankee in Little Rock, so I apologize for the northern arrogance that tends to come out. But then I live in Arkansas, so it actually is quite accurate.
October 30th, 2018 at 7:54 AM ^
I don't disagree with this overall. If this game was @ psu I'd still slightly lean UM, but it would be a nudge. I feel much better this game at home after psu playing Iowa.
I am concerned with the matchup as well. I don't like this matchup vs a running qb team with the caveat that psu doesn't even have close to the weapons they had last year. There really isn't a star on offense to compliment the qb. He had three last year and one was a mega-superstar.
Agree OSU matchup is good but playing at osu is so tough. Again, the matchup issue isn't as bad because psu on offense has become a one trick pony. If you shut down TM running it's game over.
Your analysis seems pretty good overall and I enjoyed the listen I predict a 27-20 game UM.
October 30th, 2018 at 9:22 AM ^
Thanks for the contribution, much appreciated!
I really like the part of the model that gives us that per play yardage advantage, great job there.
Question though, is there a method to "big play points"?
ex. PSU makes a "big play" once every x downs; UofM gives up a "big play" once every x, predict x points in x amount of plays
And does it account for the situation: tied, down 3/7/10/14, up 3/7/10?
UofM, especially Paterson's decision making, has been risk averse on the road and against good defenses but they've shown that situationally they can be successful hitting big plays. They hit some at ND to try to mount the comeback, used some to complete the comeback at NW and hit that 79 yarder at MSU at a crucial moment where we all thought they had fumbled away an ez win.
i could see PSU getting 10-17 points as a result of big plays, but UofM should be able to grind out 17-20, plus at least another 10 from big plays of our own.
thanks again
October 29th, 2018 at 9:41 PM ^
Not a bad listen. Seems like they are trying their damnest to find reasons how Penn will win the game. Wishful thinking boys and nice try but I've heard enough. This is the weakest opponent of these last three teams that Michigan played. Yeah Wisconsin isn't looking so hot now but up until that beat down, they were the clear cut favorite to win the west division. Enough said, McSorely isn't doing shit. He's their only hope. His x factor is gone in Barkley. Everyone also says the coordinator was a big part of past success too. Probably but that coordinator doesn't execute plays on the field.Michigan wins 64-3.
October 29th, 2018 at 9:53 PM ^
I’ll shove a lemon up my urethra if we win by more than 60 points.
October 30th, 2018 at 5:13 AM ^
"Michigan wins 64-3."
I like the cut of your jib!
October 29th, 2018 at 9:41 PM ^
So a model that predicted our routs of Wisky and MSU, but predicts a Penn State win (albeit by only a point, which must be within their margin of error) must be wrong because Don Brown. Got it.
Personally, I think this could be much too close for comfort, and it would not at all surprise me if they came out firing and we get behind early and have to fight our way back. Whether that’s enough to win the game remains to be seen. PSU is the best team we have faced since ND.
October 29th, 2018 at 9:56 PM ^
Can't find the MSU prediction video. Kindly post, otherwise it wasn't a prediction.
October 29th, 2018 at 10:01 PM ^
It wasn’t a video or posted “official” prediction but, as others have noted here, they did say at one point that we “should’ve won” by 35-7.
October 29th, 2018 at 10:04 PM ^
Thats not a prediction then. All of mgoblog thinks we shold've won by that much.
October 29th, 2018 at 9:51 PM ^
Penn State does have a big play offense and generally has one very big and one moderately big play a game which go for scores. So their point is that if Michigan stumbles offensively and leaves points on field (a la MSU) then it will be a dog fight. And as they stated, PSU has a better defense and defensive line than NW (slightly) so maybe PSU hangs around long enough to pull the James Franklin stunner (a la OSU and Wisconsin in 2016.)
So the plan is to stop McSorley (QB), Sanders (RB), and Hamler (WR). Hmmm, McSorley sure played well against App St, Kent St, and Indiana. If the offense has continued to improve during the bye week (no rust?) then an efficient Jim Harbaugh offense just might be enough.
October 29th, 2018 at 9:58 PM ^
I actually don't think UM will struggle with Miles Sanders that much. He's not the kind of back that would give them trouble. He's actually a pretty darn good RB, but he's not better than Taylor. And Taylor had a good day against UM, but below his standards. And when Wisky couldn't run for their standard, it melted the entire offense.
PSU fans were ready to riot on us when we did our OSU preview video and said Sanders was a dropoff from Barkley and the running game was suffering a bit.
They also swore he was a 4.3x speed back when we suggested he wasn't an absolute burner. One thing I've found in doing this podcast for 3 years, fans of sports really have no idea just how fast a 4.3 or even a 4.5 guy is. They all think 4.3 guys are common on 'their' team. But I imagine very few of them have been field level to see that kind of speed up close.
October 29th, 2018 at 10:28 PM ^
Interesting observation with respect to Taylor and Sanders. Most on this board were surprised that Wisconsin didn't run Taylor more and at two points it appeared that Christ outsmarted himself and called pass plays that didn't work (or punted on 4th down) when in my imagination it was a foregone conclusion that Wisconsin would get the 1st down running.
October 29th, 2018 at 10:33 PM ^
You say in your video that we have had a problem with big plays last year, and this is very true. DB has made some changes to try and address this. I am sure there are many more, but the most obvious to me is lining up our safety (when they come down to cover the slot) with outside leverage. We tend to give up a few more slants but the downfield slot routes seem to have been contained.
October 29th, 2018 at 10:57 PM ^
Out of curiosity, of the fan bases you have encountered which have been your favorite? And have you encountered any that you don’t particularly care for?
October 29th, 2018 at 11:15 PM ^
We've actually had a lot of fun doing games outside the SEC and interacting with the fans from the B10 + ND. Yes, including Ohio St fans they've actually been great.
Miami fans are, by far and universally the worst. They absolutely reek of a new money fan base that doesn't understand basic nuances of football. They were borderline apoplectic this year when we suggested LSU might beat them in a close game and that Clemson would handle them last year.
Georgia fans hate us because we really don't pull any punches and they perceived some of our criticisms of their team last year (and this year in the preseason) as slights. But we are consistent in our approach that if you're a playoff contender, we scrutinize you beyond your W/L record. Clemson fans, like UGA fans, have bristled at the notion that their team has had any weaknesses the last two seasons.
The blue bloods from both inside the SEC and outside it have the most informed fans who bring the most valuable discussion to our videos in the comment section. I can say this mirrors my experiences traveling around and interacting with fan bases from literally every conference and attending games.
We've not been able to put our finger on this year's LSU team, and the LSU fans have largely been willing to remind us about our missteps with their team but at the same time be good natured about it.
Alabama fans, for the most part, like us but that's because Bama has been good and consistent. I imagine they wouldn't like us as much once Bama falls back down to earth when Saban leaves (it's going to happen).
But our favorite thing about doing these games is that literally every show we do, fans on one side call us 'homers' for the other side - regardless of which teams we're previewing. If both sides think you're a homer for the other team, you're probably being pretty objective.
October 30th, 2018 at 12:21 AM ^
So, does your model take murderwolf into account?
October 30th, 2018 at 6:16 AM ^
Thanks for the thoughtful response and it makes sense. I suppose there will always be an adverse reaction to calling someone’s baby “ugly” even if it is the truth. I am hoping that our bye week have us more time to heal and gameplan but historically our post-bye performance has been mixed.
October 29th, 2018 at 11:02 PM ^
I don't really buy that you beat Michigan with a big play vertical passing offense. 2nd and 10 turns into 3 and out pretty quickly against Michigan, and it's underselling the offense substantially to think you can constantly cede field position against Michigan and expect to win.
October 29th, 2018 at 11:19 PM ^
It depends on how you look at it. The notion that you have to beat them with big plays is a testament to just how good the defense is. You absolutely cannot drive the field on UM's defense. This has been the same way teams would have any sort of prayer with Alabama over the last decade, so take it as a compliment.
You can either play the game straight up, and die a slow death of futility trying to run a standard offense and drive the field, or you can take shots. Shots has the potential to have you lose by a wider margin, but when it's your only shot to win you're faced with either known death by eventual, but respectable suffocation OR equal parts option of potential win by hitting some home runs vs losing big because the big swings don't pan out.
If I know my team can't win straight up, I'm taking door #2 without worrying whether or not about the final score (losing) margin.
October 30th, 2018 at 8:03 AM ^
I think that is a fair point. It's what psu did last year (granted, last year UM team wasn't nearly as good). But they used explosive weapons to go big with massive chunk plays. Many of them were the infamous TM arm punts to sensational athletes or screens to running back that looked like bo jackson.
I think TM presents a challenge UM hasn't seen yet but he doesn't have that compliment of weapons. PSU screwed the pooch last year. They should have beaten osu and been in the playoff.
In fairness I could see, after the bye, the UM offense coming out slow. I see the "D" coming our hardcore though. I mean the defensive coordinator says he thinks about psu every morning...
October 29th, 2018 at 11:51 PM ^
But that's exactly what ND did. It won't work every time, but even if it's just 1/3 of the time and gets you in scoring range that can easily be worth it. If you get 12 drives and score on 4 of them, say 2 TDs and 2 FGs, that could be enough to beat a Michigan offense that struggles. Especially if you can also get a big punt/kickoff return or int return as well.
October 29th, 2018 at 11:11 PM ^
I liked the podcast and the way it was presented.
Interesting and I think fair takes on us and PSU.
October 29th, 2018 at 11:15 PM ^
Yeah, I didn't totally agree with the conclusion, but I enjoyed it.
October 29th, 2018 at 11:24 PM ^
Wimbush got his yards in chunks. 7 of his 12 completions came in the first three drives. So the assertion that chunk plays are the only way to move/score against the Michigan defense seems good. He caught the lightning in a bottle for the first three drives. Then poof. He turned into a pumpkin and got benched.
McSorley is one of the most experienced B1G QBs. He also lost his great offensive coordinator from last year. I do not know how to include that in a statistical model.
PSU scored 75 points in the fourth quarter, about half of that coming against Illinois. Michigan on the other hand has scored about 63 points in the fourth quarter, with 17 points against the badgers being the highest. Again, not sure how to account for this in a model. May be this is what the model is picking up as a PSU advantage in a low possession short game.
I guess it will be to Michigan's benefit to score early score often and make it a game with many possessions to iron out any chance of a PSU win.
October 30th, 2018 at 1:00 AM ^
Our offense isn't a power house by any means. I think if Tarik Black comes back healthy it could be playoff caliber O.
However, If our offense comes out goofy and we don't convert in the red zone like we have not done consistently (Notre dame, Wisconsin, and MSU) and it comes down to some Trace Mcsorley jump balls it could get scary.
Fan of the show. Keep it up guys.
October 30th, 2018 at 11:55 AM ^
I think there's an intangible that the model really can't compensate for - there isn't a way to include it. Don Brown has been mindful of this game for a very long time. Give him a year to think of ways to beat you and then give him current tape and he'll find a way to scheme and defeat your team.
Yes, the D may not have Gary (honestly I'd rest him until the Indiana game and get him in a series or two there and then sit him for the rest so he's nice and fresh for OSU) but others have come on very strong from their 2017 counterparts. Another item that can't really be included is that the playbook is opening up as we progress through the season. I'd expect a few new goodies to show up against PSU that weren't used all year.
The practicality of that approach is debatable. I can see both good and bad in it. In the end so long as it translates to wins, by all means keep on keeping on.
It was noted in the discussion about how modeling in a home-field advantage would take more elaborate algorithms so that said can't really see much fault with the pure math view. Things such as emotion and extra motivation are difficult to quantify without adding in some arbitrary figure. That said, M wins and the tour continues right through to OSU.