West Lot Pirates Podcast

The Scuzz Model is back this week, just in time to preview the Big Ten Conference Season!  Not a whole lot to say about last week’s NU win (though a great thing in and of itself), so we’ll focus on looking forward.

Before I get to the usual array of projections, I wanted to highlight an adjustment I’ve just baked into the model.  It’s pertinent because as I was looking at which Big Ten teams the model sees as likely division champs, I was surprised to see Michigan still highly rated.  Some digging has led me to discover the following:

UMass Mess

As I’ve mentioned previously, I do not include results from 1-AA matchups (FCS) in the Scuzz model, primarily because those games really skew ratings;  all other FBS teams do get counted though.  This has created a small issue this year, due to the complete blow-out nature of every game Massachusetts has played.  Don’t get me wrong, I’m happy UMass and their top-notch college band have made the jump to football’s highest division, but their output on the field is screwing up my model.  This is exacerbated because both Michigan and Indiana have played UMass, and won by 40+ points (Connecticut crushed the Minutemen as well, but it’s the B1G teams I really care about for obvious reasons).  Look at how much the Wolverines and Hoosiers expected winning percentage increased on account of the UMass game:

The Indiana result in particular is really jarring.  I’ve decided to treat these matchups like 1-AA games and exclude them from the model.  As a result, NU gets a slight boost in the expected outcome against both of these squads:

NU Projections:

Note that “last week” actually represents the projections I shared two weeks ago.  You can see in the game-by-game that Michigan, Indiana, MSU, and Illinois especially are all looking a little better in terms of NU’s likelihood of winning.  Unfortunately a lot of this is being wiped out by improved results from PSU and Minnesota, which look much tougher statistically after the last two weeks.  As a result, the wins distribution has moved very little from last time out:  some slight increases in the 7 and 8 win columns, but nothing significant.

While a win this week would likely push NU into the official Top 25 rankings, it will probably have little impact on the Scuzz model’s expectations:  the Wildcats are a huge favorite, as you can see above.  On the podcast this week, John talked about how bad IU’s defense has been… here you can see it statistically – essentially a straight line from week 10 last year, though today, over 310 pts allowed per 100 possessions.  The Scuzz model average for all teams is around 220, so IU is about 30% worse than average.

You can also see that now with the UMass game excluded, IU’s profile drops in that 3^rd week of 2012 (rather than an increase, we see the offense drop off)…. That is due to an adjustment I made to account for the Tre Roberson injury.  Interestingly enough, the following week against Ball State, the offense performed pretty much as the Roberson-adjusted data point would indicate.

This graph tells you one last thing:  IU’s defense gives up almost twice as many points per possession as its offense is able to score.  This is fantastic news for a Cats team who will force IU to throw the ball with their 2^nd and 3^rd string QBs.

Big Ten Predictions:

To preview the conference overall, I’ve prepared just a couple quick charts to indicate the Scuzz Models picks in each division.  First the leaders, where the model still really believes in Wisconsin.  Some of this is the way I blend efficiencies from the prior year, but some of it is also due to the Badger defense, which is outpacing last year’s squad enough to take some of the edge off the plummeting UW offense.  OSU is a strong second, with little chance for the other four to pull even at this stage.

The Legends division still resembles the free-for-all we thought it would be.  Michigan and Nebraska are neck and neck, but 5 of the 6 teams have reasonable chances of finishing first.  I expect that a Minnesota win this week vs. Iowa would start to move the needle for the Gophers, who have not performed well enough yet to outweigh last year’s disaster ratings.

Picks Around the Country: 

All games vs. Spread: 93-85 

Week 1:  4-1
Week 2:  2-3
Week3:   3-2

Had to take week 4 off for some travel, but maybe it was a good thing… the model was not great that week.  Still over 50% for the year vs. the spread though.  Here are this week’s picks:

Wisconsin (+11.5) at Nebraska:   The model has been over estimating the Badgers all year, but this line befuddles me.  Wisconsin lost a close game on the road to Oregon State, while Nebraska got crunched by UCLA (who subsequently lost to Oregon St).  The communitive property of college football tells us these teams are not separated by two scores (really, we have one good rushing team, one good defensive team, and a lot of question marks, which leads me to think close low scoring affair… or Wisc wakes up and blows them out).  Scuzz model favors Wisc by 1.

Texas (-2) at Oklahoma State:  I don’t love this pick because it means trusting the Longhorns, but at the same time Arizona was able to slow down and outscore OSU (and then got shutout and lapped by the Ducks).  I know OSU has a good home field advantage, but I think Texas is due in this series (and say what you will about their QB situation, they have a strong D).  Model likes UT by 7.  (I’ve just learned UT will be missing 4 players including the kicker… still rolling w this pick though).

West Virgina (-11.5) vs. Baylor:   WVU has not stopped on offense.  The “air raid” is in full effect, and Geno Smith looks awesome.  On the other side is Baylor, who lost so much talent but has benefited thus far from weaker matchups.  I and the model expect WVU to smash them in this game. 

UTSA (-1) at New Mexico State:  the Road-Runners are unproven yet at the FBS level, but they are probably the best of the 4 teams that moved up from FCS this year (currently 4-0).  New Mexico State is the worst team in FBS.  Scuzz model loves UTSA (mostly because they have a ton of returning players)… seems reasonable that they have a shot to win.

Oregon (-28) at Washington State:  Uh… has anyone been watching Oregon?  Their O is still explosive and their D is better than ever. The model says 36 points and I don’t think that’s enough.  Oh and Wash St. lost to Colorado last week.   At home.

The Already Obsolete BCS Championship Hunt: 

Two more weeks and two more major competitors gone.  USC and now Oklahoma have joined the one-loss crowd and two new teams have joined this weekly tracker – Notre Dame and Louisiana Tech.  The story here though is Alabama… 75% chance of going undefeated now, which frankly doesn’t seem off – they of course have tough games, but nobody appears to be in the same league as them (and LSU’s tight win over Auburn has really hurt their profile).  As for Alabama’s potential opponent?  I like Florida State’s chances – they’ve already beaten Clemson, and while the ‘Noles have tough games against Virginia Tech and Florida, the other prime candidates have much tougher matchups;  Oregon would be the one team you could argue has an easier road, especially considering Stanford’s loss last night.

Last week was one of the more enjoyable games I’ve been to at Ryan Field in a few years.  The emergence of Mach V, the performance of the defense against a tough BCS conf foe, and the finish (weather included) really made it a helluva night.  Statistically, the NU D over performed its Scuzz Model profile, while the offense was right on par with expectations (the model forecast a 32-25 Vandy victory).  Yes there was talk about the WR corp being underwhelming, but as we mentioned there should be good opportunity for the offense to shine this weekend against BC.

The result on this week’s model is improved expectations in almost every matchup on the schedule – this is certainly driven by the poor overall performance from other conf teams this past week.  Primarily the Nebraska and Illinois games show improved outlooks.

You can also see that the projected wins distribution continues to move steadily to the right – now the model has NU winning 8 games or more in almost 60% of scenarios.

This week the model really likes NU’s chances.  BC’s offense had a ton of players coming back in 2012, but they were a really inefficient unit in 2011.  Their current offensive efficiency rating is 100 points lower than NU’s, while the defense performs about 50 points better.  (Sidenote – for the uninitiated, the Scuzz Model uses offensive and defensive scoring efficiency – points per 100 possessions to rate each teams’ O and D units):

I would expect BC to end up with a better offensive profile this year than you see above, especially after that week one game showed Rettig’s improvement from early 2011.  On the flip side we also saw that BC’s week 1 defense was a shade of what it was with Luke Kuechly last year.  I think the model is close on BC’s overall profile (expected winning percentage around 40%), but hasn’t got the units correct – offense is probably underrated while the defense is overrated.  As such, I expect this game to be a higher scoring affair than last week’s vs. Vandy, and while I feel good about this game I’m not as confident as the Scuzz model (78% winning likelihood).

Around the Country:

All games vs Spread: 45-36 
Week 1 picks:  4-1
Week 2 picks:  2-3

Foiled by my reverse jinx pick of Illinois (was totally worth it), it was a rough week for the Scuzz model’s picks last time out; but overall the model is still looking real good over 55% vs the spread across all non-FCS games.  Here are this week’s attempts (note we use opening lines for this segment): 

Louisville (-3.5) vs. North Carolina: 

Charlie Strong’s team is looking pretty darn good thus far, while North Carolina looked awful in week 1 and has enough trouble following them to be in the conversation with Miami and Penn State.  Scuzz Model is barely over the line with a -4, but I feel good that UL prevails by more.

Middle Tennessee State (-2.5) at Memphis: 

betting against Memphis has been one of the strongest elements of the Scuzz model (6-2 picking against Memphis LY).  No reason to stop now when the model likes the Middies by a full touchdown. 

TCU (-22) at Kansas: 

Weis’ team has looked bad; I don’ t know that they should be 37 point dogs to TCU at home as the model says, but that’s enough of a cushion that I feel good about taking the Frogs.

Ohio (-6.5) at Marshall: 

Ohio, darlings of the non-AQ hopes to break into the already-obsolete-BCS this year have been pretty dominant thus far, and face a not great Marshall team.  Marshall was obliterated by WVU in week 1, and the model likes Ohio by 15 points. 

Stanford (+9) vs USC: 

I’m already on the record from the podcast as expecting USC to win this game due to how they matchup vs. Stanford athletically:  I expect the Trojans speed to give them the edge over Stanford physicality.   That said, 9 points seems like a lot… especially to the Scuzz Model which favors Stanford by 2.

The Already Obsolete BCS Championship Race:

Arkansas’ totally unexpected collapse last week has boosted Alabama and LSU’s undefeated chances (LSU in particular, since their game vs. Arkansas is much closer to a toss-up).  Wisconsin of course has been eliminated with their loss – the hit to the Badgers’ statistical profile has also elevated Ohio State into this mix.  It’s my opinion that the Big Ten is at serious risk of handing a Leaders Division trophy to Ohio State at the end of the year (better than having to hand to Penn State, but embarrassing nonetheless).   The other usual suspects are all still in play… USC’s chances will improve measurably should they beat Stanford this weekend.

Just a note — the Scuzz Model is taking next week off, but will be back to preview the opening weekend of the Big Ten conference season.