Minnesota Football: Models and Bottles! - Penn State

Pat Lovell-USA TODAY Sports

Models and Bottles loves the Gophers' momentum right now. Just how likely are they to keep it?

Recapping from last week:

If we were to look at this game from a matchup perspective, it would come down to who will exploit the two glaring mismatches readily apparent: Minnesota's pass D vs. IU's passing offense, and Minnesota's run game against IU's bad run D. Depending on which Gopher pass defense shows up -- the group that's held 4/7 opponents, including the last two, to under 4.5 yards per passing dropback or the SJSU/Iowa/Michigan tilt where the secondary got bombed on -- this game could get really interesting. That the Gopher pass D has vastly improved in the last 2 games could be an aberration, though it's certainly aiding the season outlook from a modeling POV.

On the season, Indiana was averaging over 9.0 yards per dropback, even after adjusting for the strength of opposing pass defense. Against Minnesota, the Hoosiers put up 7.1 yards per dropback and were sub 4.0 YPDB in the first half. If you want to know why the Gophers won, it's because they forced IU into field goals in the first half and didn't allow Tre Roberson to go off. Nate Sudfield, on the other hand...

Whatever. "The Backwards Pass" game. Point and laugh at the Hooisers.

Models

Winning is awesome, by the way.

Gopher FBS Win Simulations (1,000,000 trials)
Mean FBS Wins Remaining 1.19
Frequency of 1+ FBS Wins Remaining 95.1%
Freq. of 2+ FBS Wins Remaining 23.6%
Freq. of 3 FBS Wins Remaining 1.2%

The simulation doesn't just like the Gophers to get 8 wins -- it LOVES them. Also, the chaos scenario of Minnesota reaching Indianapolis is above a rounding error for the first time all season. This season is fun!

For kicks, I decided to re-simulate the season using the most current data. I mean, it'd be nice to see just how much (if at all) the season would change if the Gophers played every game at the level they're rolling at now.

Gopher Season Re-Simulate (1,000,000 trials)
Mean FBS Wins 6.22
Frequency of 6+ FBS Wins 77.6%
Freq. of Undefeated Season 8.6%

Not a fluke in the least. In fact, they're right where one would expect them after 9 games.

There's simply not understating how much this current 3 game stretch (all as double digit dogs, mind you) has fundamentally changed the outlook of the 2013 season. Starting out the year with a projected win total around the 6 game mark, the Gophers were outside on the fringe bowl radar after dropping two straight ugly games to open Big Ten play. Now, not only has the season been turned around, they've improved their projected win total by two whole games. That sort of improvement is remarkable, and practically unheard-of for this program.

Gopher Win Probability 93.5%
Predicted Points Scored 23.0
Predicted Points Allowed 22.2
Predicted Margin of Victory 0.8
Current Vegas Spread Gophers -2.5
Current Vegas Over/Under 47.5

We officially have a glitch in the Matrix.

First, I'm gonna put it out there: the points model sucks. Well, caveats. The points model is kinda wonk because the whole process of scoring points in college football is highly variable. When you have that much variance, it's difficult to explain it all with a linear model -- even with "margin aware" statistics feeding it. Where I'll give the points model credit is the reasonable approximation to the Vegas spread and over/under. So there's that.

The logit model, on the other hand, I'm pretty confident in. The probabilities are more precise deeper into the season, since the level of opponent adjustment reduces in volatility and the underlying predictor variables tend to stabilize. Up until that point, however, things can get sort of wacky. Take, for example, this here probability against Penn State. It started off at 14% to begin the season (when I basing everything off preseason projections), fell to a low of 8.8% after the Iowa game and has since risen from 42.3% to 68.6% to 93.5% with each successive win. As Penn State has statistically fallen to the pack during Big Ten season, Minnesota has begun to separate from it.

Let that last line sink in one more time.

If you haven't already stopped reading out of complete laughter, I wouldn't blame you. Above 90%? Against Penn State? Surely, you're math is wrong.

Well, I quadruple checked everything, and there are no glaring alarms from neither a data perspective nor a modeling POV. When adjusting for strength of opponent, Penn State doesn't have as potent an offense as Minnesota and even though their defensive statistics are better, the model does tend to favor home teams with better offenses that don't turn the ball over -- all else equal. More things for me to validate during the offseason, though at least I can say I'm not crazy (well...).

Bottles

Hey, we're all about traditions here in M&B. That's why, when present a simple choice, you go with the classic.

Yep, the bottle of the week is Yuengling Traditional Lager. Penn State bros, make sure you bring in a case/keg or four for me if you're heading into town. Because yeah, I want it.

More from The Daily Gopher:

In This Article

Topics
Events
Teams
Players
X
Log In Sign Up

forgot?
Log In Sign Up

Forgot password?

We'll email you a reset link.

If you signed up using a 3rd party account like Facebook or Twitter, please login with it instead.

Forgot password?

Try another email?

Almost done,

Join The Daily Gopher

You must be a member of The Daily Gopher to participate.

We have our own Community Guidelines at The Daily Gopher. You should read them.

Join The Daily Gopher

You must be a member of The Daily Gopher to participate.

We have our own Community Guidelines at The Daily Gopher. You should read them.

Spinner

Authenticating

Great!

Choose an available username to complete sign up.

In order to provide our users with a better overall experience, we ask for more information from Facebook when using it to login so that we can learn more about our audience and provide you with the best possible experience. We do not store specific user data and the sharing of it is not required to login with Facebook.

tracking_pixel_9347_tracker