Predictive Win Model for Week 10
Well, many of us are probably still having nightmares of Penn State repeatedly running the counter play, the bootleg, and being McGloined down the field for touchdown after touchdown. Yes, I was there in Happy Valley on Saturday and it was awful. I knew it was a long night when at one point one fan shouted to another, “where has this offense been all year?”…….we all know the answer to that one. It was not the stellar Penn State offense, but rather the Michigan defense making the third string QB with virtually no in-game experience look like a seasoned veteran. Enough of last week. Despite more bad news for the defense we must look on to predictions for week 10. In week 9, my model had a predictive accuracy rate of 71.7% (38/53) nationally, which was slightly better than the Sagarin rating accuracy of 69.8% (37/53).
Week 9 Key Accurate Predicted Outcomes:
Tulsa (.53) at Notre Dame (-.23) = Tulsa
Missouri (2.75) at Nebraska (3.45) = Nebraska
Baylor (2.33) at Texas (.64) = Baylor
Week 9 Key Inaccurate Predicted Outcomes:
Michigan (1.21) at Penn State (-.05) = Penn State
Michigan State (3.40) at Iowa (2.65) = Iowa
Model Metrics:
Using a variety of team statistics (SOS, win percentage, turnover margin, offensive yards per play, defensive yards per play, having a home game, and so on…you name it I have it and have looked at it) on a national level (Division 1-A - FBS only), the team statistics that best predict weekly winners and losers for week 9 were, in order of importance:
- Defensive Yards Per Play
- Point Differential (avg points scored – avg points given up)
- Win Percentage
- Offensive Yards Per Play
- Turnover Margin
Week 10 Big Ten Rankings:
The Big Ten rankings for week 10 are below. All of the variables in my model are presented in z-scores (-3 to 3) that were computed on a national level, with the higher the score the better for variables for which positive results are better (offensive yards per play, win percentage, turnover margin, and point differential). For the lone variable (defensive yards per play) that is inversely related to winning, having a lower value is better. The variable PREDSCOR is the output of the model, and the game winner is determined solely by the higher of the score between the two teams. The PREDSCOR is not bound to -3 to 3.
| TEAM | PREDSCOR | Sagarin | DEFYRDS/Play | POINTDIFF | WINPERC | OFFYRDS/Play | TOMARGIN |
| Ohio State | 5.92 | 87.05 | -1.87469 | 1.92326 | 1.43224 | 1.08954 | 2.32058 |
| Iowa | 4.02 | 87.48 | -1.11964 | 1.27180 | .86744 | .89792 | 2.13952 |
| Michigan State | 3.66 | 83.84 | -.79197 | .65806 | 1.43224 | 1.35302 | 1.05316 |
| Wisconsin | 2.91 | 84.23 | -.20788 | .95753 | 1.37576 | 1.06559 | .14787 |
| Illinois | 1.28 | 80.69 | -.92019 | .56932 | .35912 | -.51528 | -.03319 |
| Northwestern | 1.00 | 69.51 | .13403 | .29203 | .86744 | -.13204 | .87210 |
| Michigan | .86 | 74.76 | .94607 | .21808 | .35912 | 2.10752 | -.57637 |
| Penn State | .22 | 75.43 | -.06541 | .03322 | .35912 | -.16797 | -.21425 |
| Purdue | -1.50 | 65.05 | -.25062 | -.82676 | -.14920 | -1.10211 | -.57637 |
| Indiana | -1.58 | 61.14 | 1.28798 | -.17974 | -.14920 | -.02425 | -.39531 |
| Minnesota | -4.51 | 57.43 | 2.22824 | -1.01754 | -1.73065 | -.29971 | .14787 |
Predictive Model Results for Week 10:
Illinois (1.28) at Michigan (.86) = Illinois (Zookered…but let’s hope I’m wrong)
Minnesota (-4.51) at Michigan State (3.66) = Michigan State
Iowa (4.02) at Indiana (-1.58) = Iowa
Wisconsin (2.91) at Purdue (-1.50) = Wisconsin
Northwestern (1.00) at Penn State (.22) = Northwestern
November 4th, 2010 at 9:29 PM ^
Northwestern will be your system's version of Tulsa this week
Underdog outright winner that catches everyone off guard
Illini could win, but at +3 and given whats going one, we wouldnt be calling it an omg upset. But the Cats we would. It's anywhere from +6 to +7
The Tulsa call was a great feather in the cap. I think you'll add another with NW
November 4th, 2010 at 9:57 PM ^
and Michigan loses.
(But very impressive. Keep up the great dairies)
November 5th, 2010 at 12:02 AM ^
What I wouldn't give to play Minnesota this year.
It's also interesting that M's defense is only 1 standard deviation (ish) below average on yards per play. I guess it makes sense though, we've been getting slain on time of possession and 3rd down conversion. I'm not a big TOP guy, but it keeps our strength off the field so, that's bad.
/understatement
November 5th, 2010 at 12:15 AM ^
If the predictor is wrong once this week it will probably be Illinois beating Michigan at home. All the other wins look like solid locks! Oh well, I suspect it will be a close game with Michigan getting their chances to win. It will be a new low if Illinois can come in and blow us out at home. Still holding out hope that the offense will be clicking and we can get a two TD lead and just keep on piling on the points. My prediction UM 42, IL 33
November 5th, 2010 at 8:02 AM ^
They "could" and will, 52-31.
November 5th, 2010 at 5:15 PM ^
My rational prediction would have been that we would be 5-3 at this point in time and would have been looking at Illinois and Purdue as our last 2 wins to go 7-5 for the regular season. Until tomorrow's game is complete, this prediction still stands as possible even though none of us would have predicted that Illinois would be this good. Let's hope we can out offense them and catch a mistake or two by their offense.
Go Blue!
Comments