Post Week 6: Yardage Analysis and Predictions + Score Predictor

Submitted by tpilews on October 12th, 2010 at 12:06 AM

A quick recap of what week 6 was and what was predicted... Week 6 predictions found in previous diary here.

Prediction vs. Actual

UM offense: 553 yards vs. 377 yards = 68%

UM defense: 495 yards vs. 536 yards = 108%

While UM ended up short of their predicted yards based on ridiculous output to date, they were still able to gain more yards than Sparty typically gives up in a game. (377 vs. 350). On the flipside, the UM defense gave up nearly 100 yards more than Sparty typically gains. Needless to say, those are not good numbers. When you combine them with three turnovers, the likelyhood of a win is slim.

On a positive note, UM left 18 points on the field in the form of a dropped TD by Hemingway and two INTs thrown by DR in the end zone. In other words, UM was three mistakes from being dead even in this game.

My predictor came up with a final score of 40.5 for UM and 39 for Sparty. Not much analysis needed here as score is based entirely on yardage. On to week 7.



Because of the poor numbers, both on offense and defense, Michigan's yardage lead slid nearly 50 yards to Iowa. The Hawkeyes join OSU as the only teams predicted to outgain UM. At this point, Michigan still has a favorable matchup against Purdue and Penn State. The recent success of Illinois has really made that game quite a tossup.

As with last week, I added an offensive and defensive ranking based on yards. MSU moved up 2 spots to number 8 after the UM game. The game this week against Iowa should prove to be another huge test for the Michigan offense.

At this point, with the score predictor, I'm inclined to throw away the "hybrid 2009/2010" stats. We have just edged past the halfway point of the regular season and teams have started to display an identity. As many yards as UM is putting up this year, they just haven't scored as much. It is definitely taking UM more yards to score the same amount of points. Because of the special teams woes [Ed: and bad defense] UM is finding itself in bad field position game-after-game.



UM - 407
Iowa - 454


UM - 28
Iowa - 35



October 12th, 2010 at 12:18 AM ^

Interested to see how the method played out over this past weekend and going forward. I've thought about doing something similar.

Based on outcomes from the first two weeks in Big Ten conference play, what is the accuracy of win-loss, total yardage, and points scored? I realize that you might not have the information needed to project every game, but I'm curious to see the accuracy for the games you're able to project.


October 12th, 2010 at 10:12 AM ^

A predicted loss - yikes.  This week's game is huge.  If the team loses two straight going into its bye-week, the collective fanbase is going to be in total meltdown mode for the two weeks that follow.


October 12th, 2010 at 11:05 AM ^

From what I've gathered about Iowa, their special teams are an achilles heel. Unfortunately, UM doesn't appear to have the team to take advantage of that. But, if UM can pick up a TD from somewhere other than their offense, then that'll go a long way in securing a victory.


October 13th, 2010 at 1:05 PM ^

I've always liked reading your diary, but I think there are a couple of ways to make it more accurate.

The first way is simple, and not even really necessary, but I think would be cool. You could break the prediction down into rushing vs passing, to see where our yards were going to come from. 

The second way, however, I think would make the prediction much more accurate. Right now, your predictor is only telling half the story. Namely, the average number of yards a team gives up per game, and how Michigan's offense is predicted to perform against that. However, the predictor isn't including how good each team's average yardage per game is. For example, Wisconsin is giving up 338 yards per game, however what does that mean if that 338 yards per game is only say 60% of their opponents average? Then it would be easy to say that Michigan will gain 148% of our yards on their defense, but the more accurate number is more in the middle, around 104%. However, to accomplish this, you would have to create this type of spreadsheet for all of Michigan's opponents, which gets ridiculous.

Otherwise, love seeing how accurate your picks are each week.