2010 Season Probabilities Survey

Submitted by tpilews on August 5th, 2010 at 9:36 PM

Alright, there was a request to insert probabilities or likelyhoods for UM victories during the other season prediction thread.

The consensus in the other thread was a respectable, perhaps rosy, 9-3. Let's see how probabilities could change this.

So, I bring you....

Season Probabilities Survey



August 5th, 2010 at 9:44 PM ^

I almost put the UMass and BGSU games at 100%, but I thought that would be tempting fate too much. I dialed them back a little.

Was the survey supposed to spit out my guesses? I didn't see a results page or anything.

Jim Harbaugh S…

August 5th, 2010 at 9:48 PM ^

I'll be interested to see the results.

9-3 sounds like a bit of a reach.  7 or 8 wins is more realistic.

But I'm reserving final judgment until after the UCONN game.  Win that one and I think the skys the limit, lose and it might be another long season.


August 5th, 2010 at 11:36 PM ^

I'll post some results tomorrow. Unfortunately, these results are going to take time to compile. I'm going to have to put everything in an excel sheet and run an average % for each game. Nothing hard, just tedious. So far, 85 folks have filled out this survey. Got nearly 150 on the previous. Might just need to give it some time before I tally it all out.


August 5th, 2010 at 10:35 PM ^

confidence in the outcome is slightly more defined.  In actuality, I game UM more of chance against osu and Wisco than I had in mind in the first survey.  In the probability survey, I had UM with a 48 and a 47 against Wisconsin.

With no significant injuries, a lot less turnovers and the right bounce of the pigskin a few times and 9-3 or 10-2 is possible.


August 5th, 2010 at 11:05 PM ^

that i blow things up after a michigan loss: 100 percent. Probability that I get wasted after a Michigan win: 100 percent (actually, that goes for a loss too)


August 6th, 2010 at 10:30 AM ^

I would love to see a projected likelihood of winning each game on the 2010 schedule based on last year's statistics for each team in the Mathlete's expected points system.

Obviously wouldn't be an accurate predictor since neither us nor any opponent should be the exact same team, but I think it would be a very interesting data point.

Then we can pretend that our opponents haven't improved any but we have and imagine how many more wins that would equate to in an expected value predictor system.

Gulo Blue

August 6th, 2010 at 4:32 PM ^

I have to remember to check out the averages.  I can run some simulations and give probabilities for final season records using the averages.