2012 BMG Senate / Presidential User Predictions

Here’s what I got for the call for 2012 BMG predictions by my deadline (8pm tonight, i.e. 24 hrs before polls close tomorrow). Let me know if I missed any or got any of these wrong, I will do another pass tomorrow morning and update if necessary.

Ranked in order of Warren pop. vote prediction, ascending. Averages at the bottom of the image.

* = This symbol on a given name means that the numbers were assumed from the prediction language, rather than directly transcribed. I can remove if you’d rather they not be included.

Model D (#5 above) is not a user, rather it is the end result of one of the prediction models I have set up based on this data. I set up four possible ways to work with the BMG predictions data, but I chose Model D because I like the assumptions it makes best. It is based on adjusting for the predictions data from a) BMG Senate primaries 2010, b) BMG Senate predictions 2010, c) RMG Senate predictions 2010, d) the 2010 straw polls of each community, and e) the volatility of each set of predictions. I will likely do a separate post on the prediction model, since that is the fun part of working with this data on my end.

For what its worth, only one of the four models had Warren winning, but I felt it was the best one anyway:

Model A (Straight adjustment on past BMG data):
Warren: 47.52
Brown: 52.39

Model B (Adjustment based on past RMG and BMG data):
Warren: 49.89
Brown: 49.89

Model C (Adjustment on past BMG data * volatility):
Warren: 49.67
Brown: 50.1

Model D (Adjustment on past BMG and RMG data * volatility):
Warren: 50.97
Brown: 48.72


5 Comments . Leave a comment below.
  1. Let everyone play

    Keep the * players, why don’t you.

  2. Waiting for final results...

    … but this is looking like user patrick might have nailed it, getting the exact Warren/Brown mark and the exact EV count, as well as calling the tierney race as a close victory on the side.

    Full numbers once the counting is over

« Blue Mass Group Front Page

Add Your Comments

You must be logged in to post a comment.

Tue 28 Mar 11:49 AM