View Single Post
  #39   Spotlight this post!  
Unread 25-04-2011, 23:38
MikeE's Avatar
MikeE MikeE is offline
Wrecking nice beaches since 1990
no team (Volunteer)
Team Role: Engineer
 
Join Date: Nov 2008
Rookie Year: 2008
Location: New England -> Alaska
Posts: 381
MikeE has a reputation beyond reputeMikeE has a reputation beyond reputeMikeE has a reputation beyond reputeMikeE has a reputation beyond reputeMikeE has a reputation beyond reputeMikeE has a reputation beyond reputeMikeE has a reputation beyond reputeMikeE has a reputation beyond reputeMikeE has a reputation beyond reputeMikeE has a reputation beyond reputeMikeE has a reputation beyond repute
Re: 2011 Newton Division

Quote:
Originally Posted by Joe Ross View Post
In 2008, when OPR was "good", 66% of team's ranks fell within the SD that I posted. That seems pretty good to me.
Joe identifies the key factor which limits this year's statistical fun - whether OPR is a "good" predictor.

The sum of the maximum OPR of each simulated alliance may well be a good predictor of the winning alliance of a simulated match1. But this year, especially with the exaggerated and quantized contribution of the minibot race, it surely can't be a good predictor of the score of that simulated match.
Teams 148, 1503 & 525 would be a very formidable alliance, but I don't expect even them to score 201 points (200.75 rounded) in a single match

Furthermore, with 88 teams playing 10 matches each, it is quite likely that two or more teams will finish qualifications with a 10-0-0 record, and several to tie with 9-1-0 records, so RS will matter for final seeding. To predict RS we need to predict the losing alliances score (ignoring penalties).

1 I publicly commit to producing an analysis after the Championship looking at the performance of several predictive metrics. Look for a thread once I've recovered.