Girls Water Polo Menu / Archives
- Written by Dan Smith
- Published on July 21, 2006
- Hits: 28076
Computers and human often disagree on who the better teams are in a given sport. One way to combine the various ratings is to mimic what is done in college football with the BCS ratings. This article explains how it works at oregonprepsports.net.
The BCS rating is a rating method similar to that used in college football to determine which teams play for the BCS (Bowl Championship Series) games at the end of the season. The BCS ratings in college football are comprised of a sample of computer ratings and a few polls voted on by humans. In my system, I try to mimic that by using the 3 computer ratings tracked on this site (RPI, Power, Colley) as well as any available polls for the given sport/classification (Coaches, A/P, etc).
For the computer polls, an average placing is determined for each team. If a team places 4th in RPI, 2nd in the Power rating, and 9th in the Colley rating, their average computer rating is (2+4+9)/3 = 5.0 This figure is then translated into a scale from 1 to 0. A team with a 1.0 average rating (first in every rating) would be assigned a rating of 1.0 A team averaging last in every rating would have a rating of 0. All of the other teams will be assigned a rating based on the number of teams in that class for that sport. If there are 50 teams, then each "place" is worth 0.02. Thus a team averaging 5th in the computer ratings would get a rating of (50 + (1 - 5)) / 50 = 0.92.
For the human polls, a rating is calculated based on the teams' votes received divided by the maximum votes possible. In a typical poll where you vote for the top 10, the first place team receives 10 points, the second place team gets 9, etc. The maximum possible is usually 10 multiplied by the number of voters. Using the number of votes rather than the final placing in the polls helps to indicate the true intentions of the voters. If a team places 2nd in a poll with 80 votes while the 3rd place team receives 65 votes, there is a clear distinction between 2nd and 3rd place. However, if the 3rd place team received 77 votes, the gap between the teams should be smaller in the ratings.
The final BCS rating is computed by averaging the rating from the computer poll averages with each of the human poll ratings. If a team receives an average computer rating of 0.900, a Coaches poll average of 0.85 and an A/P poll average of 0.65, their BCS rating is then 0.8. Their BCS rank will depend on the BCS rating of the other teams.