--9.18.2006--
Ladder Rating System?
I've always hated the college football polling system, because as several writers have mentioned repeatedly, the people voting in the polls are mostly voting based on hearsay, and some of them even get other people to vote for them. None of the voters are really able to take the time to review all the games to really determine which team is the best. I've hated the BCS even more because despite it's claim to add computers to the mix, it still far too heavily relies on opinion polls to generate the bulk of its information.
Being an avid gamer and having played several games online in sactioned leagues, the most common ranking system used is done in what is called a "Ladder" format. Everyone begins with the same number of rating points, and as you play games against opponents, your rating adjust based on the rating of your opponent and whether you won or lost.
Why can't this be done for college football, where 119 teams are all trying to prove that they're worthy of playing for the National Championship at the end of the season? So after several years of nagging myself about it and never doing anything, this year I finally broke down and set it up.
The rules are as follows:
1) All teams begin the season with 1000 points.
2) Winning a game versus another team results in gaining 100 points, minus 10% of the difference in points when subtracting the losing team's rating from the winning team's rating. If the difference is negative (meaning the losing team had the higher rating) then the points are added.
3) Losing a game versus another team results in losing 100 points, minus 10% of the difference in points when subtracting the losing team's rating from the winningi team's rating. Again, if the difference is negative, the points are instead added.
4) Division I-AA teams count as having 0 points for the purpose of calculation.
5) Defeating a team by 21 or more points results in a bonus of 15 rating points. The losing team is not penalized for this.
Example 1: Team A (1000 pts) def. Team B (1000 pts) by a score of 10-3.
Team A = 1000 pts + (100 pts - 10% of 0 pts) = 1100 pts
Team B = 1000 pts - (100 pts - 10% of 0 pts) = 900 pts
Example 2: Team A (1250 pts) def. Team B (800 pts) by a score of 49-10.
Team A = 1250 pts + (100 pts - 10% of 450 pts) + 15 bonus pts = 1320 pts.
Team B = 800 pts - (100 pts - 10% of 450 pts) = 745 pts
Example 3: Team A (900 pts) def. Team B (1200 pts) by a score of 20-17.
Team A = 900 pts + (100 pts - 10% of -300pts) = 1030 pts
Team B = 1200 pts - (100 pts - 10% of -300pts) = 1070 pts
In the above examples, it demonstrates how many points are awarded or lost by each team based on the rules I included. In example 1, the teams have identical point values, and because of that the points awarded/lost are exactly 100 points. In example 2, the teams are 450 points apart in rating, with the winning team having the higher rating, so 10% of the rating difference is subtracted from 100, so the team is only awarded 55 points for the win, however 15 points are added for the decisive victory of 21 or more. In example 3, the teams are 300 points apart in rating, however the winning team has the lower rating, so 30 points are added to the 100, for a total of 130 points awarded for the victory.
The one thing I really like about this system is that strength of schedule is built into it already. Scheduling and defeating a bunch of pansy teams won't be as valuable as the team defeating top competitors. The other strength of this format is that it does not favor any team. There are no overrated teams because the pollsters happen to really like Notre Dame and think that they're going to play for the championship because Charlie Weis is their coach, or underrated teams because TCU is from a small conference that isn't part of the BCS Coalition.
The weakness is that, like the BCS, you can't see which teams are standing out until a few weeks of data comes in and the good teams start to distance themselves in the rating system. I'm sure other people could come up with other flaws, but you can't really fault a system too much that ranks teams based solely on the wins and losses.
That's enough for now. I'll post some of my data as the week's progress.
Being an avid gamer and having played several games online in sactioned leagues, the most common ranking system used is done in what is called a "Ladder" format. Everyone begins with the same number of rating points, and as you play games against opponents, your rating adjust based on the rating of your opponent and whether you won or lost.
Why can't this be done for college football, where 119 teams are all trying to prove that they're worthy of playing for the National Championship at the end of the season? So after several years of nagging myself about it and never doing anything, this year I finally broke down and set it up.
The rules are as follows:
1) All teams begin the season with 1000 points.
2) Winning a game versus another team results in gaining 100 points, minus 10% of the difference in points when subtracting the losing team's rating from the winning team's rating. If the difference is negative (meaning the losing team had the higher rating) then the points are added.
3) Losing a game versus another team results in losing 100 points, minus 10% of the difference in points when subtracting the losing team's rating from the winningi team's rating. Again, if the difference is negative, the points are instead added.
4) Division I-AA teams count as having 0 points for the purpose of calculation.
5) Defeating a team by 21 or more points results in a bonus of 15 rating points. The losing team is not penalized for this.
Example 1: Team A (1000 pts) def. Team B (1000 pts) by a score of 10-3.
Team A = 1000 pts + (100 pts - 10% of 0 pts) = 1100 pts
Team B = 1000 pts - (100 pts - 10% of 0 pts) = 900 pts
Example 2: Team A (1250 pts) def. Team B (800 pts) by a score of 49-10.
Team A = 1250 pts + (100 pts - 10% of 450 pts) + 15 bonus pts = 1320 pts.
Team B = 800 pts - (100 pts - 10% of 450 pts) = 745 pts
Example 3: Team A (900 pts) def. Team B (1200 pts) by a score of 20-17.
Team A = 900 pts + (100 pts - 10% of -300pts) = 1030 pts
Team B = 1200 pts - (100 pts - 10% of -300pts) = 1070 pts
In the above examples, it demonstrates how many points are awarded or lost by each team based on the rules I included. In example 1, the teams have identical point values, and because of that the points awarded/lost are exactly 100 points. In example 2, the teams are 450 points apart in rating, with the winning team having the higher rating, so 10% of the rating difference is subtracted from 100, so the team is only awarded 55 points for the win, however 15 points are added for the decisive victory of 21 or more. In example 3, the teams are 300 points apart in rating, however the winning team has the lower rating, so 30 points are added to the 100, for a total of 130 points awarded for the victory.
The one thing I really like about this system is that strength of schedule is built into it already. Scheduling and defeating a bunch of pansy teams won't be as valuable as the team defeating top competitors. The other strength of this format is that it does not favor any team. There are no overrated teams because the pollsters happen to really like Notre Dame and think that they're going to play for the championship because Charlie Weis is their coach, or underrated teams because TCU is from a small conference that isn't part of the BCS Coalition.
The weakness is that, like the BCS, you can't see which teams are standing out until a few weeks of data comes in and the good teams start to distance themselves in the rating system. I'm sure other people could come up with other flaws, but you can't really fault a system too much that ranks teams based solely on the wins and losses.
That's enough for now. I'll post some of my data as the week's progress.
Comments:
Post a Comment
