This list is based off of a slightly modified Elo Algorithm (Used for ranking everything from chess, to organized sports), into which I fed all of the match results from The Blue Alliance, and Filtered by teams at IRI.
EDIT: The only modification to the algorithm, was to allow team matches, as the original Elo was designed for 1 on 1 only.
Keep in mind that these results are strictly as shown by the program, also, Every team above the line made it at least to their division finals. 7 out of the 13 teams above the line also went on to Einstein field for the semi-finals. Four of the teams made it to finals on Einstein.
##################### Teams that went to division finals ^^
The program is written in python, if anyone would like to know their specific team rating and are not listed, or would like to see the source code, I can send you either.
Do you have results before the IRI filter?
Could you explain the Elo Algorithm?
I think this might be a little flawed when you consider it includes elimination matches (which get progressively harder so of course the world champions / finalists are on top since they beat harder players).
kstl99: Yes, I have results for every team that attended The Championships.
XaulZan11: The Elo Algorithm essentially assigns each team a number representing their skill, the higher the skill number the better the skill (theoretically). When two teams play each other, it determines the skill number “payoff” for each team playing based on their chances of winning statistically, and changes their skill based on that.
Chris is me: The teams in the finals do play teams that are higher in skill, however, because both teams in the finals are high in skill, they don’t receive a disproportionate share of points. The Teams that play through to the finals do get more chances to play, but because they are equally matched, the skill numbers should remain indicative of the skill. For example, if a team with very high skill beat a team with low skill, they would receive very low bonuses because it is a expected result.
The main reason I’m skeptical is that Elo is a system designed for two-player games, and Breakaway is a 3v3 game. In addition, Elo is based on a very important assumption: that performance can’t be measured absolutely; it can only be inferred from wins, losses and draws. By its very nature, it doesn’t take into account the score of the game, or anything other than wins and losses. For chess, this is fine, as there is no metric of showing the strength of a win, but in Breakaway there is one so not using it will only produce a ranking system that turns out results similar to the finals.
Elo would be better in my opinion for ranking alliances, and then I would rather look at things like score differentials, etc. to determine relative strength of alliances in a somewhat objective manner. I am far from an expert on ranking systems, though, and if I’m wrong I’d love to hear why.
The Elo system was originally designed for chess rankings, however it is very similar to the current system used by XBox live to match people to games that are similar in skill. The skill difference assumes that the only relevant piece of information is the win or loss record. The strength of the win would be important if you look at a single game and say “Which team is better”, but in the long run a win/loss record should suffice if you have the relative strengths of the teams. Obviously if the strategy used by a team doesn’t fit well with others, then it’s skill rating being high is irrelevant. However, the nature of the statistics makes it unlikely that a team that doesn’t do well in points, or doesn’t do well with others, will come out ahead with a high score. The system I used is to treat each match won or lost as a group of 1 on 1 matches, and is the system used by many other implementations of Elo used for team matches. The XBox Live TrueSkill is slightly different, but it’s implementation is a project which takes more free time than I have :D.
Example: If a weak team gets a “free ride” with higher rated teams, it’s skill will go up. However this is negligible because with the sheer amount of teams that play, it balances out in the long run with probability.
I’d like to see the source code. Thanks.
Seconded, also curious how you would feel on me using said code in a project.