OK... so after hours of math, calculator-punching, and me whining about not wanting to remove the possibility of a negative score for winning, we've decided to use this system. It's better to read the whole thing as it's structured as a short lesson and key variables are introduced throughout it:

CEA Ranking System

This ranking system is heavily based off the Elo System used in Chess. However, modifications have been made for the following reasons:

1) Match-ups are not predetermined
In chess, the match-ups are decided beforehand, or at least through a system, meaning the players can't choose whether or not to play a certain player in fear of the risk to their Elo Rating dropping. However, at CEA, we CAN choose which team to war, meaning that teams are more likely to refuse wars against teams that are much lower-ranked. This will be accommodated by modifying the formula used for updating the ratings.

2) Wars on CEA aren't just straight W/L
When you win, it's not just a 1-0 thing; the score should be taken into account. Winning 8-0 on a team should get more points than 8-7.

So now here we go with the math:

We'll start by defining these terms:

Code:

R_W = Initial Rating of winning team
R_L = Initial Rating of losing team

From there, we can get the difference in rating, dR:

Code:

dR = R_W - R_L

E_W and E_L are the same as in the Elo Ratings:

Code:

E_W = 1/{1 + 10^[(R_L - R_W)/400]}

The expected score of the winning team, which ranges from 0 to 1. This corresponds to the chance that the winning team had of winning a match in a war.

Code:

E_L = 1/{1 + 10^[(R_W - R_L)/400]}

The expected score of the losing team, which ranges from 0 to 1. Likewise, this corresponds to the chance that the losing team had of winning a match in a war.

E_W and E_L are related by the equation:

Code:

E_W + E_L = 1

The number of games the winning team was expected to win, X_W, is found by the following equation:

Code:

X_W = G(E_W)

Where G is the number of games played.

This leads to the equations:

Code:

R_W' = R_W + K(W - X_W)
R_L' = R_L - K(W - X_W)

Where R_W'(R_L') is the new Rating of the winning(losing) team, W(L) is the actual number of wins(losses) for the winning team. For now, we've set the K-Factor, aptly labeled K, to 15 to allow for more varied shifts from the average. It MIGHT change in the future to a lower number once things settle down and/or the number of teams increases to an acceptable amount. Most places use smaller numbers like 12. This is still the same old equation used in the Elo System, just with a substitution. As mentioned earlier, this deters higher-ranked teams from playing against lower-ranked teams.

So to appease you people, the equations have been modified (using another forum's rating system) into:

Code:

R_W' = R_W + K*Log2[W-L+1](1 - E_W)
R_L' = R_L - K*Log2[W-L+1](E_L)

With this, it's now mathematically impossible to lose points for barely beating teams you're supposed to rape, beat, and murder, repeatedly. Not ideally, not normally, but MATHEMATICALLY IMPOSSIBLE. So go ahead and beat on those lower-ranked teams, you bottom-feeders. I'm pretty sure that's all you wanted to do...

This is how I enter the stuff into the calc (using the most recent completed war as an example):

All I do is enter the initial ratings and the result, and it spews out the rest. Mathematica... it works.

Current Rankings can be found here:
https://docs.google.com/spreadsheet/ccc?key=0ArM63L8AWf7OdHVTTXg1YlVyb2cydGlzc0U3NnBUd2c#gid=1

Rankings, current rating, and ongoing wars are all listed there. Team acronyms can be changed on request. I'd do the individual matches too... but I need the matches in chronological order, since it matters in the ratings. That includes the wars conducted before the move from NEDA.

*The math used can be understood by anyone who's finished Algebra II. If you don't understand it, you're too young or should go back to school*

-------------------------EDIT-----------------------------------------
Here's a picture of how the rating of the winning team changes if they won an 8 heart war with an E_w of 0.6 (dR of about 70), depending on the amount of losses they had:


So while blowouts (the left side) aren't rewarded as much, all it means is you just need to play more to gain the separation the old Elo Formula would bring. Besides, you're always gaining points with a win - pretty sure you don't like that right side of the Elo curve - so consider it a sacrifice for that positive score. Also, a larger sample size (more wars) will provide a better sense of which teams are really better than the others.

If there are any questions/comments/requests - about the system or just about how results are displayed - feel free to PM me or send me a message on Skype (Chaossoldier92) or MSN (Soldierofchaos@live.com). It's probably best if you do at least 2 of the 3... I'm rather oblivious.

Taisakuno Shouri vs Weedy: 2-1
Taisakuno Shouri vs shorifio4321: 2-0
mouzou vs riou: 2-0
Taisakuno Shouri vs KrisKirk: 2-1
Acnologia vs challa: 2-1
Hostile vs SunMoonOR : 2-0
Esfo vs Hostile: 2-0
Esfo vs Light..: 2-0
Esfo vs ketsueki: 2-1
Esfo vs Acnologia: 2-0
Esfo vs Taisakuno Shouri: 2-1

dubina vs. ketsueki: 2-1
mouzou vs dubina: 2-1
Hostile vs dubina: 2-0
NickV vs Hostile: 2-1
LordMelchett vs ketsueki: 2-1
Sabbaticus vs Hostile: 2-0
LordMelchett vs riou: 2-0
Sabbaticus vs Taisakuno Shouri: 2-1
Taisakuno Shouri vs NickV: 2-1
mouzou vs NickV: 2-0
Sabbaticus vs riou: 2-0
Sabbaticus vs Taisakuno Shouri: 2-1