Also, if you are a college football fan, I thnk the Sagarin rankings (one of the six BCS computer rankings) are somehow based on an Elo rating system.

Yeah thats right, just win/draw to raise it.

How come Decembers aren't out by the way?

The algorithm for expected result had to be changed since this is a seven player game and behaves in a way fundamentally different to chess.

I saw the teaser on my home page and thought you were saying that the December ratings aren't out yet because you had to change the algorithm!

I read it like that too Peter lol

Nope, the reason its not out is more mundane: kestas' code for getting the data out is broken (probably because you terrible lot have played too many games) and that needs fixing first.

Ah lol ok.

does the algorithm only require a player's previous ghost rating along with all the games they have played in the last month?

I would certainly hope so, since those would be sufficient.

does this rating system know if a game was joined half way, or had a CD, or is just "player1 won the game, his rating goes up accordingly"

http://sites.google.com/site/phpdiplomacytournaments/theghost-ratingslist there's a full explanation there :P

yea but that is full of equations and reading...

I don't think the system can take into account CD takeovers at this point.

But yeah, all you really need to know is that if you do well, your rating will go up, more so if you do well in a game with superior players than inferior ones. Conversely, if you do poorly, your rating goes down, but less so in a game with superior players than with inferior ones.

nothing more than a ranking of how often you post in the forum.

pay it no mind

"I don't think the system can take into account CD takeovers at this point."

Correct

"nothing more than a ranking of how often you post in the forum.

pay it no mind"

+1

say i (rank in the 2200ish) plays a game with top 1000 players, and loses terribly.
then i play a game with <4000er's, and win.
basically im asking which would benefit me more, losing a game against pros, or winning a game against noobs?

i believe losing a game will urt your rating, but against pros it will not hurt much.

beating noobs will improve your rating but again not by much... if you keep beating the same noobs then as their ratings worsen they will give you less and less each time you beat them (at least that i my understanding...)

Mongoose: I think this is a simple explanation that's also accurate.

Think of GR like points, except that each player "bets" 2/35 of their current GR. A low-rated player bets less than a high-rated player, and so they gain more points if they win or draw.

Whoa, pastor, that's a good way of putting it... and it makes me realize, that if the point system operated on a percentage bet rather than a flat bet, and you didn't reset points if you went under 100, then it would exhibit the same behaviour as GR, no?

^^ Pretty well. However, in GR you can't choose to have a game not count.

One clarification, you don't lose less if you play against higher rated players than you'd lose against lower rated players, if you are eliminated you lose the same amount (2/35 of your current GR) regardless, you are just more likely to lose against the higher rated players. This is also true in a WTA game if you survive, you will lose 2/35 of your current GR, but if you survive a PPSC game with at least one unit then you will lose less (or win more) playing better players since the total GR put into the game will be divided by SC count to figure out how much GR you get back after you threw in your 2/35 share... as I loosely understand it... Crap, that's a bit more confusing than I had hoped for...

Does anyone know how a new player is rated? The formula explains how the rating changes after a win or loss, but I don't think it explains the initial rating at the start of their first game.

@Devonian

You start with a certain Ghost Rating. I don't remember what it is off-hand (100?), but I believe it is mentioned on the site.

You start with a GR of 100. If you lose, the number of points you lose is determined solely by the number of points you have at the end of the game. If you win, the number of points you gain is determined solely by the sum of your opponents' points at the end of the game. If you draw, the number of points you gain or lose is determined by the size of the draw, the number of points you have, and the sum of your opponents' points. All this applies to WTA . . . PPSC is a bit more complicated.

Also note that how well any specific opponent did is irrelevant: only the result, your points, and the sum of your opponent's points matter.

Seems to me that it is all pretty simple. The amount you put into the ghost rating pot for the game is 2/35ths of your ghost ratings, at the time the game ends. The pot is then split the same way the dip points are split, WTA or PPSC.

Variants, public press, and gunboat games use a lesser amount of your ghost rating. 1/35th or even 1/70th I think. I don't recall that being spelled out on the webpage. And games that end in draws within the first two or three seasons don't affect ghost ratings.

@Alderian, it is more convoluted than that for PPSC. Also, wins in the early game (due to CDs) don't count either.

Ah right, if everyone CDs and one is therefore declared the winner, there can be an early game win.

As for PPSC, I didn't realize there was more to it than how PPSC normally splits the pot. Sorry for spreading misinformation.