This was in an email from Dog Soldier to the participants of the Red Victory SB tourney:

I will try to explain ELO used at the blitz. If math and statistics make your head hurt, read this at your own peril. :-D

ELO at the blitz is based on the FIDE (Federation Internationale des Echecs) or World Chess Federation for those of you not familiar with the French French acronym system of scoring.

It is a complex, mathematical model that assigned everyone a base score at start. At the blitz IIRC we use 2,000 for new members on a ladder. The math works like this:

Rating Change = K-factor * ( Result – Expected Result )

Now that is how it works in chess. In chess there are only three levels of victory, win, lose or draw. They are assigned values of 1.0, 0.0, and 0.5 respectively.

At the blitz with our many levels of victory this becomes more complicated and I will not be going into the details, because frankly this was set up ages ago by a web master who has been gone for years. His code is buried in the mines of Moria here at the blitz. How exactly the script works is something I have not gone dungeoning here at the blitz to determine.

If anyone wants to understand the FIDE ELO system, the following link is a good one IMHO.

http://www.chesselo.com/index.html
If you can follow the math and explanation at that site, (do not blame me if you get a headache from this) then you can see that adapting the chess ELO system to the blitz would really be a matter of interpretation by the fellow who did it. He was a smart cookie because it works pretty good. The blitz does not host simple games like chess, so it is easy to say our ELO might be a bit more complex.

Bottom line, ELO at the blitz can give a rough estimate of a players skill level when compared to other players ELO in the same ladder. I say rough, because unlike chess every scenario in a war game is different at start and certainly none are as balanced as chess is at start of a chess game. So use ELO with caution, knowing that playing style and experience are nearly impossible to quantify.