Mike Abberton Wrote:ELO, as implemented at the Blitz, also has the flaw of not taking into account the "balance" of the scenario. It's obviously less of a problem in chess where the only imbalance is who goes first.
For example, if you play an unbalanced scenario from the "losing" side against a newer player or one who is less skilled than you are (in an effort to "level" the playing field some), than you will take a hit in ELO if you lose the unwinnable scenario.
Now, I am in no way suggesting that we should, or even could, try to incorporate balance into the equation (way too subjective). Just another caution that ELO is not a perfect judge of relative skill levels. A useful tool, sure, but not definitive.
'Tis true!
I did spend a number of brain cycles on how to implement a "balance" argument to our ELO. I actually came up with a workable solution. However, prior to my involvement a bug had been introduced into the scenario balance rating system which had "flipped" the meanings of the ratings, such that pro-Axis became pro-Allied and vice versa.
Many scenarios had been rated using both meanings, and as such their "Balance" ratings were out of whack (as it were).
We fixed that bug but there was so much bad data that was unknowable (you know, a unknown known unknown) that I decided not to incorporate "balance" into ELO.
The statistical argument is that if enough games are played the balance should not have a significant effect as it will even out between players. Of course, there are those masochists among us who prefer taking the underdog...
umbro