[messages] [Feature Requests] ELO player ratings for two-player games?
twixtfanatic at gmail.com
Sun Feb 2 17:59:29 CET 2014
The ELO rating system provides a way to determine a player's strength
at a two-player game of skill. It was created by the physics professor
Arpad Elo. If everyone using the Vassal server had a unique
identification, then for each module which can be played by just two
players, a record of the game results could be kept, and a database of
player ratings could be built up.
There already is a website that has a player ratings database for
Avalon Hill games, including games with more than two players. I'm not
sure if the results from Vassal or other virtual games are included. A
much simpler formula is used there. Some players have commented on a
"bouncing around" effect which a specific rating might undergo before it
tends to settle within a more narrow range of values.
This would be a lot of work, but as time goes by, the database would
have increasing value. This is rating not ranking. A higher value
indicates a stronger player. Of course the accuracy of outcome
prediction would depend on the size of the database for that game. ELO
has a sound statistical basis, and is designed to provide accurate
values relatively quickly, which is to say, in general, fewer games need
to be played in order to get reasonably accurate ratings. Once you
obtain an established rating, regardless of what it is, all you should
need to know is the ratings difference between yourself and your
opponent in order to predict the likelihood of winning.
Another advantage that a Vassal database would have would be its
automation. No one would need to manually submit game results by email.
Players would agree between themselves, before the game starts, whether
it will be rated or not. The Vassal engine could provide buttons or a
popup window to submit the game result, whether the game was played on
the server or by email. Perhaps both players would be required to submit
in this fashion, in order to reduce the temptation to abuse the system.
As long as the players are registered, the server could calculate the
new ratings. A player's rating could be displayed next to their ID on
the server. Maybe there could be two sets of ratings for each game, one
for email play and one for real time play.
As far as I know, games with more than two players are not well suited
for the ELO system. But as long as a game could be played by just two
players, then ratings could be built up for those games. I should add
that two player games with more than three possible outcomes (player A
won, player B won, or draw), that is to say, games with degrees of
victory or loss, __can__ be incorporated.
What do you think?
Read this topic online here:
More information about the messages