Skip to content

HTTPS clone URL

Subversion checkout URL

You can clone with HTTPS or Subversion.

Download ZIP
Elo player rankings for multiplayer games
Python
branch: master

Fetching latest commit…

Cannot retrieve the latest commit at this time

Failed to load latest commit information.
.gitignore
LICENSE.txt
README.txt
arena.csv
groupelo.py
test_groupelo.py

README.txt

This code calculates Elo ratings (see
http://en.wikipedia.org/wiki/Elo_rating_system) for multiplayer games.

The data file format does not currently support draws, or games where the
non-winners need to be ranked rather than all lumped together as equal losers.
Adding support for those would be pretty easy, but I don't currently need
those features and I didn't want to complicate the data file.

I saw the idea for the the technique of treating a multi-player game as a
series of two-player games (winner beat each loser, losers drew each other,
scale down the K-value to prevent games with more players from counting more)
from a 2005 post on rec.games.trading-cards.jyhad by Frederick Scott.  It's
a pretty obvious and elegant idea so it probably predates that post by a few
decades.

I don't think the algorithm for team games is correct yet.  One's teammates'
ratings should factor in, since it's easy to win with strong teammates and
hard to win with weak teammates.
Something went wrong with that request. Please try again.