openskill 5.0.0

Multiplayer Rating System. No Friction.

A faster and open license asymmetric multi-team, multiplayer rating system comparable to TrueSkill.

Description

Here are some, but not all, of the reasons you should drop TrueSkill and bury Elo once and for all:

• Multiplayer.
• Multifaction.
• Assymetric faction size.
• Predict Win, Draw and Rank Outcomes.
• 150% faster than TrueSkill.
• 100% Pure Python.
• 100% Test Coverage.
• CPython and PyPy Support.
• 5 Separate Models.
• Fine-grained control of mathematical constants.
• Open License
• Up to 7% more accurate than TrueSkill.

Installation

pip install openskill

Usage

The official documentation is hosted here. Please refer to it for details on how to use this library.

Limited Example

>>> from openskill.models import PlackettLuce
>>> model = PlackettLuce()
>>> model.rating()
PlackettLuceRating(mu=25.0, sigma=8.333333333333334)
>>> r = model.rating
>>> [[a, b], [x, y]] = [[r(), r()], [r(), r()]]
>>> [[a, b], [x, y]] = model.rate([[a, b], [x, y]])
>>> a
PlackettLuceRating(mu=26.964294621803063, sigma=8.177962604389991)
>>> x
PlackettLuceRating(mu=23.035705378196937, sigma=8.177962604389991)
>>> (a == b) and (x == y)
True

References

This project is originally based off the openskill.js package. All of the Weng-Lin models are based off the work in this wonderful paper or are the derivatives of algorithms found in it.

• Julia Ibstedt, Elsa Rådahl, Erik Turesson, and Magdalena vande Voorde. Application and further development of trueskill™ ranking in sports. 2019.
• Ruby C. Weng and Chih-Jen Lin. A bayesian approximation method for online ranking. Journal of Machine Learning Research, 12(9):267–300, 2011. URL: http://jmlr.org/papers/v12/weng11a.html.

Implementations in other Languages

• Javascript
• Elixir
• Kotlin
• Lua