Skip to main content

Implementation of the TrueSkill, Glicko and Elo Ranking Algorithms

Project description

This is a Python port of the Moserware.Skills project that’s available at

For more details on how the algorithm works, see

To install run the command:

pip install skills

The match quality function of TrueSkill will run much faster with NumPy than with the provided matrix implementation. Install with:

pip install numpy

For details on how to use this project, see the accompanying unit tests with this project. You can run the tests by running the commands:

# test all calculators
python -m unittest discover

# test just the elo calculator
python -m unittest skills.testsuite.test_elo

Calculator Objects

These objects should be created and passed into the calculators. Most of these objects will also except python tuples or lists and automatically create the correct objects.


Player is an object with a player_id (anything that is hashable) and some partial play info. Partial play is used for TrueSkill only.:




Rating is an object with a mean. GaussianRating includes a stdev and is used for the TrueSkill and Glicko calculators. EloRating includes a k_factor and is used for the Elo calculator.:


GaussianRating(25.0, 8.333)

EloRating(1200, 32)

RatingFactory creates a new Rating object of whatever type is needed. RatingFactory.rating_class can be set to the Rating class desired. Instantiating one of the calculators will set RatingFactory.rating_class automatically.:

RatingFactory.rating_class = GaussianRating
RatingFactory.ensure_rating((25.0, 8.333))


Team is a dictionary of Player objects mapped to Rating objects. The objects keys method maps to players, values maps to ratings and items maps to player_rating. The constructor can take a dictionary of player to ratings or a list of player, rating tuples to create a multi-player team.:

Team({1: (25.0, 8.333),
      2: (25.0, 8.333)})

Team([(1, (25.0, 8.333)),
      (2, (25.0, 8.333))])

The Team object has convenience functions to find a player or rating by the Player object’s player_id property.:



Match is a list of teams and a ranking for each team. It inherets from list and includes a rank property, so regular lists can not be substituted.:

Match([Team1, Team2], [1, 2])

The constructor is a convenience function that will call ensure team on each team object passed in. This allows for easy object construction.:

Match( [(Player1, Rating1),
        (Player2, Rating2)],
       [1, 2] )

The Match object has convenience functions to find a player or rating in any Team object by the Player object’s player_id property.:


Match is synonomous with teams.


Matches is a list of Match objects. It inherits from list and a regular sequence type can be substituted for it.:

Matches([Match1, Match2])

The constructor is a convenience function that will call ensure_match for each object in the list. This allows for easy object construction.:

Matches([ ([Team1, Team2], [1, 2]),
          ([Team2, Team3], [1, 2]) ])

The following is syntax that uses only tuples and lists to generate a list of Matches::

Matches([([[(1, 1200)],
           [(2, 1200)]],
          [1, 2]),
         ([(2, 1200),
           (3, 1200)],
          [1, 2])])

[Match([{Player(1): Rating(1200.0)}, {Player(2): Rating(1200.0)}],
        rank=[1, 2]),
 Match([{Player(2): Rating(1200.0)}, {Player(3): Rating(1200.0)}],
        rank=[1, 2])]

Project details

Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Files for skills, version 0.3.0
Filename, size File type Python version Upload date Hashes
Filename, size (37.2 kB) File type Source Python version None Upload date Hashes View
Filename, size skills-0.3.0-py2.py3-none-any.whl (33.5 kB) File type Wheel Python version py2.py3 Upload date Hashes View

Supported by

AWS AWS Cloud computing Datadog Datadog Monitoring Facebook / Instagram Facebook / Instagram PSF Sponsor Fastly Fastly CDN Google Google Object Storage and Download Analytics Huawei Huawei PSF Sponsor Microsoft Microsoft PSF Sponsor NVIDIA NVIDIA PSF Sponsor Pingdom Pingdom Monitoring Salesforce Salesforce PSF Sponsor Sentry Sentry Error logging StatusPage StatusPage Status page