Competitors

Elo Competitor

class elote.competitors.elo.EloCompetitor(initial_rating: float = 400, k_factor: float = 32)

Overview

Elo rating is a rating system based on pairwise comparisons. Ratings are given to competitors based on their comparisons (bouts) with peers, in which they can win, lose or tie. The change in a players rating is scaled by the rating difference between them and their competitor.

[1] Elo, Arpad (1978). The Rating of Chessplayers, Past and Present. Arco. ISBN 0-668-04721-6.

Basic Usage

from elote import EloCompetitor
good = EloCompetitor(initial_rating=400)
better = EloCompetitor(initial_rating=500)
better.beat(good)
print('probability of better beating good: %5.2f%%' % (better.expected_score(good) * 100, ))

Class Variables

Class variables are configured for all competitors in a population, not on a per-competitor basis. See the documentation on Arenas to see how to set these safely.

• _base_rating: defaults to 400.

• _k_factor: tunes the speed of response to new information, higher is faster response. Default=32

Configuration Options

Parameters:

initial_rating (int) – the initial rating to use for a new competitor who has no history. Default 400

beat(competitor: BaseCompetitor)

Takes in a competitor object that lost a match to this competitor, updates the ratings for both.

Parameters:

competitor (EloCompetitor) – the competitor that lost their bout

expected_score(competitor: BaseCompetitor)

In Elo rating, a player’s expected score is their probability of winning plus half their probability of drawing. This is because in Elo rating a draw is not explicitly accounted for, but rather counted as half of a win and half of a loss. It can make the expected score a bit difficult to reason about, so be careful.

Parameters:

competitor – another EloCompetitor to compare this competitor to.

Returns:

likelihood to beat the passed competitor, as a float 0-1.

export_state()

Exports all information needed to re-create this competitor from scratch later on.

Returns:

dictionary of kwargs and class-args to re-instantiate this object

tied(competitor: BaseCompetitor)

Takes in a competitor object that tied in a match to this competitor, updates the ratings for both.

Parameters:

competitor (EloCompetitor) – the competitor that tied with this one

Glicko Competitor

class elote.competitors.glicko.GlickoCompetitor(initial_rating: float = 1500, initial_rd: float = 350)

from http://www.glicko.net/glicko/glicko.pdf

class vars:

• _c: default 1

• _q: default 0.0057565

Parameters:

• initial_rating – the initial rating to use for a new competitor who has no history. Default 1500

• initial_rd – initial value of rd to use for new competitors with no history. Default 350

beat(competitor: GlickoCompetitor)

Takes in a competitor object that lost a match to this competitor, updates the ratings for both.

Parameters:

competitor (GlickoCompetitor) – the competitor that lost thr bout

expected_score(competitor: BaseCompetitor)

The expected outcome of a match between this competitor and one passed in. Scaled between 0-1, where 1 is a strong likelihood of this competitor winning and 0 is a strong likelihood of this competitor losing.

Parameters:

competitor – another EloCompetitor to compare this competitor to.

Returns:

likelihood to beat the passed competitor, as a float 0-1.

export_state()

Exports all information needed to re-create this competitor from scratch later on.

Returns:

dictionary of kwargs and class-args to re-instantiate this object

tied(competitor: GlickoCompetitor)

Takes in a competitor object that tied in a match to this competitor, updates the ratings for both.

Parameters:

competitor (GlickoCompetitor) – the competitor that tied with this one

DWZ Competitor

class elote.competitors.dwz.DWZCompetitor(initial_rating: float = 400)

class vars:

• _J: default 10

Parameters:

initial_rating – the initial rating to use for a new competitor who has no history. Default 400

beat(competitor: BaseCompetitor)

Takes in a competitor object that lost a match to this competitor, updates the ratings for both.

Parameters:

competitor (DWZCompetitor) – the competitor that lost thr bout

expected_score(competitor: BaseCompetitor)

The expected outcome of a match between this competitor and one passed in. Scaled between 0-1, where 1 is a strong likelihood of this competitor winning and 0 is a strong likelihood of this competitor losing.

Parameters:

competitor – another EloCompetitor to compare this competitor to.

Returns:

likelihood to beat the passed competitor, as a float 0-1.

export_state()

Exports all information needed to re-create this competitor from scratch later on.

Returns:

dictionary of kwargs and class-args to re-instantiate this object

tied(competitor: BaseCompetitor)

Takes in a competitor object that tied in a match to this competitor, updates the ratings for both.

Parameters:

competitor (DWZCompetitor) – the competitor that tied with this one

ECF Competitor

class elote.competitors.ecf.ECFCompetitor(initial_rating: float = 40)

class vars:

• _delta: default 50

• _n_periods: default 30

Parameters:

initial_rating – the initial rating to use for a new competitor who has no history. Default 40

beat(competitor: BaseCompetitor)

Takes in a competitor object that lost a match to this competitor, updates the ratings for both.

Parameters:

competitor (ECFCompetitor) – the competitor that lost their bout

expected_score(competitor: BaseCompetitor)

The expected outcome of a match between this competitor and one passed in. Scaled between 0-1, where 1 is a strong likelihood of this competitor winning and 0 is a strong likelihood of this competitor losing.

Parameters:

competitor – another EloCompetitor to compare this competitor to.

Returns:

likelihood to beat the passed competitor, as a float 0-1.

export_state()

Exports all information needed to re-create this competitor from scratch later on.

Returns:

dictionary of kwargs and class-args to re-instantiate this object

property rating

Returns:

tied(competitor: BaseCompetitor)

Takes in a competitor object that tied in a match to this competitor, updates the ratings for both.

Parameters:

competitor (ECFCompetitor) – the competitor that tied with this one

BlendedCompetitor

class elote.competitors.ensemble.BlendedCompetitor(competitors: list, blend_mode: str = 'mean')

Parameters:

• competitors –

• blend_mode –

beat(competitor: BaseCompetitor)

Takes in a competitor object that lost a match to this competitor, updates the ratings for both.

Parameters:

competitor (BlendedCompetitor) – the competitor that lost their bout

expected_score(competitor: BaseCompetitor)

The expected outcome of a match between this competitor and one passed in. Scaled between 0-1, where 1 is a strong likelihood of this competitor winning and 0 is a strong likelihood of this competitor losing.

Parameters:

competitor – another EloCompetitor to compare this competitor to.

Returns:

likelihood to beat the passed competitor, as a float 0-1.

export_state()

Exports all information needed to re-create this competitor from scratch later on.

Returns:

dictionary of kwargs and class-args to re-instantiate this object

tied(competitor: BaseCompetitor)

Takes in a competitor object that tied in a match to this competitor, updates the ratings for both.

Parameters:

competitor (BlendedCompetitor) – the competitor that tied with this one