API

All raters extend the abstract Rater base class and share common options.

Shared options (`IRaterOptions`)

Option	Default	Description
`failHard`	`false`	Throw on errors instead of collecting them
`minRounds`	`3`	Minimum `moves.length` to rate a game
`respectUnrated`	`true`	Skip records with `header.unrated: true`

Shared result shape (`IRaterResults`)

interface IRaterResults {
    ratings: Map<string, IRating>;
    recsReceived: number;
    recsRated: number;
    warnings?: string[];
    errors?: string[];
}

Each IRating includes userid, rating, recCount, wins, losses, and draws.

ELOBasic

Sequential ELO rating with a configurable K-factor.

import { ELOBasic } from "@abstractplay/recranks";

const rater = new ELOBasic({
    ratingStart: 1200,
    K: (p1Rating, p1Games, p2Rating, p2Games) => 30,
    minRounds: 3,
});
const results = rater.runProcessed(records);

Option	Default	Description
`ratingStart`	`1200`	Initial rating for new players
`K`	`30`	K-factor function (receives both players' rating and game count)

Glicko2

Batch rating-period Glicko-2 via glicko2-lite.

import { Glicko2 } from "@abstractplay/recranks";

const rater = new Glicko2({
    ratingStart: 1500,
    rdStart: 350,
    volatilityStart: 0.06,
    tau: 0.5,
    knownRatings: existingRatingsMap,
});
const results = rater.runProcessed(records);

Option	Default	Description
`ratingStart`	`1500`	Initial rating
`rdStart`	`350`	Initial rating deviation
`volatilityStart`	`0.06`	Initial volatility
`tau`	`0.5`	System constant
`knownRatings`	empty	Seed ratings from a previous period

IGlickoRating extends IRating with rd and volatility.

Trueskill

Sequential TrueSkill via ts-trueskill.

import { Trueskill } from "@abstractplay/recranks";

const rater = new Trueskill({
    muStart: 25,
    sigmaStart: 8.333,
    betaStart: 4.166,
    tauStart: 0.083,
    drawProbability: 0.1,
});
const results = rater.runProcessed(records);

Option	Default	Description
`muStart`	library default	Initial mean
`sigmaStart`	library default	Initial uncertainty
`betaStart`	library default	Performance variance
`tauStart`	library default	Dynamics factor
`drawProbability`	library default	Draw probability

ITrueskillRating extends IRating with sigma.

Entry points

// Validate JSON strings and rate
rater.run(batch: string[]): IRaterResults;

// Rate pre-parsed records (no schema validation)
rater.runProcessed(batch: APGameRecord[]): IRaterResults;

Player IDs are keyed as {site.name}|{userid}.

API