Skip to main content

Introduction

Idem provides matchmaking for multi-player games through an API. Connect, and deploy a full-fledged matchmaking and rating system in minutes. Instant access to the best-in-class algorithm, with everything pre-configured and ready to scale.

If you have more specific needs, we will help you customise the matchmaker to any possible use-case. Think: Team-games, parties, latency-requirements, regions, platforms etc.

The product consists of three core components:
Matchmaking, Rating, and Ranking

Matchmaking​

Performance: The heart of Idem is the matchmaker which operates a probabilistic algorithm. It finds optimal matches by looking into currently queueing players, and also predicting incoming players and their respective skill level. Hence, it can deliver better matches at shorter waiting times. The matchmaker ensures a predictable waiting time for players, based on the game studio's target.

Top-players: The matching algorithm is particularly good at catering to top-players. In most games, they are systematically at a disadvantage when it comes to waiting time, due to the outstanding rating they have. However, top-players are the most valuable players in any launch of a product, due to streaming and the importance of influencers on word-of-mouth. The design of the Idem matchmaker ensures reliable waiting time for these top players.

Ease of integration: The matchmaker is accessed through a simple API. The integration requires minimal exchange of data, which enables studios to onboard matchmaking without any deep integration or refactoring. Our backend is battle-tested and build entirely serverless, to ensure even the greatest overnight-success of a game can be handled. Don't worry about calibrating and adjusting the matchmaker - we automatically tune parameters for you. Your demand spikes or you release a new version? The matchmaking automatically adjusts to the demand for your game to deliver a consistent experience, no matter the situation.

Key features:

  • Any gamemode Supports any PvPvP mode with any number of players per team and number of teams, and potentially asymmetric game modes (talk to us!).

  • Parties: Players can join as parties and be matched.

  • Team balancing: The matchmaker will select players for team-games based on their skill but also on the quality of how the two (or more) teams balance against each other.

  • Latency-based matching: You can define latency requirements or geography constraints for players to be matched.

  • Custom criteria for matching: Any particularity in your game can be considered for matchmaking.

Rating system​

The following rating systems can be selected:

1. Bayesian rating (similar to e.g. Glicko-2, TrueSkill™️)
Note: Elo-rating is not supported because it is inferior to more modern Bayesian rating systems and cannot natively support games larger than 1v1.

Supported game outcomes:

  1. win/loss - binary game outcome
  2. score-dependent - "winning 5-0 better than 1-0"
  3. ranking-dependent - "finishing 2nd is better than 3rd" (e.g. car race and elimination race)

Additional features:

  • Team dynamics: Able to model team dynamics on a continuum from "weakest link" to "carry" type of games.
  • Game randomness: Able to model randomness in a game (luck vs skill) and a player's performance (good vs bad day).
  • Automated learning: Is tuned over time to find the best parameters to represent the true probability of the different events upon which ratings are based.
  • Staleness prevention: Prevents rating from becoming "stale" by enforcing rating uncertainty floor - we are never "too sure" so players can always climb rapidly enough if they improve.
  • Segmentable: Can be segmented within a given game mode to have a separate rating for each faction (or any other dimension upon which it makes sense to separate ratings) if needed.

Ranking system​

Activity incentives: Can trade off player skill (rating) with activity (number of wins) in a formula to keep players engaged if they want to rank high. Makes sure that continuing to play on is always incentivized, and "camping a rank" is not rewarded.

Customization: Can be customized in multiple ways to maximize engagement without having to risk the integrity of the rating system (e.g. no need for rating decay, or rating penalties).

Seasons: Custom seasons can be defined and players' ratings reset.