Skip to content
How You Rank
Sign in
All ranking methods
Uncertainty-aware

Glicko-2

coming soon scales globally

Elo plus a confidence band and a volatility term. A newcomer sits at 1500 ± 350 and moves fast; a regular at 1800 ± 50 barely twitches. Time away widens the band again.

R ± RD with volatility σ; converges adaptively

Best at

Ladders where people come and go, so returning players re-find their level quickly.

Where it misleads

Two sides only, and the volatility parameter needs tuning for your pool before the numbers behave.

How it works

Glicko-2 extends Elo with two additional parameters: Rating Deviation (RD, measuring uncertainty) and Volatility (measuring consistency). It converges faster for new players and grows uncertain when players are inactive.

How it works: Like Elo, but with a confidence interval. A new player might be 1500 ± 350, while an established player is 1800 ± 50. Rating changes are larger when uncertainty is high, allowing the system to quickly find a new player's true skill. Inactivity gradually increases uncertainty.

When to use it: For competitive ladders where players come and go. Glicko-2 handles uneven activity gracefully -- a player who hasn't played in months will see faster rating adjustments when they return.

Watch out for: More complex to implement and explain than Elo. The volatility parameter (tau) needs tuning for your player pool. Only supports 1v1 and team-vs-team.

Works well with

Glicko-2 — How You Rank