Each player has a rating and an uncertainty score.
The rating is the system’s estimate of a player’s strength. The uncertainty score is called rating deviation, or RD. A new player starts with a high RD because there are not enough results to know how accurate the initial rating is.
Every completed match updates both numbers.
Winning raises your rating and losing lowers it. The size of the change depends on your opponent’s rating and RD. Beating a stronger player raises your rating more than beating a weaker player. Losing to a stronger player lowers it less.
Ratings become more stable as you play.
As you complete more matches, RD falls and the system becomes more certain about your rating. New results still change the rating, but a single match has less influence. If you stop playing for a while, RD rises because an older rating is less certain.
The leaderboard uses a conservative score.
Poker has more short-term variance than chess. To avoid placing a player with a high but uncertain rating at the top, the leaderboard ranks a score we call Skill.
Skill = Rating − 2 × RD
A lower RD produces a Skill score closer to the underlying rating. Players remain provisional until their RD falls below 75. Every completed match updates both monthly and all-time ratings.
Why use a public rating?
A rating gives each completed match context and gives players a reason to return and improve. It also makes the public table a more useful place to practice decisions against a range of opponents.