System comparison
Elo vs Glicko
Elo and Glicko are the two most important rating systems in competitive chess, but they solve the measurement problem differently. Traditional Elo gives you a single number representing estimated strength. Glicko gives you that number plus a rating deviation (RD) that explicitly measures how confident the system is in the estimate. This fundamental difference explains why your FIDE rating and your Chess.com rating behave so differently — even when you are playing at the same level.
Elo: One Number, No Confidence Interval
The classic Elo system assigns every player a single rating number. When you play a game, the system compares the result to the expected score and adjusts the rating. The K-factor controls how fast the number moves. There is no built-in mechanism to express doubt about the rating — a player rated 1500 after 500 games looks identical to a player rated 1500 after 5 games. For a side-by-side view of the differences, see Why ratings differ across pools.
This simplicity is Elo's greatest strength and its most significant limitation. The number is easy to understand and communicate, but it cannot tell you whether the system has high confidence or is still guessing. For a side-by-side view of the differences, see FIDE vs US Chess Ratings.
Glicko: Rating + Rating Deviation + Volatility
Glicko (and its successor Glicko-2, used by Lichess and Chess.com) tracks three values for each player: rating, rating deviation (RD), and rating volatility. RD starts high for new players and decreases as more games are played. If a player stops playing, RD gradually increases again, reflecting growing uncertainty about their current strength.
This means Glicko can react more aggressively to results from uncertain players and more conservatively to results from well-established ones — automatically, without needing fixed K-factor bands. A returning player who has been inactive for months will see larger swings than an active player at the same rating, because the system is less certain about the inactive player's level.
Practical Differences You Will Notice
- After a break: In Elo, your first game back behaves identically to your last game before the break. In Glicko, the system increases your RD during inactivity, so your first games back produce larger rating swings.
- New account volatility: Glicko systems are intentionally volatile for new accounts, converging quickly. Elo with K=40 achieves something similar but less elegantly.
- Stability over time: Long-active Glicko players have very low RD, making their ratings extremely stable — often more stable than Elo with K=10.
- Cross-system comparison: A 1500 Glicko rating with RD=30 is a very confident measurement. A 1500 Glicko rating with RD=200 is essentially a guess. Elo has no way to express this distinction.
Why You Cannot Convert Between Them
Elo and Glicko ratings operate on the same general scale (a 200-point advantage produces similar expected scores in both systems), but the populations, update mechanics, and confidence handling create real divergences for individual players.
A player who is highly active on Lichess (Glicko-2) and moderately active in FIDE events (Elo) will naturally develop different numbers in each system. Neither is wrong — they are measuring the same player through different lenses against different opponents with different mathematical tools.