Converter / Rating Systems

How Chess Rating Systems Work: Elo, Glicko, and Glicko-2

A chess rating is a single number that estimates your playing strength from your game results. You gain rating when you beat opponents rated above you, and you lose more when a lower-rated opponent upsets you. The three systems you will meet online are Elo, Glicko, and Glicko-2. They share that basic idea but differ in how much they trust your current number, which is exactly why the same rating can mean different things on different sites.

What a chess rating actually measures

Think of a rating as a running best guess. Every game is a small experiment: it tests whether your number is too high, too low, or about right, and the system nudges it based on the outcome. Beat someone far above you and the system was clearly underestimating you, so your rating jumps. Beat someone you were expected to beat and it barely moves, because the result told the system nothing new. Over many games the number settles near your true strength.

Because the whole point is comparison, a rating only has meaning inside the group of players it was calculated against. That group is called a rating pool, and it is the key to understanding why numbers do not transfer cleanly between platforms.

How Elo works

Elo is the classic system, devised by Arpad Elo, a physics professor and chess master. FIDE adopted Elo ratings in 1970, and the name has been shorthand for "chess rating" ever since.

The core idea is elegant. Each gap between two ratings implies an expected score: a favorite is expected to score more than half, an underdog less. After the game, your rating moves in proportion to the difference between your actual result and that expected result. Do exactly as well as expected and nothing changes. Overperform and you gain; underperform and you lose. The larger the surprise, the larger the swing.

Elo's limitation is that it treats every player's number as equally trustworthy. A grandmaster who has played thousands of games and a newcomer who just joined are both nudged by the same kind of arithmetic, even though the system has far less certainty about the newcomer. That is the gap Glicko was built to close.

How Glicko and Glicko-2 differ

Glicko was created by statistician Mark Glickman. Its central addition is rating deviation, usually written as RD, which is a measure of how uncertain the system is about your rating right now. A brand-new player, or one returning after a long break, carries a high RD, so their rating swings in big steps as the system quickly hunts for the right level. A player with a long, steady history carries a low RD, so their rating barely twitches from any single game. In other words, Glicko does not just track your strength; it tracks how sure it is about that strength.

Glicko-2 is Glickman's refinement. It adds a volatility term that captures how erratic a player's results have been. A player whose recent scores bounce around unpredictably is assigned higher volatility, which lets their rating adapt faster, while a consistent player is treated as more stable. Together, RD and volatility make the system responsive to new information without overreacting to a single lucky or unlucky game.

This is why a fresh account can climb or fall so quickly at first and then seem to lock in place. Early on, the deviation is high and every result moves the number a long way as the system searches for your level. Once you have played enough games for the deviation to shrink, the same win or loss barely registers. The rating has not stopped responding to you; the system has simply grown confident, so it now demands more evidence before it revises its estimate.

What each platform uses

Lichess uses Glicko-2. A brand-new Lichess rating starts at 1500 plus or minus 1000, where that 1000 is the 95% confidence interval around the estimate. While the system is still uncertain about you, your rating is marked provisional and shown with a question mark; it stays provisional until its deviation drops to 110 or below. Lichess also assigns time controls by estimated game duration rather than by the raw clock alone: 179 seconds or less counts as Bullet, 479 or less as Blitz, 1499 or less as Rapid, and 1500 or more as Classical.

Chess.com uses a Glicko-based system with rating deviation, and it keeps separate ratings for each format such as bullet, blitz, rapid, and daily. When you make a new account, you choose an approximate skill level, and that choice sets where your rating starts so the system does not have to discover your level entirely from scratch.

FIDE, the international chess federation, maintains the official over-the-board Elo ratings. These are the ratings earned in in-person tournament play, and FIDE publishes them as separate standard, rapid, and blitz lists.

Why the same number means different things

Each platform is a separate rating pool. It has its own starting points and its own player population, so a 1600 on one site is not the same strength as a 1600 on another. The numbers were never calibrated against each other, and no amount of staring at them will make them line up.

The honest way to compare across sites is percentile matching: find where a rating sits within its own pool, then find the rating that holds the same position in the other pool. That is exactly what this site does, and you can read the full approach on our methodology page.

Ready to translate a real number? Try the converter on the home page, or jump straight to the common case of going from Chess.com to Lichess.