Chess Rating Systems Explained
Every major chess rating system — Elo, Glicko, and Glicko-2 — works on the same basic idea: win against someone rated higher than you and your rating goes up more; lose to them and it goes down less. But the systems differ in how they handle uncertainty and how quickly ratings move, which is why FIDE, USCF, Chess.com, and Lichess all give the same player noticeably different numbers.
The Elo system: the original
The Elo rating system was devised by Arpad Elo, a physics professor and chess player, and was adopted by FIDE (the World Chess Federation) in 1970 to replace an older, cruder ranking system. Elo's core idea is simple and elegant: your rating is a prediction of your expected score against any given opponent, based on the rating difference between you.
If you beat someone rated much higher than you, your rating jumps significantly, because that result was unlikely under the current ratings. If you beat someone rated much lower, your rating barely moves, because that result was expected. Draws split the difference. FIDE and USCF (the U.S. Chess Federation) both use variations of the Elo system for classical over-the-board ratings, which is why FIDE and USCF numbers, while calculated slightly differently, are conceptually the closest to each other of any two systems in wide use.
Glicko and Glicko-2: adding uncertainty
The Glicko system, developed by statistician Mark Glickman, extends Elo with one key addition: a "ratings deviation" — essentially a confidence interval around your rating. A brand-new player has a high ratings deviation because the system does not yet know much about them, so their rating can swing hard on just a few games. An established player with hundreds of rated games has a low ratings deviation, so their rating moves more slowly and predictably game to game.
Glicko-2 refines this further by also tracking a "volatility" measure — how erratically a player's results have been swinging recently — which lets the system react faster to a player who is genuinely improving or declining, rather than treating every game as equally informative. Chess.com uses a Glicko-based system for its ratings, while Lichess uses Glicko-2.
The practical upshot of both refinements is the same: your very first rated games matter far more to your number than games 400 through 410 do, because the system is explicitly less confident early on. This is also why an experienced player returning after a long break sometimes sees their rating move faster than they expect — the system re-widens its uncertainty during long inactivity, since it no longer knows if your strength has changed.
Elo vs. Glicko vs. Glicko-2 at a glance
| System | Used by | Key feature |
|---|---|---|
| Elo | FIDE, USCF (classical) | Expected-score model based on rating difference; no built-in uncertainty tracking. |
| Glicko | Chess.com | Adds a "ratings deviation" — a confidence range that shrinks as you play more games. |
| Glicko-2 | Lichess | Adds a "volatility" measure on top of Glicko, reacting faster to a player who is streaking or slumping. |
The practical effect: newer accounts on any Glicko-based platform can see bigger rating swings early on than an equivalent FIDE-rated newcomer would, because the system is explicitly modeling its own uncertainty about you.
Why the same player gets different numbers everywhere
This is the question almost everyone eventually asks: "I am 1400 on Chess.com, so why am I only 1200 on Lichess, and would I even be near that in a FIDE tournament?" A few real reasons:
- Different formulas — Elo, Glicko, and Glicko-2 update ratings differently, so identical results produce different numbers.
- Different player pools — each platform's population of opponents has a different overall skill distribution, and rating is always relative to who you actually played.
- Different time controls and conditions — bullet, blitz, rapid, and classical ratings are tracked separately and do not transfer directly, and online play removes some of the pressure and preparation that over-the-board FIDE play involves.
As a rough rule of thumb, Lichess ratings typically run somewhat higher than Chess.com ratings for a comparable level of actual skill, partly due to Glicko-2's faster-moving nature and partly due to pool differences — but "somewhat higher" varies by player and is not a fixed offset you can apply universally. Comparing your rating on one platform to your rating on another as if they were the same scale is one of the most common misunderstandings among improving players.
FIDE and USCF vs. online ratings
FIDE and USCF ratings come from classical, slow, over-the-board tournament games with real preparation, real nerves, and real consequences — a fundamentally different experience than online rapid or blitz. A player's online rating on Chess.com or Lichess is not a reliable predictor of what their FIDE rating would be, even on the same platform's "rapid" pool, because the populations and playing conditions differ so much. Many strong online players have never played a single FIDE-rated game, so there is no official conversion between the two — only rough, unofficial estimates that vary widely.
If your goal is genuine improvement rather than chasing a specific platform's number, focus on the fundamentals that transfer across every system — see our full how to improve at chess guide, or the level-specific paths from 1200 to 1600 and 1600 to 2000.
The one part of the system that is fixed: titles
While platform-to-platform comparisons are messy, FIDE title thresholds are one of the few genuinely fixed reference points in chess ratings, because they are defined directly in FIDE's Elo system rather than any online platform's pool. The main ones: a FIDE rating of roughly 2300 plus the required norms earns the FIDE Master (FM) title; 2400 plus norms earns International Master (IM); and 2500 plus norms earns Grandmaster (GM), chess's highest title. These numbers do not shift with platform, player pool, or time control the way online ratings do — they are tied to the single, unified FIDE rating list.
Because these thresholds are fixed and well known, they make a useful, stable yardstick when you want to describe how far "up" the rating scale a number sits, even if you personally play mostly online rather than in FIDE-rated events.
The practical takeaway
You do not need to fully understand the math behind Elo, Glicko, and Glicko-2 to use your rating well — you mainly need to remember three things. First, your rating is only meaningful relative to the specific system and pool that produced it, so do not directly compare a Lichess number to a Chess.com number or a FIDE number without expecting a gap. Second, newer accounts on Glicko-based systems (Chess.com, Lichess) will see bigger swings early on because the system is explicitly modeling its own uncertainty about you — that is normal, not a sign anything is broken. Third, if you want a fixed, cross-platform reference point, use the FIDE title thresholds (2300 / 2400 / 2500) rather than trying to eyeball a conversion between platforms.
It also helps to remember that every one of these systems, however it is calculated, is trying to answer the same underlying question: given everything known about your results so far, what score would you be expected to get against a given opponent right now? Elo, Glicko, and Glicko-2 differ in how quickly and how confidently they update that estimate, not in the basic question they are trying to answer. Once that clicks, differences between platforms stop feeling like a mystery and start feeling like exactly what they are — different amounts of statistical caution applied to the same idea.
Frequently Asked Questions
What is the difference between Elo and Glicko rating systems?
Elo estimates your rating purely from the expected score against your opponent's rating. Glicko adds a "ratings deviation" — a confidence interval that is wide for new or inactive players and narrow for established ones — so newer accounts can see bigger swings while established ratings move more steadily. Glicko-2 adds a further "volatility" measure that reacts faster to a player who is clearly improving or declining. FIDE and USCF use Elo-based systems; Chess.com uses Glicko; Lichess uses Glicko-2.
Why is my Lichess rating higher than my Chess.com rating?
This is common and expected. Lichess uses Glicko-2, which tends to produce somewhat higher numbers than Chess.com's Glicko-based system for a comparable level of play, and the two platforms also have different overall player pools. Neither number is more "correct" — they are different scales measuring you against different populations with different math, so a direct comparison is not meaningful.
Is a Chess.com rating the same as a FIDE rating?
No. Chess.com ratings come from online games, often at fast time controls, using a Glicko-based system, while FIDE ratings come from classical over-the-board tournament games using an Elo-based system. The two are calculated differently, drawn from different player pools, and reflect different playing conditions, so there is no official conversion — only rough, widely varying unofficial estimates that should not be treated as precise.
Who invented the chess rating system?
The Elo rating system was devised by Arpad Elo, a physics professor and chess enthusiast, and FIDE officially adopted it in 1970. The Glicko and Glicko-2 systems, used by Chess.com and Lichess respectively, were developed later by statistician Mark Glickman as a refinement that explicitly accounts for how much uncertainty the system has about a given player's true strength at any point in time.
