What Is Elo in Chess?
Arpad Elo's rating formula, the K-factor, FIDE vs USCF vs Lichess Glicko, the highest ratings in history, and four techniques to actually gain points. No signup. No paywall.
TL;DR — Elo Ratings in 5 Bullets
- Elo is a single integer between roughly 100 and 2900 that predicts your expected score against another rated player — invented by Arpad Elo, adopted by FIDE in 1970.
- The formula: expected score E = 1 / (1 + 10^((opp − you) / 400)), then new rating R' = R + K × (actual − E).
- The K-factor controls how fast your rating moves: 40 for new players, 20 for adults under 2400, 10 for titled players 2400+.
- FIDE, USCF, Lichess (Glicko-2), and Chess.com (Glicko-1) are all Elo or Elo-derived — but the pools are calibrated differently so cross-pool comparisons need a converter.
- All-time peaks: Carlsen 2882 (2014), Kasparov 2851 (1999), Fischer 2785 (1972).
What Is Elo?
The Elo rating is a single integer that estimates a chess player's strength. It was invented by Hungarian-American physics professor Arpad Elo in the 1950s to replace the older Harkness system used by the US Chess Federation. USCF adopted his method in 1960; FIDE followed in 1970. Every serious chess rating system today — FIDE, USCF, Lichess Glicko-2, Chess.com Glicko-1 — is either classical Elo or a Glicko refinement of it.
The mental model is simple: your rating minus your opponent's rating predicts the expected score of a match between you. A 200-point favourite is expected to score about 0.76 (roughly 3 wins to 1 loss over 4 games). After every rated game the system compares your ACTUAL score to the EXPECTED score, multiplies the difference by a constant called the K-factor, and adjusts your rating. Do better than expected → gain points. Worse than expected → lose points. The rating IS the strength estimate — there is no other number under the hood.
The Three Ingredients of an Elo Rating
Any working Elo system has exactly these three pieces — anything more is a refinement (Glicko, Glicko-2), anything less is not Elo.
1. Every player has a number
Your Elo rating is a single integer — usually somewhere between 100 and 2900 — that estimates your playing strength. Higher is better. New players start at a provisional rating (typically 1200 on FIDE, 1500 on Lichess, 800 on Chess.com's rapid pool) and the number moves after every rated game. The rating IS the strength estimate — there is no other number under the hood.
2. The number predicts the score, not the win
From the rating gap between two players, Elo's formula spits out an 'expected score' between 0 and 1 (0 = certain loss, 0.5 = draw, 1 = certain win). A 200-point favourite is expected to score about 0.76 over many games — not a guaranteed win in any single game. This is why upsets in a single game are common but almost never happen over a 10-game match.
3. Each game updates the number toward reality
After a game you compare your ACTUAL score (1, 0.5, or 0) to the EXPECTED score. Multiply the difference by a constant called the K-factor — usually 10, 20, or 40 depending on your title and the tournament — and add it to your rating. Overperform your expectation → gain points. Underperform → lose points. Draw with a stronger player → gain points. Draw with a weaker player → lose points.
The Elo Formula, In Two Lines
This is the entire mathematics of every modern chess rating system — one expected-score formula, one update rule.
Worked example: you are rated 1500, opponent 1700. E = 1 / (1 + 10^(200/400)) = 0.24. You draw (actual = 0.5). At K = 40, you gain 40 × (0.5 − 0.24) = 10.4 → new rating 1510. Your opponent loses the same 10.4 points → new rating 1690. Every rated game on every site in the world runs some version of this. Our free elo calculator does the arithmetic for any pair of ratings and K-factor.
FIDE vs USCF vs Lichess vs Chess.com — The Five Rating Pools
Different bodies run different rating pools. All are Elo-based but the calibration, K-factor, and floor differ — which is why the same player can be 1600 FIDE, 1650 USCF, 1900 Lichess Rapid, and 1750 Chess.com Rapid all at once.
FIDE (world governing body, over-the-board)
The international standard since 1970. Rating floor 1000. K-factor is 40 for players under 30 games (or under 2300 and under age 18), 20 for adults under 2400, and 10 for titled players 2400+. Ratings update monthly. FIDE ratings are the ones quoted in every media headline ('Magnus Carlsen's 2882 peak', 'Fischer's 2785 in 1972') and are the coin of the realm for real over-the-board tournaments.
USCF (United States Chess Federation)
American national body. USCF ratings run roughly 50–100 points higher than FIDE at the club level (a 1600 USCF is typically 1500–1550 FIDE) and closer at master level. Uses a modified Elo with an anti-inflation bonus formula and rating floors that ratchet up as you cross thresholds. USCF publishes a public online rating lookup at uschess.org.
Lichess (Glicko-2)
Lichess uses Glicko-2 — Mark Glickman's 2001 refinement of Elo that adds a 'rating deviation' (RD) to model how confident the system is in your number. New players start at 1500 with high RD, so ratings swing wildly at first and settle as RD shrinks. Lichess ratings run roughly 100–300 points higher than FIDE at intermediate level because the population skews younger and more online.
Chess.com (Glicko-1 variant)
Chess.com uses a Glicko-1 variant with separate pools for Bullet / Blitz / Rapid / Daily. Ratings typically run 100–200 points lower than Lichess in the same time control and roughly equal to or a bit below FIDE at rapid. Provisional ratings settle after about 20 games. Chess.com does not publish exact K-factors — the formula is proprietary but Elo-based.
chess.rodeo & other online pools
Independent online pools each maintain their own Glicko or Glicko-2 rating. Cross-pool comparisons are rough — the base 1500 is the same but the population, K-factor, and floor differ. Use our free rating converter for approximate cross-pool translation.
For a rough cross-pool translation, use our free rating converter. It handles FIDE ↔ USCF ↔ Lichess ↔ Chess.com with the standard offset tables.
Four Famous Elo Peaks
The all-time record book of FIDE classical Elo. Every serious chess fan should know these four numbers by heart:
Magnus Carlsen — 2882 (May 2014) — highest FIDE rating ever
Carlsen crossed 2800 at 19 and hit his peak of 2882 in May 2014 — the highest classical FIDE rating any human has ever held. He has stayed above 2800 for more than a decade, a level of consistency unmatched in rating history. His run demonstrates the Elo ceiling in practice: even a generational talent needs 200+ rated games at the very top to squeeze the last 30 points out of the system.
Garry Kasparov — 2851 (July 1999) — held the record for 15 years
Kasparov's 2851 stood as the all-time peak from 1990 to 2013. He held the world #1 spot for 255 out of 258 months between January 1984 and January 2006 — the longest #1 streak in FIDE history. Kasparov's rating was inflation-adjusted less than modern peaks because his generation had fewer engine-trained opponents to farm points from.
Bobby Fischer — 2785 (July 1972) — the pre-Elo era comparison
Fischer's 2785 was the peak FIDE rating in 1972 — the year he won the World Championship in Reykjavík. The rating gap to the field was massive (about 125 points above #2 Boris Spassky). By strength-adjusted metrics, some analysts rate Fischer's 1972 form as effectively 2820+ in modern terms, but the raw FIDE number is 2785. Fischer promptly stopped playing rated chess and his rating was frozen.
Judit Polgár — 2735 (July 2005) — highest ever for a female player
Polgár broke into the top 10 in 1996 and reached 2735 in 2005 — a rating that would still crack the world top 30 today. She refused the women-only title track (WGM/WIM) and competed exclusively in open events, making her Elo directly comparable to every male grandmaster of her era. Her rating was hit with the standard K-factor and no bonus system.
Elo vs Glicko vs Glicko-2 vs Bayeselo
Every modern rating system in chess is either classical Elo or a Glicko refinement. Here is how they differ:
| System | Adds | Used by |
|---|---|---|
| Classical Elo | Nothing — just rating + K-factor | FIDE, USCF |
| Glicko (1995) | Rating deviation (RD) — confidence in the rating | Chess.com |
| Glicko-2 (2001) | RD + volatility for inconsistent players | Lichess, chess.rodeo |
| Bayeselo | Bayesian posterior over an unknown true rating | Engine-vs-engine tournaments (CCRL, etc.) |
The refinements matter most for new / inactive players (where confidence is low) and for players with wildly inconsistent form. For a stable, active club player with 200+ rated games, all four systems produce ratings within 20–30 points of each other.
How to Gain Elo — Four Rules That Actually Work
Rating gain is a long game — the K-factor caps how fast it can move. But four rules produce the vast majority of measured climbs from beginner to expert:
1. Play players slightly above your rating
The formula rewards you asymmetrically: beating a 200-point-higher opponent nets +16 to +32 points depending on K-factor, while beating a 200-point-lower one nets only 4 to 8. Losses are the mirror image. If you play up 100–200 points, you gain more per win than you lose per loss over the long run — provided your true strength is genuinely near the higher pool. That is why 'always play up' is the single most common rating-climb advice.
2. Play more games during your rising phase
Elo is a moving average of your recent form. If you have improved but not played many rated games since, your rating LAGS your strength. Play 20–30 games during a rising phase and the rating catches up — often 100–200 points in a month. The classic pattern: study 3 months, play a weekend tournament, gain 80 points. Study 3 more months, gain 60 more. The rating floor rises with you.
3. Analyse every loss with an engine
Every loss contains the single move that flipped the eval from equal to lost. Finding it is the mechanism of improvement — engine analysis makes that move obvious. Free options with no signup: paste your PGN into our chess.rodeo Stockfish analysis or use our free PGN Viewer. The players who gain 200 rating in a year are almost always the ones analysing every loss with an engine within 24 hours of the game.
4. Pick a small opening repertoire and stick to it
Under 2000 Elo, most players lose more games to time-scrambled opening moves than to any tactical shot. Two openings as White (one against 1...e5, one against everything else), two as Black (one vs 1.e4, one vs 1.d4) will do until 1800+. Every extra opening dilutes preparation without adding wins. Use our free Opening Explorer to pick and study your four lines.
Common Rating Mistakes
The same five errors slow rating climbs at every level. Fix any of them and the rating catches up in 20–30 games:
Chasing a milestone rating in a single tournament
The K-factor caps how fast the rating can move: at K=40 you can gain at most 20 points per win against a peer. Even a perfect 9/9 gains only about 60–80 points in most systems. Trying to force a 200-point climb in one tournament by playing wild openings against strong opposition just costs points. Rating gain is a long game — plan across 30–60 rated games, not one weekend.
Refusing to play up because 'losses will tank my rating'
The math is against this fear: at 200 points below the opponent, you are EXPECTED to score only 0.24. Every draw you scrape is +8 to +16 rating points. Every win is +30 to +40. Losses are only −4 to −8. Playing down is where ratings actually stall: the expected score is 0.76 and you must WIN cleanly just to break even.
Comparing your Lichess rating to a FIDE rating one-to-one
The two pools are calibrated differently. A 1900 Lichess Blitz is roughly a 1700 FIDE classical player. A 2200 Chess.com Rapid is roughly a 2000 FIDE. Cross-pool conversion is fuzzy — the population, K-factor, and time control all matter. Use our free rating converter for approximate translations rather than assuming the numbers are the same coin.
Ignoring the K-factor when picking events
A K=40 tournament (open to sub-2300 adults) moves ratings twice as fast as a K=20 one. If you are on a rising streak, K=40 events help you climb faster; if you are on a losing streak, they hurt faster too. Titled players stuck at K=10 have the smallest per-game rating swing in chess — a well-known reason grandmaster ratings barely move even in bad tournaments.
Treating a provisional rating as final
New players start at a provisional rating (usually 1200 FIDE or 1500 Lichess). It takes 20–30 rated games for the number to stabilise — before that the rating deviation is huge and the system moves your rating aggressively toward your true strength. Don't screenshot your rating on game 5 and call it your 'real' Elo. Wait until you have at least 20 rated games before drawing any conclusion.
Model Your Rating for Free
Three free tools, no signup, no paywall — model the exact Elo swing of any game before you play it:
- Free Elo Calculator — enter your rating, opponent's rating, and the K-factor. See exactly how many points you gain per win or lose per loss. The fastest way to see why playing up 200 points is the mathematically correct strategy.
- Free Rating Converter — cross-pool translation between FIDE, USCF, Lichess, and Chess.com ratings. Use it before comparing yourself to a player from a different site.
- chess.rodeo — no account, no paywall — play unlimited free rated games with Stockfish post-game analysis. Track your rating trajectory alongside engine-flagged blunders so every rating swing comes with a ‘here’s the move that caused it’ explanation.
Frequently asked questions
Everything club players ask about Elo — the formula, the K-factor, FIDE vs USCF vs Lichess vs Chess.com pools, Glicko refinements, all-time peaks, title thresholds, and how to gain points.
What is Elo in chess?
Elo is a rating system invented by Hungarian-American physics professor Arpad Elo in the 1960s and adopted by FIDE in 1970. Each player has a single integer rating — higher is better — that updates after every rated game based on the result vs the expected result. The rating gap between two players predicts the expected score of a match: a 200-point favourite is expected to score about 0.76 over many games. All modern chess ratings (FIDE, USCF, Lichess Glicko-2, Chess.com Glicko-1) are Elo or Elo-derived.
Who invented the Elo rating system?
Arpad Elo (1903–1992), a Hungarian-American physics professor at Marquette University and an active tournament player himself. He designed the system in the 1950s to replace the older Harkness system used by the USCF; USCF adopted his method in 1960 and FIDE followed in 1970. The name is his surname — not an acronym. Common misspellings 'ELO' and 'E.L.O.' are wrong; the correct capitalisation is 'Elo'.
How is the Elo rating calculated?
The core formula has two steps. Step 1: compute the expected score E = 1 / (1 + 10^((opponent_rating − your_rating) / 400)). Step 2: compute the new rating R' = R + K × (actual_score − E), where actual score is 1 for a win, 0.5 for a draw, 0 for a loss, and K is the K-factor (usually 10, 20, or 40 depending on your title and tournament). Every rated game applies this formula. Our free elo calculator does the math for you.
What is the K-factor in chess?
The K-factor controls how much your rating moves per game. FIDE uses K=40 for new or under-18 players below 2300, K=20 for adults below 2400, and K=10 for titled players at 2400+. A higher K means bigger rating swings per game — good if you are improving fast, painful if you are on a losing streak. Titled players are pinned to K=10 to prevent short-term form fluctuations from moving elite ratings by 30-40 points per event.
What is the difference between Elo and Glicko?
Glicko (1995) and Glicko-2 (2001), both invented by Boston University statistician Mark Glickman, extend Elo by adding a 'rating deviation' (RD) — a confidence interval around your rating. Elo treats every rating as equally certain; Glicko says a rating from 200 rated games is more trustworthy than one from 5. New players have huge RD (ratings swing fast) and it shrinks as they play more. Glicko-2 further adds a 'rating volatility' term for players whose form fluctuates. Lichess uses Glicko-2, Chess.com uses Glicko-1, FIDE and USCF still use classical Elo.
Why is my Lichess rating higher than my FIDE rating?
Different pools calibrate differently. Lichess starts everyone at 1500 (vs FIDE's 1000 floor), the population skews younger and more online-native, and Blitz / Bullet ratings pull online games toward tactical strength over classical strength. Rule of thumb: subtract 200–300 from a Lichess Blitz rating to approximate FIDE classical. Our free rating converter has the fuller table. Note that above 2400 the gap narrows and elite Lichess ratings map closer to FIDE 1:1.
What is the highest Elo rating ever?
Magnus Carlsen — 2882 in May 2014, the all-time peak in classical FIDE ratings. Garry Kasparov held the previous record at 2851 from July 1999 until Carlsen broke it in 2013. Bobby Fischer's peak was 2785 in July 1972. On online pools, elite grandmasters routinely reach 3200+ Blitz on Lichess (Magnus Carlsen has topped 3400 Bullet), but online ratings are inflated relative to FIDE — the 2882 is the harder number to beat.
What Elo do you need to be a chess master?
FIDE title thresholds: Candidate Master (CM) 2200, FIDE Master (FM) 2300, International Master (IM) 2400, Grandmaster (GM) 2500. USCF titles: National Master 2200 lifetime peak, Senior Master 2400, plus 'Life Master' after 300 games above 2200. Reaching the CM 2200 milestone typically takes 5–15 years of serious training for an adult starting at 1200. GM (2500) takes an additional 3–7 years plus norm tournament results and is achieved by roughly 0.02 % of active tournament players.
What is a good chess Elo rating for a casual player?
Casual online player (a few games per week, no formal study): 800–1400. Serious hobbyist (regular study, club member): 1400–1800. Strong club player (weekly tournaments, opening prep, endgame study): 1800–2200. Expert / candidate master: 2200+. Percentiles on FIDE roughly: 1000 is the 15th percentile of rated players, 1500 the 40th, 1800 the 65th, 2000 the 80th, 2200 the 92nd. On Lichess Rapid the same percentiles are all about 300 points higher — a 1800 Lichess Rapid is roughly a 1500 FIDE.
How fast can you gain Elo?
The K-factor caps per-game gain. At K=40 (new / under-2300 / under-18), a maximum realistic climb is 100–200 rating points per year for a dedicated adult improver. Beginners racing from 800 to 1200 can move faster (+300 in six months is achievable). Above 2000, gains slow to 50–80 per year at best, and above 2400 the average player gains fewer than 20 per year. The single fastest way to gain: analyse every loss with an engine within 24 hours and drill the recurring mistakes. Our free PGN Viewer + chess.rodeo Stockfish analysis is a zero-cost route.
Does Elo work for chess variants like Chess960 or bughouse?
Yes — the Elo formula is variant-agnostic. Any two-player, zero-sum game with a clear win / draw / loss can use Elo. Lichess and Chess.com both maintain separate rating pools for Chess960, King of the Hill, Three-check, Antichess, Bughouse, and other variants. Ratings across variants are not directly comparable to standard chess ratings — the strengths correlate but a 2000 Antichess is not a 2000 standard chess player. FIDE maintains an official Chess960 rating pool as of 2019.
What is rating deflation vs inflation in chess?
Rating inflation is when the average rating of the top players drifts upward over decades without a corresponding jump in strength — often blamed on more juniors entering the pool at low ratings and losing points to higher-rated players. Rating deflation is the reverse: strong newcomers extract points from established players and the top ratings drift down. FIDE has fought both since the 1990s with anti-inflation bonuses and rating floors. Whether Carlsen's 2882 is 'really' equivalent to Fischer's 2785 is a long-running debate — engine-strength studies suggest Carlsen is stronger, but not by 100 rating points.
Where can I play rated games and gain Elo for free?
Every major online chess site offers free rated games with no signup paywall. Lichess and chess.rodeo both give you unlimited free rated play + Stockfish post-game analysis — no account, no paywall on either. Chess.com's free tier caps some rapid analysis features but rated games are free. For over-the-board FIDE-rated games, play in any FIDE-rated open tournament — entry fees are usually $30–$100 per weekend and the rating you earn is the official one that lets you chase FIDE titles. Track your rating trajectory at chess.rodeo and use our free elo calculator to model 'if I score X against a Y-rated field, what's my new rating?' before entering an event.
Model your next rating swing — free Elo calculator, no signup
Enter your rating, opponent, and K-factor. See exactly how many points you gain per win or lose per loss. Then play unlimited free rated games at chess.rodeo with Stockfish flagging every move that cost you Elo.