Elo rating: chess term definition
Elo rating
Definition
The Elo rating is a numerical measure of a chess player’s strength based on game results against other rated opponents. Devised by Hungarian-American physicist Arpad Elo (1903-1992), the system predicts the probability of one player scoring against another and updates each participant’s rating after every rated game or event.
Historical background
Prior to 1970, FIDE relied on the less precise Harkness system. Recognizing its shortcomings, FIDE officially adopted Elo’s model at the Siegen Congress (1970) and published the first international rating list in 1971, topped by Bobby Fischer at 2760. Since then, national federations (e.g., USCF, German, Russian) and major online platforms have adapted variants of the Elo formula for different time controls and large player pools.
How it works
- Expected score: Each 400-point rating gap translates to an expected score of roughly 10:1 in favor of the higher-rated player; a 200-point gap gives a 75 % expected score.
- Rating update: NewRating = OldRating + K × (Actual Score − Expected Score). The constant K varies (e.g., 40 for new players, 10 for established elites).
- Distribution: The Elo model assumes performance follows a normal (or logistic) distribution around a player’s rating.
- Separate pools: FIDE now maintains distinct Elo lists for Classical, Rapid, and Blitz. Online servers add Bullet and Chess960 pools, often using Glicko or hybrid systems.
Strategic and practical significance
Elo ratings influence tournament invitations, title norms (FM, IM, GM), pairings, prize funds, and seeding in knockout events. For example, a player must reach a 2500 FIDE rating, in addition to achieving three GM norms, to earn the Grandmaster title.
Rating classes (over-the-board)
- 2600 +: Super-Grandmaster
- 2500-2599: Grandmaster (GM level)
- 2400-2499: International Master (IM level)
- 2200-2399: FIDE Master / National Master territory
- 2000-2199: Expert / Candidate Master
- Below 2000: Class A (1800-1999) down to Class E (<1200)
Milestones & examples
• Garry Kasparov became the first player to breach 2800 in 1990 (peak
2851 in 1999).
• Magnus Carlsen holds the all-time record of 2882 (2014).
• Judit Polgár’s 2735 (2005) remains the highest rating achieved by a
woman in classical chess.
• The largest recorded rating gain in a single FIDE list update was
achieved by GM Teimour Radjabov, who soared 146 points in January 2001
after an outstanding year as a junior.
Illustrative calculation
Suppose Player A (2600) defeats Player B (2400) in one classical game.
Expected score for A = 1 / (1 + 10((2400−2600)/400)) ≈ 0.76.
Actual score = 1. With K=10 (elite), A’s new rating = 2600 + 10 × (1−0.76)
≈ 2602.4, while B loses the same amount.
Interesting facts & anecdotes
- Bobby Fischer’s “Perfect 6-0” sweeps over Taimanov and Larsen in the 1971 Candidates propelled him from 2740 to an unprecedented 2785 on the next list.
- In Kasparov vs. Deep Blue, 1997, IBM tentatively assigned its computer an “unofficial” Elo of 2700+ based on training matches, highlighting how the system can extend beyond humans.
- Rating inflation is debated: average top-10 ratings have crept upward, yet statistical studies show overall pool growth is a major factor.
- Online rapid ratings can diverge wildly from OTB numbers—Hikaru Nakamura’s Chess.com blitz often exceeds 3200, illustrating the impact of shorter time controls and larger K-factors.
At-a-glance chart
[[Chart|Rating|Classical|2000-2023]]