Table of Contents
What Is the Elo Rating System?
The Elo rating system is a method for calculating the relative skill levels of players in zero-sum games such as chess, Go, and competitive video games. Originally devised by physicist Arpad Elo in 1960 for the United States Chess Federation, the system has since been adopted by FIDE (the international chess federation) and countless other competitive organizations worldwide.
The fundamental principle behind Elo is that a player's rating should reflect their probability of winning against another rated player. After each game, the winner takes rating points from the loser. The number of points transferred depends on the difference in ratings between the two players. If a higher-rated player beats a lower-rated opponent, only a few points change hands because the result was expected. If the lower-rated player wins (an upset), a larger number of points are transferred.
This self-correcting mechanism ensures that ratings converge toward an accurate representation of true skill over time. New players may experience volatile ratings initially, but as they play more games, their ratings stabilize and become reliable indicators of performance.
Elo Rating Formula
Where RA and RB are the current ratings, EA is the expected score (win probability) for player A, SA is the actual score (1 for win, 0.5 for draw, 0 for loss), and K is the K-factor that controls the maximum possible rating change per game.
Understanding the K-Factor
| K-Factor | Use Case | Max Change per Game |
|---|---|---|
| 10 | Top-level players (FIDE 2400+) | ±10 points |
| 20 | Standard rated players | ±20 points |
| 32 | USCF standard / newer players | ±32 points |
| 40 | Provisional / rapid calibration | ±40 points |
A higher K-factor makes the system more responsive to recent results but also more volatile. FIDE uses K=40 for a player's first 30 games, K=20 for players below 2400, and K=10 for players who have ever reached 2400. Many online platforms use K=32 as a balanced default.
Elo Rating Ranges
| Rating Range | Chess Level | Description |
|---|---|---|
| 2700+ | Super Grandmaster | World elite, top 30 players globally |
| 2500-2700 | Grandmaster | Professional-level competitors |
| 2200-2500 | Master | Expert-level, tournament winners |
| 1800-2200 | Class A-B | Strong club players |
| 1400-1800 | Class C-D | Intermediate players |
| 1000-1400 | Beginner | Casual or novice players |
Frequently Asked Questions
What happens when two equally rated players play?
When two players have the same rating, each has an expected score of 0.5 (50% win probability). The winner gains K/2 points and the loser drops by K/2 points. With K=20, the winner gains 10 points and the loser drops 10 points.
Can my Elo rating go below zero?
Theoretically yes, but in practice most systems set a floor (commonly 100). In FIDE chess, the minimum published rating is 1000. The formula itself does not impose a lower bound, but organizations typically enforce one.
Is Elo used outside of chess?
Absolutely. Elo or Elo-derived systems are used in online gaming (League of Legends, Dota 2), table tennis, football (FIFA rankings use a modified Elo), Scrabble, Go, and many other competitive domains. The core math is the same; only the K-factor and implementation details differ.
How many games until my rating stabilizes?
Most systems consider a rating provisional until 20-30 games have been played. After roughly 50 games against opponents of varying strength, the rating typically converges to a stable value that accurately reflects skill level.