How Chess Ratings Work: Elo, K-Factor & More Explained
iCan is reader-supported: if you buy through links on this page, we may earn a commission at no extra cost to you. How we review.
Whether you’re new to competitive chess or looking to understand the number next to your name, the chess rating system can seem like a mystery. That number isn’t just a scoreboard; it’s a dynamic prediction of your playing strength. At its heart, a chess rating has one primary purpose: to predict the outcome of a game. Developed by Hungarian-American physics professor Arpad Elo, this system, and its variations, is used by organizations worldwide, including FIDE (the World Chess Federation) and the US Chess Federation (USCF). It’s not a simple tally of wins and losses but a sophisticated calculation that weighs the significance of each result. Beat a much stronger player, and your rating will jump. Lose to a much weaker one, and it will drop significantly. Understanding how this works is key to tracking your progress and setting realistic goals.
The Core Principle: Expected Score vs. Actual Score
The entire rating system revolves around a simple comparison: how you were expected to perform versus how you actually performed. Your rating and your opponent’s rating are used to calculate an expected score, which is a statistical prediction of your result.
The greater the rating difference, the more lopsided the prediction. For example, according to the standard calculations:
- If two players have the same rating, each has an expected score of 0.5 (a 50% chance).
- If Player A is 100 points higher rated than Player B, Player A’s expected score is about 64%.
- A 200-point gap leads to an expected score of about 76% for the stronger player.
- A 400-point gap means the higher-rated player is expected to score about 91%.
After the game, you have your actual score: a win (1 point), a loss (0 points), or a draw (0.5 points). The rating change is determined by the gap between this actual score and the expected score.
The Math Behind the Rating Change
The formula for updating a rating is straightforward. Your new rating is your old rating plus a change, which is calculated as follows:
New Rating = Old Rating + K × (Actual Score – Expected Score)
This simple line of arithmetic is where the magic happens. Let’s break down the components, especially the crucial K-factor.
Understanding the K-Factor
The K-factor acts as a “speed limit” for how much your rating can change from a single game or event. It represents how responsive the system is to new results.
- High K-factor: For new players or those with fewer games, the K-factor is higher (e.g., 40 in the FIDE system). This allows their rating to adjust quickly to find their true skill level.
- Low K-factor: For established players, the K-factor is lower (e.g., 20 for most FIDE-rated adults). This stabilizes their rating, as a single unexpected result is less likely to dramatically change a rating built over many games.
- Very Low K-factor: For elite players (e.g., those rated over 2400 by FIDE), the K-factor can be as low as 10, making their ratings very stable.
A Practical Example
Imagine a player rated 1500 (with a K-factor of 20) plays against a player rated 1700. The 1500-rated player’s expected score is only about 0.24. Now, imagine they win!
- Actual Score = 1
- Expected Score = 0.24
- Difference (Actual - Expected) = 1 - 0.24 = 0.76
- Rating Change = K × Difference = 20 × 0.76 = 15.2 points
The 1500-rated player would gain about 15 points for this upset victory. Conversely, the 1700-rated player, who was expected to score 0.76, would lose a similar amount for the surprising loss.
Different Rating Pools: FIDE, Online, and US Chess
It’s important to know that your rating is only meaningful within its specific “pool” of players. Your rating on Chess.com is separate from your rating on Lichess, which is separate from your official FIDE or US Chess rating.
| Rating System | Key Characteristics |
|---|---|
| FIDE / US Chess | Uses a K-factor system based on Arpad Elo’s original design. Ratings are “provisional” until a player has played enough games (e.g., over 25 for US Chess), after which they become established. |
| Chess.com | Uses a proprietary variant of the Glicko system, which is similar to Elo but incorporates a “rating deviation” to represent uncertainty. |
| Lichess | Uses the Glicko-2 system, which also tracks a rating deviation and volatility. A “?” next to a rating indicates the system is still uncertain about the player’s true strength. |
These pools are not calibrated against each other. A blitz rating of 1800 on Chess.com, a rapid rating of 1800 on Lichess, and a standard FIDE rating of 1800 represent three different skill levels relative to their respective player pools. You cannot directly compare the numbers. This is also why your first official tournament rating might be significantly different from your online rating—it’s a completely different set of opponents and a different calculation system.
Provisional vs. Established Ratings
When you first start playing in a rated system, your rating is labeled provisional. This is the system’s initial guess at your strength, and it is designed to change rapidly as you play more games. Don’t be alarmed by large swings early on; this is the system efficiently homing in on your true level.
Once you have a sufficient number of games under your belt (e.g., more than 25 rated games in US Chess), your rating becomes established. Changes will typically be smaller and more gradual because the system has more data and is more confident in its assessment.
Frequently Asked Questions
Why did I lose points for winning a game?
You cannot lose points for a single win. However, in a tournament, your final rating change is based on your total performance across all games. If you win a game against a lower-rated opponent but lose several others to much lower-rated players, your overall tournament performance may be below the system’s expectation, resulting in a net rating loss. The win itself would have contributed a positive change, but it may have been outweighed by negative changes from the losses.
Is a 1200 rating good?
“Good” is relative. A 1200 rating on a major site like Chess.com places you above many beginners, indicating a grasp of basic tactics and strategy. In the context of US Chess tournaments, a 1200 rating would typically make you eligible for “Under-” sections designed for developing players. It’s a solid starting point for serious improvement.
Can I improve my rating by only playing weaker players?
No, this is a common misconception. If you only play players rated far below you, your expected score will be very high (e.g., 0.91 against a player 400 points lower). Winning these games will meet expectations, resulting in very small rating gains. A single loss, however, would cause a significant drop. Sustainable rating growth comes from consistently scoring above expectations against a variety of opponents, including those rated similarly or higher than you.
Why does my online rating differ from my official tournament rating?
As mentioned, they are different pools with different calculations. The player strength and competitive environment on an online platform are not the same as in an over-the-board tournament. Furthermore, online systems like Glicko-2 factor in the speed of play (bullet, blitz, rapid) separately. Your official FIDE or US Chess rating is specific to classical time controls in a regulated tournament setting.