How The Elo Chess Rating System Works

The Elo chess rating system allows chess players to determine how strong they are compared to other players. Chess ratings are straightforward once you see them in action. However, it can be confusing to understand the theory behind the chess rating system.

In general, when chess players play rated games, they gain or lose points on their USCF chess rating based on their performance. The stronger their opponent, the more they can raise their rating.

The Elo chess rating system is also a factor used in awarding chess master titles. So if you want to become a chess grandmaster one day, you should get familiar with the system.

Here's how the Elo chess rating system works.

White chess piece knocks over black king on chess board.
Learn how the chess rating system works. Credit: Pixabay/Pexels

The Birth of the Elo Chess Rating System

The Elo rating system was created by Arpad Elo, a physics professor in the United States. Elo was also a chess master and wanted to improve how the US Chess Federation evaluated players' skill levels. 

The general idea behind the rating system is that one player gives some of their points to another player based on the outcome of a game.

Before the Elo system, USCF used something called the Harkness system from 1950 to 1960. Kenneth Harkness, a tournament organizer, developed the system and wrote about his invention in Chess Review. In 1960, the Elo rating system became the standard behind the US chess rating system. Then, in 1970, FIDE adopted the rating system.

Today, the Elo rating system is not used only to measure strength in chess but in other sports and games like Go, Scrabble, football, and tennis.

Example of Go board.
Go is an example of a game that uses the Elo rating system. Credit: Elena Popova/ Unsplash

The Elo chess ratings algorithm is pretty sophisticated, and FIDE and USCF use slightly different variations. Plus, players may perform differently in FIDE tournaments than in tournaments by their national chess federation. For these reasons, players often have slightly different USCF and FIDE ratings.

But the Elo rating system isn't the only system chess players use today. The Glicko rating system, for example, is a different version of the Elo system. It uses more variables to determine a player's rating to improve ratings' reliability. Chess.com uses the Glicko rating system. Lichess and the Internet Chess Club uses the Glicko-2 rating system. 

In 1995, Mark Glickman created this system. While it was created predominantly for chess, Online-go.com and other online game servers use the chess rating system.

How the Elo Rating System Works in General

In general, the Elo system calculates the likely outcome of a person's games against other players.

A principle of the Elo system is that the higher-rated player is more likely to beat the lower-rated player. 

For instance, a player rated 100 points higher than their opponent will likely win 64% of their games or five in eight. Someone rated 200 points higher than their opponent will win 75% of their games. 

Based on these odds, the Elo system decides how many points two players gain and lose with every game.

Players' ratings will change based on the outcome of the game and the difference between their ratings.

Hand moves a white pawn on a tournament-style black and white chess board.
Playing rated chess games will adjust a chess player's USCF chess rating, which is based on the Elo rating system. Credit: Anastasia Shuraeva/ Unsplash

For instance, when a lower-rated player beats or draws a higher-rated player, they gain rating points. The bigger the difference in ratings, the greater the achievement for the lower-rated player. Therefore, they will see a greater increase in rating points.

On the other hand, the higher-rated player will lose rating points from this scenario. Even if they draw, the higher-rated player will lose rating points. This is because the higher-rated player should be able to beat the lower-rated player. 

When a higher-rated player beats a lower-rated player, they gain fewer points than the other way around. They were expected to win against the lower-rated player, and the result doesn't demonstrate that the higher-rated player dramatically improved their performance.

When two players with the exact same Elo draw, they gain no rating points.

Limitations of the Elo Chess Rating System

For a number of reasons, players can be severely underrated or overrated based on who they have faced in the past. So it's important to remember the Elo system is an estimate, not an absolute measure, of whether a player is improving.

For instance, ratings change based on wins and losses. If an opponent makes the absolute best moves throughout most of a game but blunders forced mate towards the end, they lose that game. They also lose at least some rating points, even if they played like a chess master right until the end.

There are also a few other things to consider that affect ratings.

Consider the Ratings Floor and the K Factor

Rating floors also change how many points a chess player can lose. This means that no matter how poorly a player does, their rating cannot decrease below a certain level.

Marble black and white checkered floor.
The ratings floor prevents players from losing Elo after they achieve a certain chess rating in their career. Credit: Pete Alexopoulos/ Unsplash

One reason rating floors exist is sandbagging. Sandbagging is when competitive players keep their scores lower to qualify for lower-rated sections. This allows them to have an easier time competing and winning prize money.

In general, a player's rating floor is 200 points lower than their highest rating ever achieved.

Another factor that impacts rating is the K factor. This uses the number of prior games and the prior rating to determine whether a player's rating can change more rapidly. The algorithm can also factor a player's age, giving them greater rating boosts for their wins if they are younger because they might be in a period of rapid growth.

The higher the K factor, or the younger the player, the more rating points they will earn from positive outcomes.

You can use this estimator from US Chess to determine how a USCF chess rating will be changed by different scenarios.

Examples of Rating Adjustments by Performance

Let's create a scenario and see how ratings can change. A 1000-rated player goes to a 5-round tournament. We are not factoring in age and we are assuming they have played 50 games prior. Using the online US Chess estimator, here is what would happen to this player's score based on these different scenarios:

Earns 2.5 points after playing five 1000-rated players- new rating: 1000

This player drew all their games against players that are considered equally strong.

Earns 0.0 points in 5 round tournament against all 1800-rated players- new rating: 997

This player lost all their games. But they faced tough competition and statistically had very little chance of winning. They lose only 3 rating points.

Earns 0.5 points in 5 round tournament against all 1800-rated players- new rating: 1022

As you can see, this player experienced four losses and a draw. Without taking ratings into consideration, this sounds like a terrible performance. Yet they managed to avoid a loss against an 1800! This is an achievement for the 1000-rated player. They are rewarded for their perserverance in the face of an asymetrical battle with 22 rating points.

The Elo ratings of Chess Masters

The chess grandmaster title is the premier title in chess. To become a titled player under FIDE, you need to reach the following Elo chess ratings:

  • Chess Grandmaster: 2500 or higher
  • International Master: between 2400- 2500
  • FIDE Master: between 2300-2400

You'll also need to earn norms in tournaments.

Women's FIDE chess titles have slightly lower rating requirements:

  • Woman Grandmaster: 2200 or higher
  • Woman International Master: between 2100- 2200
  • Woman FIDE Master: between 2000-2100

There are also a few other titles that don't require norms.

For instance, to become a Candidate Master under FIDE, you need a 2200 FIDE rating. National Master titles vary by each country's federation, but as a player in the United States, you need to achieve a USCF chess rating of 2200. You are considered an Expert chess player once you reach a rating of 2000.

One informal title players can reach is Super Grandmaster, which describes when a chess grandmaster surpasses a rating of 2700. For instance, Magnus Carlsen, the greatest player alive today and the highest-rated player in FIDE, had a peak Elo of 2882 in 2014.

Magnus Carlsen in a white button-up shirt in front of a chess board at a tournament with a water bottle next to him.
Magnus Carlsen has reached the highest chess Elo of all time. Credit: Lennart Ootes/Wikimedia Commons

Chess Ratings for the Rest of Us

So even if you haven't cracked a 2000 rating, your chess rating still shows your skill level. 

In general:

The bottom line is that the chess rating system is ultimately a useful tool to help you determine your strength and measure improvements. So use it to your advantage, improve your skills with practice, and enjoy the game!

FAQ: How The ELO Chess Rating System Works

What is a good chess rating score?

A rating between 1600-2000 is considered strong, while above a 2000 is expert to master level. You can learn more about what a good chess rating is here.

Is 1200 a high chess rating?

A 1200-rated player is an intermediate player. It's not a bad rating, but there is room for improvement.

Who has the highest Elo rating ever?

Magnus Carlsen, the greatest chess player in history, holds the record for the highest Elo rating ever achieved by a human player. He reached 2882 in 2014. Right now, he is at 2837 and remains the highest rated player today.

Is Elo rating 3000 possible?

A FIDE or USCF chess rating of 3000 is theoretically possible. However, in practice, it is likely not possible, at least not with the current rating system design. Magnus Carlsen himself said he likely won't reach a rating of 2900.

Rating inflation is hard to explain, but over decades, the FIDE ratings of top players have slowly climbed up. Elo wasn't designed to compare players across different eras but to compare players who face each other. This all means that over future decades, the chance of a player reaching 3000 will become more possible, even if it doesn't happen in our lifetimes.