ELO Calculator

How the Elo Rating System Works

The Elo rating system measures relative skill between two competitors in zero-sum games. Arpad Elo, a Hungarian-American physics professor and chess master, designed the method in the 1960s to replace older, less reliable ranking formulas.

Unlike points in a league table, Elo ratings do not simply accumulate with every match you play. Instead, your rating goes up when you win and goes down when you lose, with the exact point shift depending on who you faced.

The core idea is simple: beating an opponent rated 400 points above you is a major upset, so you gain a large number of points. Beating a player rated 400 points below you is expected, so your rating barely moves. If you lose to that lower-rated player, your rating drops significantly because the system expects you to win that matchup almost every time.

Today, Elo and its variants run behind the scenes in traditional board games, video game matchmaking, and sports analytics. Chess organizations like FIDE and USCF rely on it for official rankings. Many competitive titles, including tactical shooters, battle arenas, and digital card games, use modified versions to build fair multiplayer lobbies.

How to Calculate Elo Rating Changes

Calculating an Elo rating update requires two distinct mathematical steps. First, you calculate the expected score for each player. Second, you adjust each player's rating based on the difference between their actual score and their expected score.

Step 1: Calculate the Expected Score

The expected score represents the probability that a player will win, plus half their probability of drawing. The system calculates this value using a logistic curve based on the rating difference between the two competitors:

In this formula:

• $E_A$ is Player A's expected score, expressed as a number between 0 and 1.
• $R_A$ is Player A's current rating.
• $R_B$ is Player B's current rating.
• 400 is the standard scale factor used in chess and gaming.

A rating advantage of 400 points gives the stronger player an expected score of roughly 0.91 (a 91% expected score). If both players share the exact same rating ($R_A = R_B$), the exponent becomes zero, $10^0 = 1$, and both players have an expected score of 0.50 (50%).

Player B's expected score is simply:

Step 2: Calculate the New Rating

Once the match concludes, compare the actual match result to the expected score. The rating update formula is:

In this formula:

• $R'_A$ is the updated rating for Player A.
• $R_A$ is the starting rating.
• $K$ is the development coefficient, known as the K-factor.
• $S_A$ is the actual match score (1 for a win, 0.5 for a draw, 0 for a loss).
• $E_A$ is the expected score calculated in Step 1.

The difference $(S_A - E_A)$ determines the direction and scale of the shift. If you outperform expectations ($S_A > E_A$), your rating rises. If you perform worse than expected ($S_A < E_A$), your rating falls. Because the system is zero-sum between the two players, whatever points Player A gains, Player B loses:

Understanding the K-Factor

The K-factor determines how volatile rating changes are after a single contest. A high K-factor makes ratings swing rapidly after every victory or defeat. A low K-factor keeps ratings steady, requiring a sustained track record of results to move up or down.

Different organizations assign different K-factors based on age, tournament experience, and rating brackets:

FIDE (World Chess Federation)

• $K = 40$ for players new to the rating list until they have completed at least 30 games, as well as for all players under age 18 whose rating remains under 2300.
• $K = 20$ for most established players rated under 2400.
• $K = 10$ for top-tier master players who have reached a published rating of at least 2400.

USCF (United States Chess Federation)

Uses variable K-factors that scale with a player's current rating and game count, generally ranging from 32 for developing players down to 10 for senior masters.

Competitive Video Games and Casual Clubs

Video game placement matches frequently use effective K-factors of 32 to 50 to help new accounts settle quickly into their skill rank. Once rank stabilizes, the effective K-factor drops to 16 or 20 to prevent wild swings.

Selecting the right K-factor in our calculator allows you to replicate your exact league rules.

Step-by-Step Worked Examples

These examples walk through realistic match scenarios to show how ratings shift in practice.

Example 1: An Even Matchup

Two players with identical ratings of 1500 play a tournament game with a K-factor of 20. Player A wins.

Calculate the rating difference:

Calculate Player A's expected score:

Determine the actual score: Because Player A won, $S_A = 1.0$.

Calculate the rating change:

Update both ratings: Player A new rating: 1500 + 10 = 1510. Player B new rating: 1500 - 10 = 1490.

When opponents have matching skills, a win always yields exactly half the total K-factor in rating points.

Example 2: The Underdog Scores an Upset

Player A (rated 1350) faces Player B (rated 1650) with a K-factor of 32. Player A pulls off an unexpected victory.

Calculate the rating difference:

Calculate Player A's expected score:

Player A had only a 15.1% expected chance of winning.

Determine the actual score: Player A won, so $S_A = 1.0$.

Calculate the rating change:

Update both ratings (rounded to nearest integer): Player A new rating: 1350 + 27 = 1377. Player B new rating: 1650 - 27 = 1623.

Because the upset was so unlikely, Player A collects almost the maximum possible points available under a K-factor of 32.

Example 3: A Favorite Draws Against an Underdog

Player A is rated 1800, and Player B is rated 1600. They play with $K = 20$, and the game ends in a draw.

Calculate the rating difference:

Calculate Player A's expected score:

Player A was expected to score 76% of the points (win or draw heavily favored).

Determine the actual score: A draw awards half a point: $S_A = 0.5$.

Calculate the rating change:

Update both ratings: Player A new rating: 1800 - 5.2 = 1794.8 (rounds to 1795). Player B new rating: 1600 + 5.2 = 1605.2 (rounds to 1605).

A draw against a significantly lower-rated opponent causes the higher-rated player to lose points. For the underdog, drawing against a stronger opponent acts as an overperformance and earns rating points. You can check expected point distributions across longer series using our winning percentage calculator.

Expected Score and Win Probability Reference Table

The table below illustrates how rating gaps translate into expected scores and win chances under the standard 400-point scale:

All point-shift columns assume K = 20.

When the gap between two competitors reaches 400 points, the stronger player is expected to win roughly nine out of ten games. When the gap reaches 735 points, the expected score for the weaker player drops below 1%.

How Elo Differs Across Different Sports and Games

While Arpad Elo originally built this framework for board chess, modern competitive ecosystems adapt the formula to match their mechanics:

• Chess (FIDE and USCF): Classical chess relies on pure Elo. Official updates apply at the end of each monthly rating period or tournament, rather than game-by-game, treating the entire event as a collective pool of matches.
• Team Video Games (League of Legends, Valorant, Overwatch): These titles use matchmaking rating (MMR) systems inspired by Elo. Because individual players queue in five-player squads, the game calculates an average team Elo before the match. Individual performance factors or hidden confidence values often modify final gains.
• Fighting Games (Brawlhalla, Street Fighter): 1v1 fighting games apply Elo rules directly. Players climb visible ranked tiers where point gains mirror standard K-factor shifts.
• Association Football and International Rankings: FIFA uses a modified Elo formula for world rankings. Their equation incorporates match importance weights (friendly games carry lower K-factors than World Cup knockout matches) and adjusts for goal differentials.

Common Mistakes When Using an Elo Calculator

• Using the wrong K-factor: Entering an arbitrary number like 32 when your federation uses 10 or 20 distorts your projection. Check your organization's handbook before calculating.
• Treating score as goals or rounds: In the Elo formula, score ($S_A$) is always binary or fractional for the entire match: 1 for a win, 0.5 for a draw, and 0 for a loss. Entering the number of goals or chess pieces captured produces completely invalid results.
• Applying game-by-game updates inside a single tournament: FIDE calculates tournament rating reports by comparing your total tournament score against your total expected score across all opponents, using your pre-tournament rating for every calculation. Calculating each game sequentially with an updated rating will produce a small discrepancy compared to official FIDE lists.
• Assuming 0 rating points means absolute zero skill: Elo is an ordinal scale, not a ratio scale with an absolute zero. A player with 1200 Elo is not "twice as skilled" as a player with 600 Elo. The difference between two ratings is the only number that holds statistical meaning.

Limitations of the Classic Elo System

While Elo is dependable and mathematically elegant, it operates under specific assumptions:

• No measure of rating uncertainty: Elo treats a player who has played 1,000 matches with the exact same certainty as an unranked player who just completed their third match, provided they have the same K-factor. Modern algorithms like Glicko address this by tracking a Rating Deviation (RD) number.
• Rating deflation and inflation: Over decades, active player pools can experience inflation (average ratings drifting upward) or deflation (new players leaving the pool before distributing their points). Federations occasionally adjust minimum rating floors to rebalance the pool.
• Draw frequency bias: Elo assumes a logistic distribution of performance. In high-level chess, master players draw a large percentage of games, which can slightly compress rating progression compared to decisive games.

Frequently Asked Questions

Can an Elo rating ever be negative?

Mathematically, the formula allows negative numbers if a player loses enough games against weak opposition. In practice, most organizations enforce a rating floor (such as 1000 or 1400 in FIDE) so ratings never drop below a set positive threshold.

What is the 400 rule in FIDE chess?

For rating calculations, FIDE caps the maximum counted rating difference between two players at 400 points. If an unrated or lower-rated player faces a Grandmaster rated 800 points higher, the game is processed as if the gap were exactly 400 points, protecting both players from extreme mathematical distortions.

How does a draw affect my Elo rating?

A draw awards a score of 0.5. If your opponent had a higher rating than you, your expected score was below 0.5, meaning your rating will go up. If your opponent had a lower rating, your expected score was above 0.5, meaning your rating will go down.

What does an expected score of 0.75 mean?

An expected score of 0.75 means you are projected to win 75% of the total points available over a series of matches against that opponent. In a four-game match, you would be expected to score 3 points (for example, three wins and one loss, or two wins and two draws).

What is a good Elo rating in chess?

A rating of 1000 to 1200 represents an intermediate casual club player. Ratings between 1600 and 1800 reflect strong tournament competitors. Players above 2000 are candidate masters, while international Grandmasters typically hold ratings of 2500 and above.

Why do some video games hide my exact Elo number?

Video game studios often separate public visual ranks (such as Gold, Platinum, or Diamond) from behind-the-scenes matchmaking rating (MMR). Hidden MMR prevents players from gaming the system, while visual ranks protect players from the emotional fatigue of watching points fluctuate after every single round.

Is Elo used for team sports?

Yes. Analysts apply Elo to football, basketball, and baseball by treating the entire roster as a single combined rating. Ratings update after every game based on home-field advantage and final win-loss outcomes.

Related Calculators

• Probability Calculator
• True Shooting Percentage Calculator
• Magic Number Calculator