Elo Calculator

How to use the Elo rating calculator

What this calculator does

This calculator estimates the expected score and Elo rating change for a player after one game or a batch of up to ten games. It applies the standard base-10 Elo expectation curve and the update rule rating change = K × (actual score – expected score). The result is a reproducible mathematical estimate, not an official federation rating. Governing bodies may use rating floors, score-probability tables, special provisional rules, rounding conventions, or event-level adjustments that are outside this compact model.

When to use it

Use it to preview how a chess tournament could affect an established rating, compare the rating impact of a win versus a draw, audit a small league's hand calculation, or test how a different K-factor changes rating volatility. The underlying approach also works for many head-to-head games and sports that use an Elo-style rating scale.

How to calculate

1. The calculator opens with a complete demonstration: a 1,750-rated player faces a 1,810-rated opponent twice, earning one win and one draw. The results and an example Excel workbook are available immediately.
2. Replace Player's rating with the rating before the game or rating period. Choose Single game or Tournament / multiple games under Game type.
3. For each row, enter the Opponent's rating and choose a Score: win equals 1 point, draw equals 0.5, and loss equals 0.
4. Leave Adjust K-factor manually off to use the calculator's established-player convention, or enable it and enter a Manual K-factor that matches your league's rules.
5. Read New Elo rating, Rating change, Expected points, Actual points, and Performance vs expected. The table shows the contribution from each game.
6. Select Download Excel to export the current canonical inputs and outputs as a validated .xlsx workbook. Reset clears the demonstration and all result data; export remains disabled until a complete valid state is entered again.

Input guide

Player's rating is required and accepts a plain number from 0 to 5,000, such as 1750 or 1750.5. Commas are accepted only as thousands separators, so 1,750 is valid but 1,5 is rejected as ambiguous. A higher player rating raises the expected score against the same opponent, which reduces the reward for an ordinary win and increases the penalty for a loss. Game type is required: Single game keeps one row, while Tournament / multiple games lets you add up to ten rows. Switching to single-game mode removes extra game rows.

Opponent's rating is required for every game and uses the same 0-to-5,000 format. A stronger opponent lowers your expected score, so beating that opponent usually produces a larger gain. Score is required and must be Win, Draw, or Loss from the player's perspective; do not enter a match score such as 3 – 1. Adjust K-factor manually is optional. When off, this calculator uses K = 20 below 2,400 and K = 10 at or above 2,400 as a transparent established-player convention. Manual K-factor becomes required when the option is enabled and accepts 1 to 100; for example, 40 makes every change twice as large as K = 20. Official chess rules distinguish provisional, youth, and elite cases, so confirm the applicable coefficient in the FIDE Rating Regulations.

Output guide and worked example

The header pills labeled Current rating, Games, Expected points, and K-factor repeat the live state in a compact form. New Elo rating is the starting rating plus the total calculated change. Rating change is positive after outperforming expectation, negative after underperforming, and near zero when actual points closely match expected points. Expected points is the sum of the model's expected score for each game; it is an estimate, not a guaranteed win probability. Actual points is the exact score total from the selected results. Performance vs expected equals actual minus expected points and has the same direction as the rating change. In the table, Expected score and Actual score are per-game values, while Running rating shows the cumulative effect for readability.

For the startup example, the rating difference is 1810 – 1750 = 60. The expected score for each game is 1 ÷ (1 + 10^60/400) ≈ 0.4145. Across two games, expected points are about 0.8290, while the win and draw produce 1.5 actual points. With K = 20, the total change is 20 × (1.5 – 0.8290) ≈ +13.4, so the first-open result is 1,763.4. Small display differences can occur because the calculator keeps full precision internally and rounds only for presentation.