---
title: "What Is a Good Chess Accuracy for Your Rating? · Chess Goat"
description: "Median and typical accuracy by rating for bullet, blitz and rapid, measured on 981,000 Lichess games with engine analysis. Plus why chess.com’s number differs."
canonical: "https://trychessgoat.com/improve/good-accuracy-by-rating/"
last_updated: 2026-11-18
---

# What is a good chess accuracy for your rating?

By Chess Goat · Updated November 15, 2026

At 1500 in Lichess blitz, the median game is played at 84.7% accuracy, and half of all games land between 80–89%. We measured 981,000 Lichess games that have Stockfish analysis, so you can see what is typical at your own rating and time control.

## What accuracy is typical at each rating?

A good accuracy for your rating is anything inside the typical range for your band, and above the median is better than most players at your level. In blitz, the median goes from 80.7% at 800–899 to 89.4% at 2400–2499. The typical range below is the middle half of games: a quarter of games score lower, a quarter higher. Ratings are Lichess ratings, and each game counts once for each player.

Bullet

Blitz

Rapid

Line: median. Shaded: the middle half of games (25th to 75th percentile). Lichess rating, 100-point bands.

Median accuracy and typical range (25th–75th percentile) by Lichess rating band and time control

Lichess rating

Bullet

Blitz

Rapid

400–499

—

—

**78.7%** (68–86%)

500–599

—

**78.6%** (70–85%)

**77.7%** (69–85%)

600–699

**80.3%** (72–87%)

**78.9%** (70–86%)

**78.4%** (69–86%)

700–799

**81.5%** (74–88%)

**79.8%** (72–86%)

**78.9%** (71–86%)

800–899

**82%** (75–88%)

**80.7%** (74–87%)

**80%** (73–87%)

900–999

**82.2%** (75–88%)

**81.5%** (75–87%)

**81.1%** (74–87%)

1000–1099

**82.2%** (76–88%)

**82.2%** (76–88%)

**81.8%** (75–88%)

1100–1199

**82.4%** (76–88%)

**82.7%** (77–88%)

**82.8%** (77–88%)

1200–1299

**82.7%** (76–88%)

**83.3%** (78–88%)

**83.3%** (77–88%)

1300–1399

**83%** (77–88%)

**83.6%** (78–88%)

**83.9%** (78–89%)

1400–1499

**83.1%** (77–88%)

**84.1%** (79–89%)

**84.4%** (79–89%)

1500–1599

**83.6%** (78–89%)

**84.7%** (80–89%)

**85.1%** (80–90%)

1600–1699

**83.9%** (78–89%)

**85%** (80–89%)

**85.4%** (81–90%)

1700–1799

**84%** (79–89%)

**85.4%** (81–90%)

**86%** (81–90%)

1800–1899

**84.2%** (79–89%)

**85.8%** (81–90%)

**86.5%** (82–91%)

1900–1999

**84.4%** (79–89%)

**86.4%** (82–90%)

**87.3%** (83–91%)

2000–2099

**84.7%** (79–89%)

**87%** (83–91%)

**88%** (84–92%)

2100–2199

**85%** (80–90%)

**87.6%** (84–91%)

**88.9%** (85–92%)

2200–2299

**85.5%** (81–90%)

**88.3%** (85–92%)

**89.7%** (86–93%)

2300–2399

**86%** (81–90%)

**88.9%** (85–92%)

**90.6%** (87–94%)

2400–2499

**86.5%** (82–91%)

**89.4%** (86–93%)

**91.4%** (88–95%)

2500–2599

**86.9%** (82–91%)

**89.9%** (87–93%)

**92.2%** (89–95%)

2600+

**87.9%** (84–92%)

**90.8%** (88–94%)

—

Median in bold, typical range (middle half of games) in brackets. A dash means fewer than 500 games in that band. Lichess time controls: bullet under 3 minutes, blitz 3 to 8 minutes, rapid 8 to 25 minutes (base time plus 40 moves of increment).

## What do the numbers show?

-   **Accuracy rises slowly with rating.** Across the whole blitz range, from 500–599 to 2600+, the median moves only from 78.6% to 90.8%.
-   **The ranges overlap a lot.** A typical 800–899 blitz game (74–87%) and a typical 2400–2499 game (86–93%) share several points. One game’s accuracy says little about a player’s rating.
-   **From about 1100 up, slower games score a little higher.** At 1500–1599: bullet 83.6%, blitz 84.7%, rapid 85.1% median. Lower down it flips: at 600–699, bullet 80.3% and rapid 78.4%.
-   **One blunder costs a few points, not the whole game.** Accuracy averages the winning chances lost per move: a 90% game of 40 moves drops to about 86% with one move that throws away 40 points of winning chances.

## Why is my chess.com accuracy different?

Your chess.com accuracy uses a different formula, and your chess.com rating is on a different scale, so compare with care. Chess.com’s accuracy is CAPS2, a proprietary score it has not published ([chess.com on CAPS](https://www.chess.com/article/view/better-than-ratings-chess-com-s-new-caps-system)). These numbers use the curves on [Lichess’s accuracy page](https://lichess.org/page/accuracy): each move’s loss in winning chances, averaged over the game, turned into a 0–100 score. Our own game review uses exactly this formula, so its accuracy is directly comparable with this table. Lichess’s analysis page now weights moves slightly differently, so its number can differ from ours by a few points.

Lichess ratings also run higher than chess.com ratings at the same playing strength, and by an amount that changes with level and time control. We do not convert between them. If you play on chess.com, read your own band with that in mind: the matching Lichess band is usually higher than your chess.com number.

## What can these numbers not tell you?

-   Only games someone asked Lichess to analyse are in the data (about 1 game in 11). Those are not a random sample: people analyse games they care about, wins and painful losses.
-   Short games are left out (under 20 moves in total): a 10-move resignation scores near 100% and says nothing.
-   Lichess’s server analysis searches deeper than a quick in-browser review, so the same game can score a few points differently.
-   One month of one site. Players on chess.com, in other months or in other pools may differ.

## How was accuracy measured?

The data is the [Lichess open database](https://database.lichess.org/) (CC0), month 2025-09, read from [its Parquet copy on Hugging Face](https://huggingface.co/datasets/Lichess/standard-chess-games): 8 of the month’s 62 files. We kept rated bullet, blitz and rapid games between human players (no bots) that ended normally or on time, had at least 20 moves, and carry a [Stockfish](https://stockfishchess.org/) evaluation after every move: 981,166 games, 1,961,186 player results. Each player’s accuracy uses the same code as our [game review](https://trychessgoat.com/improve/free-game-review/), checked against it in a test. Bands with fewer than 500 games are hidden. Data built 2026-11-15.

## What is your accuracy, game by game?

A free [game review](https://trychessgoat.com/improve/free-game-review/) scores any of your games with this same formula; for a Lichess game it shows the typical range for your rating next to it. To see which habits pull your accuracy down across many games, [analyze all your games](https://trychessgoat.com/improve/analyze-my-games/).

Free · 5 seconds · no email

### See what is costing you accuracy, free

Type your chess.com username. Your card shows the habits behind your mistakes in five seconds.

Want every mistake across up to 2,000 of your games, not one at a time? The [Goat Report is $5](https://trychessgoat.com/buy), paid once. See [a blitz sample](https://trychessgoat.com/r/sample-blitz/) and [a rapid sample](https://trychessgoat.com/r/sample-rapid/) first.

## Keep going

-   [Game review limitOne free review a day on chess.com, and how to review every game anyway.](https://trychessgoat.com/improve/free-game-review/)
-   [Analyze my gamesHabits across your games, one game with Stockfish, or all of them at once.](https://trychessgoat.com/improve/analyze-my-games/)
-   [Rating not going upIs it skill, noise or tilt? How to read a flat rating line.](https://trychessgoat.com/improve/rating-not-going-up/)
-   [Stop blunderingA 10-second blunder check: checks, captures, threats, and loose pieces.](https://trychessgoat.com/improve/stop-blundering/)

Chess Goat is not affiliated with or endorsed by any coach, author, channel or site named on this page. We link to them because their free material works. Quotes link to their sources.
