ChessOnyx

A system that is both FAST and STRONG. Become a different animal and the same beast. Dominate the game & reach success.Powered by Stockfish 18 | 3M+ Puzzles & free Analysis
Play Puzzles
New
New: Chrome Extension is Live!

Play a game on Chess.com or Lichess and get a direct link to analyze it here — instantly. Free to install.

ChromeAdd to Chrome — It's FreeLike it? A rating & review on the store means a lot to me.
ChessOnyx extension on Chess.com
ChessOnyx extension on Lichess

Improve at Chess

In-depth articles on chess improvement and understanding engine analysis, plus a variety of openings to explore.

Articles

View All Articles

Popular Openings

View All Openings

How it Works

This platform allows you to evaluate your chess games and positions directly in your browser using the Stockfish Web Computer. With integrated analysis features, you can adjust computer parameters to find the right balance between speed and depth for your needs.

Please note that evaluation results, accuracy, and performance metrics should be seen as directional trends rather than absolute facts. All evaluations can fluctuate based on your hardware performance and calculation depth.

Quick Guide

To help you get started, here is a simple walkthrough on how to run your first analysis:

  • Copy Game PGN: After finishing a match on major platforms like Chess.com or Lichess, you can usually export or share the game as a PGN string.
  • Import PGN: Click the "Import" button and simply paste your PGN (Ctrl+V). The board will immediately update.
  • Select a Computer: Choose your preferred Stockfish version. You can fine-tune parameters like threads or hash memory by clicking the gear icon in the computer panel.
  • Start Analysis: Once ready, hit the "Start Analysis" button to begin the move-by-move evaluation.

The interface is designed to be intuitive—I am sure you will figure it out in no time!

For Members

If you want to skip the manual copy-pasting, you can take a shortcut: Use the "Fetch Games" button to directly link your Lichess or Chess.com accounts. Free members can save up to 10 games, while Pro members enjoy unlimited storage and full history integration.

The Evaluation Philosophy

The system is intentionally strict. The moves are compared directly with the computer's top choice, often disregarding "human" logic or practical play. The system uses the Centipawn Loss (CPL) principle to calculate your accuracy.

Technical Overview: The Accuracy Model

Rather than relying on linear material shifts, this system measures move quality by how it alters the mathematical win probability. By stepping away from standard Centipawn Loss (CPL), the model mirrors the practical swings of a real chess game.

1. From Evaluation to Win Probability (WP)

Computer evaluations (measured in centipawns) are converted into a win probability (WP) using a sigmoid function. This approach recognizes the diminishing returns of a material advantage: gaining +1.00 in an equal game can be game-changing, whereas gaining +1.00 when already leading by +8.00 is negligible.

The relationship is expressed as:

WP=11+e−k⋅cpWP = \frac{1}{1 + e^{-k \cdot cp}}

Where cp represents the computer evaluation and k is a scaling constant governing the curve's sensitivity.

2. Contextual Severity: Opening vs. Endgame

The model factors in the game phase and the remaining complexity on the board. This explains why an identical material loss can yield different accuracy scores depending on when it occurs:

  • High Complexity (Opening): The "drawing margin" is broad in the opening. A small loss, such as a pawn, often results in a relatively minor drop in WP because the position remains sufficiently complex to allow for positional or tactical compensation.
  • Low Complexity (Middlegame/Endgame): As pieces are traded and the board simplifies, the game becomes more "solved." A material loss here causes a sharp drop in WP, because there are fewer variables to obscure the disadvantage, making it more straightforward for an opponent to convert a lead into a forced win.

3. Accuracy Score Calculation

The final accuracy is determined by the Win Probability Loss (L), which is the difference between the computer's best possible move and the move actually played:

L=WPbest−WPplayedL = WP_{best} - WP_{played}

To map this loss into a human-readable percentage, an exponential function is applied. This ensures the system remains sensitive to small inaccuracies while harshly penalizing blunders that significantly alter the outcome.

Accuracy=100⋅e−α⋅LAccuracy = 100 \cdot e^{-\alpha \cdot L}

Key System Benefits

  • Phase Awareness: The system distinguishes between theoretical inaccuracies in the opening and decisive blunders in the endgame.
  • Non-Linearity: It recognizes that maintaining a large advantage is fundamentally different from finding a singular winning resource.
  • Stability: The model provides a consistent metric across various computer depths and different playstyles.

Why did I build this?

I am a student with zero background in programming or web development. In my free time, I love playing meaningless chess. I always enjoyed tracking my progress through numbers and statistics, but I felt that high-quality analysis features shouldn't always come with a steep subscription price.

With absolutely no initial knowledge of how to build a website, I decided to create my own platform. I wanted a place that provides the features I love as a player—tools that offer motivation and help me see exactly where I can improve my game.

Transparency & Support

I know it might look like just another site trying to hide features behind a paywall. But I give my best to keep the heart of this project—the Analysis—free for everyone. I only charge for features that cost me real money to maintain, like cloud storage for your history or server maintenance.

This is a one-man operation. Maintaining and expanding this site as a hobby takes a lot of time and resources. It is more than enough for me if you simply try out the site. A "thumbs up" or, even better, constructive feedback helps me immensely.

If you do decide to support the project financially through a membership or donation, please know that I truly appreciate it. It is not something I take for granted. It helps me scale my computing power and servers, as this is just the beginning. I have many more features planned for the future.

Special Thanks

JPTheKuma

A huge shout-out to JPTheKuma, a Twitch streamer who uses ChessOnyx live on stream. Seeing the platform used in real-time on stream means everything to me. If you enjoy chess content, check out his stream — twitch.tv/jpthekuma

Lichess.org

An extra special shout-out to Lichess. Their API powers game fetching, their opening explorer drives the theory database, and their commitment to open-source software has saved me from many sleepless nights. The PGN viewer, the board logic, the opening data — Lichess gives it all away freely so others can build on it.

Chess.com

Thank you to Chess.com for their public API that allows ChessOnyx to fetch games and player data directly. Without access to this data, a huge part of what this platform does simply wouldn't exist.

Stockfish

The beating heart of this entire platform. Stockfish is the world's strongest open-source chess engine, and it runs entirely in your browser via WebAssembly. None of the analysis, accuracy scores, or engine lines would exist without this incredible piece of open-source software.

Chessground

Every chess board you see on this site is powered by Chessground, the open-source board UI library built by the Lichess team. It handles piece rendering, move animations, arrows, and highlights — all the visual magic that makes the board feel alive.

chess.js

All the chess logic under the hood — move validation, PGN parsing, FEN generation, legal move calculation — is handled by chess.js. A reliable and battle-tested open-source library that quietly powers everything behind the scenes.