I see the accuracy score in chess as about as valuable as a count of how many times your team's players kicked the ball in a soccer match, or the average speed of a player's shots during a tennis match. It can be interesting, and there's some correlation with the quality of the play, but don't take it too seriously.
@Toadofsky said in #6:
> ... because your moves were 62% accurate. 0% would represent "every move is a blunder" and you're nowhere near that.
You lost me. I am totally fine with this 62%. My question is why in two games with 42 moves each I have 89% accuracy for 5 blunders and 62% accuracy for 4 blunders?
I see that it more depends on the CPL and real change in position as the result of the move, but I am challenging this approach, since it hardly gives something valuable as the result. Regardless of the resulting changes, in the first game I made 5 blunders and this couldn't be 89% accuracy. This is my point.
> ... because your moves were 62% accurate. 0% would represent "every move is a blunder" and you're nowhere near that.
You lost me. I am totally fine with this 62%. My question is why in two games with 42 moves each I have 89% accuracy for 5 blunders and 62% accuracy for 4 blunders?
I see that it more depends on the CPL and real change in position as the result of the move, but I am challenging this approach, since it hardly gives something valuable as the result. Regardless of the resulting changes, in the first game I made 5 blunders and this couldn't be 89% accuracy. This is my point.
I use lucus chess gui to analysis some games. The analysis matrix is impressive and the elo graph is particular interesting to get an over all impression of the game. I wish we had that type of matrix and graph layout on lichess.
__________________________________________________________________________________________________
Quoted from www.perplexity.ai/search/value-of-accuracy-Ff84reYcSIWM82qfUN3euQ
"Based on the search results, the key points regarding the relationship between accuracy metrics and blunders in chess are:
1. Accuracy metrics alone do not provide a complete picture of the quality of a chess game. A high accuracy percentage does not necessarily mean the game was played well.[2]
2. Accuracy is calculated based on the difference between the player's move and the engine's recommended best move. However, this does not account for strategic errors or poor overall play.[2][3]
3. Blunders, which are egregious mistakes that significantly worsen the position, are more important than the raw accuracy percentage. A game with few blunders but many inaccuracies can be considered better played than one with a high accuracy but critical blunders.[1][4][5]
4. The definitions of blunders, mistakes, and inaccuracies are not standardized across chess platforms. They are based on the magnitude of the evaluation difference compared to the engine's recommendation, but the exact thresholds vary.[1][4][5]
5. The search results suggest that accuracy is not the sole determinant of the quality of a chess game. Factors like strategic thinking, positional understanding, and the ability to avoid critical blunders are more important.[2][3]
In summary, the search results indicate that accuracy metrics should not be viewed in isolation when evaluating chess performance. The presence and severity of blunders, as well as the overall strategic quality of play, are more important factors in assessing the true quality of a chess game.
Citations:
[1] www.reddit.com/r/chess/comments/ug8qt9/is_there_an_exact_inaccuracy_percentage_that/
[2] www.chess.com/forum/view/general/what-they-dont-tell-you-about-accuracy
[3] www.chess.com/forum/view/game-analysis/are-the-game-analysis-blunders-always-accurate
[4] lichess.org/forum/general-chess-discussion/difference-between-inaccuracies-mistakes-and-blunders
[5] www.chess.com/forum/view/game-analysis/what-is-an-inaccuracy
__________________________________________________________________________________________________
__________________________________________________________________________________________________
Quoted from www.perplexity.ai/search/value-of-accuracy-Ff84reYcSIWM82qfUN3euQ
"Based on the search results, the key points regarding the relationship between accuracy metrics and blunders in chess are:
1. Accuracy metrics alone do not provide a complete picture of the quality of a chess game. A high accuracy percentage does not necessarily mean the game was played well.[2]
2. Accuracy is calculated based on the difference between the player's move and the engine's recommended best move. However, this does not account for strategic errors or poor overall play.[2][3]
3. Blunders, which are egregious mistakes that significantly worsen the position, are more important than the raw accuracy percentage. A game with few blunders but many inaccuracies can be considered better played than one with a high accuracy but critical blunders.[1][4][5]
4. The definitions of blunders, mistakes, and inaccuracies are not standardized across chess platforms. They are based on the magnitude of the evaluation difference compared to the engine's recommendation, but the exact thresholds vary.[1][4][5]
5. The search results suggest that accuracy is not the sole determinant of the quality of a chess game. Factors like strategic thinking, positional understanding, and the ability to avoid critical blunders are more important.[2][3]
In summary, the search results indicate that accuracy metrics should not be viewed in isolation when evaluating chess performance. The presence and severity of blunders, as well as the overall strategic quality of play, are more important factors in assessing the true quality of a chess game.
Citations:
[1] www.reddit.com/r/chess/comments/ug8qt9/is_there_an_exact_inaccuracy_percentage_that/
[2] www.chess.com/forum/view/general/what-they-dont-tell-you-about-accuracy
[3] www.chess.com/forum/view/game-analysis/are-the-game-analysis-blunders-always-accurate
[4] lichess.org/forum/general-chess-discussion/difference-between-inaccuracies-mistakes-and-blunders
[5] www.chess.com/forum/view/game-analysis/what-is-an-inaccuracy
__________________________________________________________________________________________________
I wrote this blog lichess.org/@/JoaoTx/blog/exploring-the-python-chess-module/P0nb4FEs which explores Lichess' accuracy measure. I don't think will help you a lot but maybe a little.
@Toscani said in #13:
> 2. Accuracy is calculated based on the difference between the player's move and the engine's recommended best move.
This shows the biggest issue with accuracy: The engine's recommendation is not one fixed move in every position. It varies alot depending on your engine, your CPU/system, analysis time and depth and so on. Even today's top engines change their evaluation depending on how long they run. Openings analysed with Fritz 5.32 back in 1998 show terrible suggestions compared to today's top engines. But back then that was "the truth". And now a lot of people think that SF's recommendation after letting it run for 2 seconds on their mobile phone is "the truth". But there is none and there will probably never be one. So if you want to compare accuracy between players or games you should always use the same settings and once you get a better CPU or another SF version all results from older tests are obsolete.
> 2. Accuracy is calculated based on the difference between the player's move and the engine's recommended best move.
This shows the biggest issue with accuracy: The engine's recommendation is not one fixed move in every position. It varies alot depending on your engine, your CPU/system, analysis time and depth and so on. Even today's top engines change their evaluation depending on how long they run. Openings analysed with Fritz 5.32 back in 1998 show terrible suggestions compared to today's top engines. But back then that was "the truth". And now a lot of people think that SF's recommendation after letting it run for 2 seconds on their mobile phone is "the truth". But there is none and there will probably never be one. So if you want to compare accuracy between players or games you should always use the same settings and once you get a better CPU or another SF version all results from older tests are obsolete.
Accuracy is not a fixed metric, but rather a relative one that can vary depending on the game. So it's not an absolute percentage, but more of a relative percentage value. A player can make accurate choices without being efficient in their game. The popular move is not always the best engine move. The key to reach a goal is to find the right equilibrium between accuracy and efficiency.
The performance of a player is not their accuracy percentage either. Maybe the accuracy of the game needs to be compared to the opening database of popular moves, then the middle game by an engine and finally by an EGTB. The decision to make an accurate move needs some form of consistency. If it's played often and the WDL ratio is favorable, than it's an accurate move to play, until something else becomes more or less popular.
The performance of a player is not their accuracy percentage either. Maybe the accuracy of the game needs to be compared to the opening database of popular moves, then the middle game by an engine and finally by an EGTB. The decision to make an accurate move needs some form of consistency. If it's played often and the WDL ratio is favorable, than it's an accurate move to play, until something else becomes more or less popular.
Obviously it doesn't mean a great deal. It behooves one not to become an accuracy-monkey. ;)