Correlation of Accuracy to Rating • page 1/1 • General Chess Discussion • lichess.org

Not everyone will be as obsessed with accuracy as I am. That being said, I check and recheck my insights to see whether my average centipawn loss per game is falling or dropping, because now I've been playing more than ever and I like to think that I'm improving. Usually it stays around 64cp per game. This is odd because I've checked out the Insights of some other rapid players, some rated above 2000 ( I'm barely 1700 by the way ) and they have similar centipawn losses. Obviously it doesn't mean that I'd trounce them in a match, and at their level, they barely hang pieces. So, what does the whole thing mean? If you don't play so accurately, how come one can be so good?

ShiningDrongo

#2

you miscalculate a complicated tactic or you just hang a knight, it's all the same for an engine. a blunder is a blunder.

"So, what does the whole thing mean?"

It's as meaningful as your phone number divided by shoe size, unless you look in details: what openings give you the most trouble, time management and so on.

you miscalculate a complicated tactic or you just hang a knight, it's all the same for an engine. a blunder is a blunder. "So, what does the whole thing mean?" It's as meaningful as your phone number divided by shoe size, unless you look in details: what openings give you the most trouble, time management and so on.

kajalmaya

#3

To elaborate on what @ShiningDrongo said above, higher rated players may be making deeper mistakes like losing a pawn to a 5 move tactic or overlooking opponent's deep mistake, and the engine evaluates these mistakes in the same way - as a number equal to the drop in the evaluation of a position. But it is possible that if you play an opponent 400 points above you, they may play with lower average centipawn loss. For example, if you hang a pawn, they may not miss it, and very likely you will give them many more opportunities to make correct moves. The point is that in some positions correct moves are obvious and in some positions they are not. A strong opponent playing against a weak opponent will often create positions in which correct moves are also the obvious ones. But they may not be able to do that against another strong opponent.

One solution to this apparent inconsistency is to measure how shallow and narrow is a mistake, where shallow refers to the depth of the tree searched and narrow refers to the width of the tree searched.

Disclaimer: I don't know the details of the stockfish evaluation, so what I am saying is based only on some general considerations.

To elaborate on what @ShiningDrongo said above, higher rated players may be making deeper mistakes like losing a pawn to a 5 move tactic or overlooking opponent's deep mistake, and the engine evaluates these mistakes in the same way - as a number equal to the drop in the evaluation of a position. But it is possible that if you play an opponent 400 points above you, they may play with lower average centipawn loss. For example, if you hang a pawn, they may not miss it, and very likely you will give them many more opportunities to make correct moves. The point is that in some positions correct moves are obvious and in some positions they are not. A strong opponent playing against a weak opponent will often create positions in which correct moves are also the obvious ones. But they may not be able to do that against another strong opponent. One solution to this apparent inconsistency is to measure how shallow and narrow is a mistake, where shallow refers to the depth of the tree searched and narrow refers to the width of the tree searched. Disclaimer: I don't know the details of the stockfish evaluation, so what I am saying is based only on some general considerations.

Forrest01

#4

This is an interesting article.

https://kwojcicki.github.io/blog/CHESS-BLUNDERS#a1

This is an interesting article. https://kwojcicki.github.io/blog/CHESS-BLUNDERS#a1

Rookitiki

#5

One pretty important thing to keep in mind is that in the endgame you can make huge and crucial mistakes, but the ACPL will be diluted by the fact it took more moves to get there.
In other words: it matters when you make a mistake and ACPL therefore is not definitive (at least not across different players).

One pretty important thing to keep in mind is that in the endgame you can make huge and crucial mistakes, but the ACPL will be diluted by the fact it took more moves to get there. In other words: it matters when you make a mistake and ACPL therefore is not definitive (at least not across different players).

NM MrPushwood

#6

You shouldn't be so obsessed with "accuracy" on something that's not terribly accurate. ;)

Akbar2thegreat

#7

Acpl is relative figure that shows rough idea and it's something not to be thought of again and again.