Average Centipawn Loss – An Introduction and Explanation of the Concept

Just when you think you are getting to grips with the game of chess, improving your skills, and seeing your ELO rating, someone comes along and mentions Average Centipawn Loss and you have no idea what it is. This article will explain what centipawn loss is as a unit of measure and how it is used in chess.

A centipawn is 100/1th of a single pawn and centipawn loss is a calculation and numerical score given by a chess engine to the difference between the move you actually play against the strongest move available at that time. A GM may score Average CPL of under 20, a new chess player, 150!

Using the average centipawn loss figure across an entire game provides an idea of the overall skill of a player, however – Read on to find out more about Centipawn loss, how it works, and how it could be used.

Chess Piece Values in Centipawn Standard Valuation

PieceStandard ValueCentipawns
Queen9900
Rook5500
Knight3300
Bishop3300
Pawn1100

The use of a scoring system in chess is very much frowned upon by many in the game. Attributing a value or points to individual pieces arguably has no bearing on the direction a game may take or the skill of the players involved.

Although even a master will be making calculations around exchange and sacrificial situations as to whether there is a material advantage on the other side.

When players make calculations around exchanges, it is a simple process of considering the worthiness of what pieces may be given up in exchange for a more advantageous and attacking position on the board; it does not use Centipawn scoring directly and no human could assess the centipawn loss potential of individual moves within the game anyway.

So back on the evaluation, the above table of centipawn value uses the standard accepted points value of each chess piece and multiplies it by 100, therefore arriving at one centipawn being one 100th of a pawn.

Whilst these standard values are widely known, some pre-eminent players had a different value for chess pieces, which could also adjust during the game.

For instance, one theory of the value of a Pawn is that is it worth 2pts at the start of the game, but towards the end-game, the value increases to 3.75. Few would argue that a pawn can be a very valuable asset in the end game and this is an interesting take. In some cases the King is given a value when used in attack.

Emanuel Lasker suggested that starting positions of pieces would also affect their value, with pawns closer to the center being worth more than those on the flanks and even the two bishops having different values based on being either kingside or queenside.

There are a host of other chess piece valuations which, if you are interested, are covered in a piece on Fandom.

How is Centipawn Loss Calculated

Centipawn loss is calculated by first measuring the values of both players’ positions, using chess engines.

Then each player moves and a new position is computed with any loss in centipawn signifying the accuracy and worth of the move. An example would be a move that loses 0.25 pawns worth of advantage, and your centipawn loss would be calculated as a centipawn loss of 25.

The score has no formal bearing on the game nor the position, although over the course of the entire game the average centipawn loss of each move, will give some indication to your skill level according to the chess engine analysis of each move and the game as a whole.

What is Average Centipawn Loss

Average Centipawn loss (ACPL) can be as low as 25, signifying close to perfection across a given game, or it can be high as you could imagine.

To keep things in scale, an average centipawn loss over 300 would suggest that you have either never played the game before or you are playing suicide chess.

The lower the average centipawn loss a player scores in a game, reflects on the accuracy of the moves made and the advantage maintained in the position of all pieces according to chess programs.

I played a perfect game but still had Centipawn loss!

It is virtually impossible to have zero average centipawn loss in a game. Indeed, an individual move may not be a mistake nor blunder, but still score centipawn loss according to the engine. You can even have a centipawn loss move that are not inaccurate that would still score 40.

It is dependant on many different factors and the position of every chess piece on the board.

Even a Brilliant move could incur a small loss because the engine reassesses the board with a potentially deeper evaluation than was available prior to the move.

I am Rated 1500 Elo – What Centipawn loss Should I expect?

Just because you have a certain rating does not dictate what centipawn loss you will incur in your next game, but this table may provide a guide alongside player ratings. You would expect Stronger players to incur smaller losses, weaker players a higher figure, and for the majority of players, take a look at the average player rating in the table.

LevelELOExpected ACPL
Grandmaster2500+10-20
Master2200-249910-30
Expert Player1800+15-50
Average Player1200-180050-100
Beginner<120080-200
Monkey300

The mathematics starts getting crazy when trying to plot the ELO rating tangibly to centipawn loss. It is interesting no doubt but begins to deflect from the real aim of just trying to play your best game and best possible move each time.

Magnus Carlsen Centipawn Loss

Magnus needs no introduction and one would expect that whilst a grandmaster can no longer beat the best chess engines,  the world champion can score a very accurate centipawn loss score.

Taking the 2018 championship into account Carlsen played 12 games and had an average centipawn loss over under 10. In one game, [Drawn – all the classical games were] Carlsen scored 5 and Caruana only 4! This is incredibly accurate.

Centipawn loss figures have got lower and lower over the years.

With improved chess engines and the use of them as part of training and game analysis, average centipawn scores have been coming down, and it could be argued that the increased accuracy of player ability these days is what has led to more and more games ending in draws, the 2018 championships a prime example of this.

The Centipawn System is not Perfect

Consider one player with a huge advantage on the board playing a near perfect positional game. There is room to not select the best move and incur centipawn loss without a major shift in the advantage or chance to win the game. Theoretically, the engine will calculate a shift, but in reality, little will have changed such is the superiority of one side.

What effect does missing a mating opportunity have on Centipawn loss?

OK, we are getting a bit too deep into this now, to be able to answer, and given that one engine may have a slightly different assumption than another, things start to get hugely muddy.

I had to look this up myself, and found this complicated set of answers that delve into mathematics which draw you in so far as to start moving me away from studying my chess and playing which has to be inherently worse for my overall game than actually playing and practicing!

Mathematics in chess is great, and I don’t dispute that it is important to know but just not for me to this depth. If we concentrate on looking for the best move, every move, then any centipawn loss value will look after itself, and should not be used as a consideration or target.

Just concentrate on trying to find the best move on the board and not consider how that will be scored.

Happy Chessing!

Similar Posts