How many moves are there in chess! • page 4/4 • General Chess Discussion • lichess.org

Here is a quick and dirty estimate (lower bound) - a little more than 6300.

You can do the following.

Start with a not too small database. I used pgnmentor, which has about 3.56M games.
Remove the metadata (event, site, players, ...) of each game, and write each move without move number in enhanced long algebraic notation (e.g., f5e6ep is the en passant capture with pawn going from f5 to e6, and f5e6ep# is en passant capture giving checkmate, etc.). I did this with pgn-extract as follows.

pgn-extract --nomovenumbers --notags -w1000 -Welalg --quiet input.pgn -o output.pgn

Count the number of distinct words in output.pgn as follows.

grep -wo '^\S*' output.pgn | sort | uniq -c > move-freq.txt

move-freq.out has the unique words and their frequencies. If you want, you can sort by frequency. The number of lines minus 3 is the estimate (lower bound). I had to ignore the words 0-1, 1-0, 1/2-1/2.

6300 is only a lower bound, but it may not be off by much. I am sure some moves may have never occurred (for example, weird moves like b5c6 en passant capture giving checkmate). I would be interested in comparing the estimate with an accurate calculation.

Here is a quick and dirty estimate (lower bound) - a little more than 6300. You can do the following. 1. Start with a not too small database. I used pgnmentor, which has about 3.56M games. 2. Remove the metadata (event, site, players, ...) of each game, and write each move without move number in enhanced long algebraic notation (e.g., f5e6ep is the en passant capture with pawn going from f5 to e6, and f5e6ep# is en passant capture giving checkmate, etc.). I did this with pgn-extract as follows. pgn-extract --nomovenumbers --notags -w1000 -Welalg --quiet input.pgn -o output.pgn 3. Count the number of distinct words in output.pgn as follows. grep -wo '^\S*' output.pgn | sort | uniq -c > move-freq.txt move-freq.out has the unique words and their frequencies. If you want, you can sort by frequency. The number of lines minus 3 is the estimate (lower bound). I had to ignore the words 0-1, 1-0, 1/2-1/2. 6300 is only a lower bound, but it may not be off by much. I am sure some moves may have never occurred (for example, weird moves like b5c6 en passant capture giving checkmate). I would be interested in comparing the estimate with an accurate calculation.

#32

DuMussDieUhrDruecken

#33

"accurate calculation" ... music in my ears

DetectiveDumbo

#34

between 10 to the power of 111 to 10 to the power of 123 according to some research.(Of course the number includes illegal moves.) If you rule the illegal positions out, it will probably be 10 to the power of 40.

DuMussDieUhrDruecken

#35

@DetectiveDumbo said in #34:

wot u torkin bout

@DetectiveDumbo said in #34: > wot u torkin bout

There’s this YouTube video about what the rarest move is