Resources

Lichess database exports (database.lichess.org)

Lichess is the leading free and open-source online chess platform and all of the rated games played on lichess (over 7 billion so far) are available for download under the very permissive Creative Commona CC0 license.

Files are broken out by month with most months containing 90-100 million games for standard chess. Lichess also supports several popular chess variants and the rated games for those variants are available for download as well. CQLi supports all of the variants available on lichess.org.

The Site tag included in each game contains a link to the game on lichess.org. ECO, Opening, and TimeControl tags are provided. Clock information (the amount of time remaining as each move was played) is provided as a comment after each move. These comments look like this:

    1. e4 { [%clk 0:01:00] } 1... c5 { [%clk 0:01:00] }
    2. Nf3 { [%clk 0:00:59] } 2... Nc6 { [%clk 0:01:00] }

Some games have engine evaluation comments as well which look like this:

    1. e4 { [%eval 0.15] [%clk 0:03:00] } 1... c5 { [%eval 0.25] [%clk 0:03:00] }
    2. Nf3 { [%eval 0.22] [%clk 0:03:00] } 2... d6 { [%eval 0.26] [%clk 0:03:01] }

These comments can sometimes be useful but if they are not needed they just take up a lot of space. Removing all of these comments reduces the size of these files by about 65% (the monthly standard rated files are usually over 200GB uncompressed so this adds up quickly). These comments can be removed using the CQLi command:

    cqli -i input.pgn -o output.pgn -silent -nosort -cql 'removecomment'

To remove only the clock comments, leaving the evaluation comments (which results in a reduction of about 60%) use:

    cqli -i input.pgn -o output.pgn -quiet -noheader -nosort -cql 'if hascomment {removecomment $newcomment = replace((originalcomment) "\[%clk .*?\]" "") if $newcomment != "" {comment $newcomment}}'

The Week in Chess (theweekinchess.com/twic)

The Week in Chess is a free online weekly chess magazine that includes a PGN file with several thousand high-quality games in each installment. This is a fantastic resource of mostly OTB games (internet games, such as chess.com Titled Tuesday tournaments, are differentiated with INT appearing in the Site field) that provides access to recent and high-profile tournaments.

HHdbVII - Endgame Study Database (hhdbvii.nl)

The Harold van der Heijden endgame study database (HHdbVII) is the most comprehensive collection of endgame studies available. The meticulously curated studies contain a wealth of information presented in a standard format including study composers, publication information, details of known cooks, corresponding composition tournament results, etc. The most recent version of the database (released in 2025, updated every 5 years) contains 103,157 studies consisting of 5,209,799 positions across all variations.