I am looking for a relatively large PGN-formatted database (with at least 500k+ games), that is downloadable, with possibly the following properties:

  • Simple formatted (no time indication after moves, etc), i.e. game info headers on top, followed by the moves, an example: (it doesn't have to contain all the headers of this example) enter image description here

  • Few such, but very small (2-3 thousand games) databases can be found here.

  • It would be preferential if the database was diverse, for example not being only composed of man vs. machine games or somesuch. No real constraint on the ratings, whether the levels (amateur to super grandmaster) are fairly balanced or whether it's a 2000+ rating database.

  • Although a single-file database would be easiest to work with, say for parsing purposes, one that is divided onto multiple PGN files would also be perfectly fine, as long as they were formatted similarly and did not contain too many duplicates.

In other posts, one mentioned database with similar properties to the above is the ICOFY database, but unfortunately it does not seem to be maintained any longer, nor can a trustworthy download link for it be found.

up vote 3 down vote accepted

Do not know if it really suits you but on http://www.theweekinchess.com/twic you can download a lot of PGN files. I once saw an explanation how somebody used this to create an opening book out of it. I could not find one big file with all games but there are quite a lot of files that date back to 25/06/2012. You can always make 1 big file out of it.

  • THIS! is the best database you can have. I have two game/opening databases: the million database one top-5000.nl/pgn.htm which itself is updated with games from TWIC up to january 2013, and another one I've build myself with TWIC games. The latter is the one I most use (as I have there the last contributions to opening theory), while the former is great to look for classic games. – sharcashmo Aug 25 '15 at 15:58
  • @sharcashmo the Million base 2.2 is indeed useful, but it is only composed of man vs. machine games, so lacks in diversity. As for TWIC itself, games have to be downloaded individually, doing that for building a 500k+ size database is definitely not a viable solution! I was more hoping to come across one similar to ICOFY. Thanks for the link by the way! – Phonon Aug 25 '15 at 16:17
  • The million base 2.2 is NOT composed of man vs machine games in any way. And TWIC allows you to download all the games of every week in a single file (one file per week). It took only a couple of hours to me to download all the old files, and once I had an up to date database I have only to download one file per week... not so hard :D – sharcashmo Aug 25 '15 at 16:21
  • @sharcashmo oh, I was reading the description of the column directly left to it :), anyway the millionbase is a windows EXE file, I'm working on a Linux machine. Is there a *.pgn format of it available? (or even a text file). – Phonon Aug 25 '15 at 16:38
  • @sharcashmo I think the weekly pgn's on TWIC, once put together will do the trick. I will do this then. Thanks for the suggestion! – Phonon Aug 25 '15 at 20:30

Here! Indexed by openings ECO code. http://www.logicalchess.com/learn/lessons/openings/pgndb/index.html.

  • You should click into the ECO section a-e and then click again for the download eg A00 here bookuppro.com/ecopgn/a.html. Just a heads up this database looks about 10 years old but fulfils all your other conditions :) – magd Aug 23 '15 at 21:22
  • Thanks for the clarification. Sadly most of the games in each file seem to be empty (no moves after the headers, only the result). – Phonon Aug 23 '15 at 21:27
  • 2
    +1 for the link. I may end up using it, but it is a very sparse database (so many files), with anomalies (admittedly rare for the remaining files) such as empty games. – Phonon Aug 24 '15 at 18:10

Your Answer

 

By clicking "Post Your Answer", you acknowledge that you have read our updated terms of service, privacy policy and cookie policy, and that your continued use of the website is subject to these policies.

Not the answer you're looking for? Browse other questions tagged or ask your own question.