A correspondence chess book: IECG

[Mike S.]

The International Email Chess Group, IECG, has seized their activities per Dec. 31st, 2010. More than 166,000 correspondence chess games have been collected. The IECG provides them as a free download. The site is up still:

http://iecg.org/index.htm

I have made a reasonable selection of their games:

- both opponents have 2000+ IECG ratings
- no very short draws
- no long wins (have no relation anymore with the opening)

From the resulting 50.737 games, I have generated a CTG book. Of course, I provide it for free just as the IECG game database is provided. The download contains the CTG book for Fritz and the databases it is based upon, CBV and PGN:

http://remixshare.com/download/4p1dj
(very short waiting time)

Max. book depth is 30 moves. Fritz users can reduce this to smaller values anytime.

[Norm Pollock]

Good find Mike.

I made an edited collection of the games among the top players. There are 40,223 (of the original 166,659) games in the collection.

Among the editing I did was:
removed double games
minimum elos of 2000
minimum plies of 31
removed any comments
new ECO codes from SCID
sorted by ECO codes
removed unnecessary disambiguities
removed games with irregularities from PGN Standards
removed excess tags

the link for the zipped download:

https://dl.dropboxusercontent.com/u/662 ... -edited.7z

-Norm

[Albert Silver]

The International Email Chess Group, IECG, has seized their activities per Dec. 31st, 2010. More than 166,000 correspondence chess games have been collected. The IECG provides them as a free download. The site is up still:

http://iecg.org/index.htm

I have made a reasonable selection of their games:

- both opponents have 2000+ IECG ratings
- no very short draws
- no long wins (have no relation anymore with the opening)

From the resulting 50.737 games, I have generated a CTG book. Of course, I provide it for free just as the IECG game database is provided. The download contains the CTG book for Fritz and the databases it is based upon, CBV and PGN:

http://remixshare.com/download/4p1dj
(very short waiting time)

Max. book depth is 30 moves. Fritz users can reduce this to smaller values anytime.

Many thanks for the resource!

[jdart]

This is a good starting point, but you have to be careful about the game results. Many games, especially in correspondence, are lost by time forfeit, so the game result is not a reliable indication of how good the opening is.

Here is just one random example:

[pgn]
[Event "CP.2004.P.00032"]
[Site "IECG"]
[Date "2004.10.01"]
[Round "?"]
[White "Hansen, John"]
[Black "Speisser, Patrick"]
[Result "1-0"]
[ECO "A07"]
[WhiteElo "2243"]
[BlackElo "2094"]
[PlyCount "39"]
[EventDate "2004.??.??"]

1. Nf3 Nc6 2. g3 e5 3. d3 d5 4. Bg2 Nf6 5. O-O Be7 6. Nbd2 O-O 7. c4 d4 8. a3
a5 9. Rb1 a4 10. Qc2 h6 11. h3 Re8 12. Kh2 Bd7 13. Ne1 Bc5 14. Ne4 Nxe4 15.
Bxe4 Na5 16. Bg2 Ra6 17. Nf3 Nb3 18. Nd2 Rb6 19. Nxb3 Rxb3 20. Bd2 1-0

[/pgn]

Black is fine, in fact better, in the ending position, but lost, probably by forfeit.

--Jon

[Mike S.]

Many games, especially in correspondence, are lost by time forfeit, so the game result is not a reliable indication of how good the opening is.

That is a problem, and for now I can just hope that the big number of normally finished games "evens out" a potential distortion of the result percentages at least in the most common variations.

As a "hotfix", I recommend to set the min. games requirement to a reasonable big value in the [F4] book options, e.g. 20, which reduces the probability of distortion by forfeits. But of course, that will reduce the length of book variations, too.

Related to that, meanwhile I noticed that the number of games the tree itself gives, is smaller than the number of games in the database (48830 / 50737). Many games had either all moves "missing" (somewhat unclear) and/or were forfeited at start because the first move wasn't done in time, obviously.

I am not familiar with IECG and with corr. in general, that is why I was not aware of that problem. Otherwise I could have easily removed these "no moves games" (I only removed very short draws but not very short wins.)

I am considering a 2nd, greatly improved edition.

[jdart]

I have made an even more filtered set:

Both players >=2200, average rating >= 2250, filtered to remove games whose end position differs greatly in eval from what the result suggests:

http://www.arasanchess.org/iecg-jdart.zip.

(somewhere between 11,000 and 12,000 games in this set).

--Jon

[Graham Banks]

I have made an even more filtered set:

Both players >=2200, average rating >= 2250, filtered to remove games whose end position differs greatly in eval from what the result suggests:

http://www.arasanchess.org/iecg-jdart.zip.

(somewhere between 11,000 and 12,000 games in this set).

--Jon

Hi Jon,

any chance you could do a zip file of games between opponents 2400+ (if there are any)?
I realise that this would be quite a tiny file, but I'd still be keen on having it if it was available.

Graham.

[Mike S.]

Hi Graham, just a few mouse clicks:

http://remixshare.com/download/y0od6

1010 games 2400+ (I have excluded games with less than 10 moves.)

[Mike S.]

Since this thread may be bumping up (sorry), I just want to mention that there is a greatly improved 2nd edition of the CTG book:

http://talkchess.com/forum/viewtopic.php?t=48093

Please use that one, not the initial one.

[Graham Banks]

Hi Graham, just a few mouse clicks:

http://remixshare.com/download/y0od6

1010 games 2400+ (I have excluded games with less than 10 moves.)

Thanks. Much appreciated.