Fun with CQL - Triple Pawn Battery!

Recently several of us at our local Seattle Chess Meetup were curious how often tripled pawns appear in high level games.

As it turns out, this is relatively easy to determine using Chess Query Language (CQL).  CQL is incredibly powerful for searching chess databases - much more so than a simple position search.  CQL may be installed as a stand alone program accessed through the command line.

http://www.arves.org/arves/index.php/en/endgamestudies/chess-query-language-5

Here is an example of CQL code to find all games in a database with tripled pawns for white:

cql(input om_top_otb.pgn)
shifthorizontal
countsquares 3 P[b2-7]

For black use a lower case p.  We could also include a flipcolor command and it would build a databases with results for BOTH black and white in one pgn file.  But I prefer keeping them separate for now.  The result of a CQL query is a new PGN containing the results of the query.

CQL found 1.762 games for white and 2,064 games for black with tripled pawns in games where both OTB players elo >2500. This is only about 1% of games played with both players > 2500.

White won ~ 37% of the games in which they had tripled pawns, lost ~ 23%, and drew ~ 40%.

Black won ~ 21% of the games in which they had tripled pawns, lost ~ 35%, and drew ~ 44%.

For the total population of games in the database sampled for players >2500 white won ~ 30%, lost ~ 19% and drew ~ 51% out of a total population of approximately 267,000 games.

(With no restriction on ELO, there are significantly fewer draws with white winning ~ 40%, black ~ 33% and ~ 27% draws out of over 8 million OTB games in the database but the interest here was top level games.)

I find this interesting.  Games with tripled pawns among top level players reduce the draw % and increase the probability of a win for both sides regardless of which side has the tripled pawns.  I suppose this makes sense.  At this level, players won't allow tripled pawns unless they think they are getting some compensation for it.

So George - I guess you may be on to something with your "Triple Pawn Battery"

Here is one - Magnus beats Anand in Leuven in 2016 with tripled pawns on the e-file starting at move 21...

Another - Magnus beats Karjakin with tripled pawns on the f-file starting at move 23...

And now for some examples with Black...

Aronian beats Caruana with tripled pawns on the c-file...

Kramnik beats So with tripled pawns at Leuven 2016 tripling on the c file at move 8...

Another random tip of the day.  Because pgn files are text files it is very easy to combine them using the command prompt in windows using something like:

copy file_1.pgn+file_2.pgn+file_3.pgn combined.pgn

I find this is MUCH faster than combining within a chess database program.