Ktulu's endgame play

[Graham Banks]

The way that Ktulu plays rook and pawn endgames in particular seems quite impressive.
Just an observation.

[Frank Quisinsky]

Hi Gabor,

I made some endgame statistics with the TOP-21, 1 Core, 32-Bit. Ktulu seems to be the number 3 in endgame playing strength. Made the most points if the position in move 80 is = and win up to move 200.

Stockfish is a little bit stronger, but Ktulu seems to have clearly more knowledge for a lot of endgames. Same if I compare with Rybka.

Problem is ... Ktulu lost the most games up to move 55 from the TOP-21. Also each other TOP-21 engine won more games up to move 55 as Ktulu. Late middlegame = around place 12-14 in the TOP-21.

With other words.
Ktulu hope to get the endgame ... and the monster is born

The reason why this engine is very important. Its a pity that Ktulu have different problems with stability.

Nice to have such fine chess programs. More interesting as clones, here all is the same.

Last week I made 31 statistics (middlegame, endgames, different with remis games and so on). 12 of the TOP-21 engines plays completly different chess (completly other statistics). Will not say more, or we have a new clone discuss. So I hold in the information privat.

But Ktulu is on place 1, most differences to all other TOP-21 engines. On two is Hiarcs, on three is Spark. Also a great statistics in late middlegames with rocks and pawns. Seems that the rock is the favorite figure from Rahman Ktulu give often knight / bishop for pawn and rock in middlegame and lost very fast

Best
Frank

[hgm]

The way that Ktulu plays rook and pawn endgames in particular seems quite impressive.

On the other hand, the way it plays KQBK is the pits.

[Frank Quisinsky]

Hi,

interesting is also:
Ktulu don't use endgame databases.

I have 2 or 3 games in the SWCR-32 bit database (after more as 1.400 games) Ktulu lost with KN vs. KR.

Nothing is perfect

Best
Frank