Tord Romstad wrote:Don't forget Sjeng, who played a very strong tournament, and Buzz, the only engine apart from Rybka and Sjeng which didn't lose a single game.Spock wrote:Yes - and congratulations to Tord and the Stockfish team for achieving third place !!
I had a brief look at Stockfish's games tonight, and it is clear that its third place was entirely undeserved. It got a free point in the first round (it seems that the opponent didn't show up), and somehow managed to get 1.5/2 from technically lost positions out of the book against Pandix and Spark. Sjeng, Hiarcs, Thinker, Crafty and Diep didn't play anywhere close to their best against Stockfish.
Stockfish was also helped by a massive hardware advantage compared to most of the participants, which surely explains some of its luck.
I would like to take a minute and thank Tord for Stockfish and of course for his permission to allow me to operate Stockfish/Glaurung the past 3 tournaments so far! Also special thanks to Joona Kiiski, Marco Costalba for helping out with Stockfish.
I think I need to clarify few things about Stockfish games in ACCA. Round 1 Stockfish did play TornadoX The engine logged in a bit late and the game went all the way to the endgame Stockfish Had the better position and up a pawn, worst case scenario the game would end with a draw, but it is true Torando logged off and forfeit win awarded to Stock.
Now regarding the opening book that Stockfish used. It is actually an opening book I created back for CCT (back in CCT10). It is only selective of openings and mainlines/sidelines that I picked to be played all the time in the tournament! It wasn't meant to play different openings or with different variations. The trick was to allow stockfish to play unpredictable opening/rare once with the idea that some book authors may not be prepared for because they mainly give their effort toward main popular openings. Therefore I decided to let Stockfish play 1... g5 against 1.c4 for one it is an actual opening. Yes, it could weaken the king but it is tricky because it arises unfamiliar positions out of the book for engine. Reaching Dark square pawn structure, g5, h6, c5, d6 gives a good control of the dark squares! However I haven't created enough book moves toward 1.c4 g5 And indeed Pandix took Stockfish out of book quite early, but the 3 or 4th time g5 was played, I got positive record whether a win or draw! In CCT11 and ACCA but more prepartion is indeed for sure! Trompsky and Ruy lopez I think I was prepared, Hiarcs only had beaten Stockfish in that opening if I remember, but this time stockfish made some interesting saves against hiarcs, it may have looked loss but that d5 passer and the way it played the game awarded the draw! Diep and IkarusX games is a bit of proof that Trompsky is not too easy to predict and stockfish had the good positions right after the opening book! Diep game is really showing it. The Ruy lopez, I actually managed to take out Rybka out of book in CCT11 but Rybka managed to win the game, but I felt it may have been a book draw. Sjeng game proved that stockfish/glaurung play Ruy very well! Lastly, I really think stockfish played badly against rybka in a very equal position in my view, but much better handling against Crafty and it won!
P.S. Spark game I agree, it was book error from my part!