Not logged inRybka Chess Community Forum
Up Topic The Rybka Lounge / Test Positions / TCEC S9 Stage 3, games 26/54 and 72/100 (IDeA tree analysis)
- By MaLaoshi (*) [tw] Date 2016-10-07 18:13
[White "Stockfish 160716"][Black "Fire 5"]
[FEN "r1bq1rk1/1ppnbppp/p1np4/4p3/B3P3/2PP1N2/PP3PPP/RNBQR1K1 w - - 5 9"]
9. Be3 (9. Bc2 Nb6) b5 10. Bc2 Bb7 (10. ... Nb6) 11. Nbd2 Re8 12. a4 Nc5 13. axb5 axb5 14. Rxa8 Qxa8
15. b4 Ne6 16. Nf1 Bc8 17. h4 (17. h3) Nf8 18. Bd2 Bf6 19. g3 Be6 20. Ng5 Bxg5 21. hxg5 d5 22. f4 dxe4 23. dxe4

In the above position, arising out of the Ruy Lopez Averbakh variation in games 26 and 54 of TCEC Season 9, Stage 3, SF's play in game 54 was criticized because the trades along the a-file led to an early draw. I have tried to analyze this opening (and its sister opening with h3 not d3 below) by setting up an IDeA tree for it, with > 10000 "tasks", each consisting of > 2 min engine time with ASMfish.

This leads me to suspect that the early book-exit moves in fact were OK. Rather, it was 17. h4 and 20. Ng5 which led in a second round of exchanges, after which the position became quite drawish (eval +0.20). Had SF continued more quietly with say 17. h3, the white advantage would have lasted longer (eval +0.41). Also the book-exit move 9. Be3 (eval +0.37) looks solid, using that d3 has been played already. In contrast, Fire's book exit in game 26, 9. Bc2, looks premature, and can be answered advantageously with 9. ... Nb6 (eval +0.27). Granted, this tree analysis checks the moves of TCEC SF with another SF, albeit to a greater depth. Also, white's moves in the above line are not "only moves", and say 12. a3, 12. h3, 12. Qe2, 14. b4, 14. Qe2 etc could all be tried. Still, Stockfish got a good position out of the opening against fairly accurate defense, without missing alternatives that are obviously better--about as much as can be expected of it.

It is interesting to compare the above with the "h3" variation:

[White "Stockfish 160716"][Black "Andscacs 0.872b"]
[FEN "r1bq1rk1/1ppnbppp/p1np4/4p3/B3P3/2P2N1P/PP1P1PP1/RNBQR1K1 w - - 5 9"]
9. Bc2 (9. d4 b5 10. Bc2 Nb6 11. Nbd2 (11. a3) Re8 (11. ... exd4) 12. b3 b4 13. d5 bxc3 14. dxc6
cxd2 15. Bxd2) 9. .. b5 10. a4 Bb7 11. d3 Bf6 (11. ... b4 12. a5) 12. Be3 (12. Nbd2) Nb6 13. axb5 axb5 14.
Nbd2 (14. Rxa8 Qxa8) Rxa1 15. Qxa1 Qd7 16. Bxb6 cxb6

Here, the 9. Bc2 book exit (eval +0.29) is not premature, with the advantages that it (A) postpones the choice of playing the d-pawn to d3 or d4, and (B) it makes it possible to answer 9. ... b5 with 10. a4 if so desired. In this position, 9. d4 may be standard theory, but even then Bc2 is a move that often needs to be played anyway, so Bc2 right out of book looks at the least not worse. In Andscacs's 9. d4 line of game 72, 11. a3 (eval +0.28) was probably more accurate to avoid losing control of b4 after a pawn trade on d4; 11. Nbd2 exd4 would already have been drawish (eval +0.22).

For the next move in game 100, 10. a4 looks to be a sound choice, although both d3 and d4 were possible as well. This move does not yet necessitate the opening of (and trades on) the a-file; in fact, one of the most accurate continuations is 10. ... Bb7 11. d3 b4 12. a5 (eval +0.28). After 11. ... Bf6, not all of SF's moves may have been the most accurate, but by move 17 (eval +0.29), it at least got something concrete out of the opening in black's destroyed queenside pawn structure. In several variations, white can launch a make-believe kingside attack with the remaining pieces using moves like Bb3, Re1-e3-g3 etc, and use the resulting pressure to capture one of black's weak pawns. After this, black must be quite careful to keep the draw. It's not so much that 10. a4 and 13. axb5 hold the promise of a concrete win (they don't), but at least they offer an advantage and a chance for black mistakes when, again, it isn't clear that alternative moves would have been much better.

The bottom line about both book exits but especially the "h3" variation is that white's advantage is a good deal smaller than it looks at normal depth. Black's 8. ... Nd7 may not be the main line of the Averbakh variation, but after b5 and Nb6 it gives black meaningful queenside counterplay.
Up Topic The Rybka Lounge / Test Positions / TCEC S9 Stage 3, games 26/54 and 72/100 (IDeA tree analysis)

Powered by mwForum 2.27.4 © 1999-2012 Markus Wichitill