Two annotated Keres Attack games

My last two games on this site have been in the Keres Attack. Though far from perfect, both games have theoretical value, so I have chosen to annotate them and post the annotations. To aid analysis, I have used the chess engine SPIKE (version 1.0/960), which, though undoubtedly not the strongest engine on the market, is certainly quite sufficient for my purposes. I accept evaluations by SPIKE only after a minimum of 14 ply analysis.

Comments on the games or my annotations are welcome (as long as you're not too critical). 😉

Game #1
Game 3904776
1. Nf3 c5 2. e4 Nf6 3. Nc3 d6 4. d4 cxd4 5. Nxd4 e6 {Via a rather unusual order of moves, the game has transposed into the Scheveningen variation of the Sicilian Defense.} 6. g4 {This move initiates the Keres Attack. It has such a dangerous reputation, that many former Scheveningen devotees (such as GM Kasparov) will only play the Scheveningen via a transposition from the Be2 variation of the Najdorf.} h6 {This move is the most frequently played response to the Keres Attack. The alternatives here (with various values) include a6, Nc6, Be7, e5, and d5.} 7. h4 {This has become the most popular move (overtaking 7. g5).} Be7 {Possibly even more popular is Nc6} 8. Bg2 {The main line is 7. Rg1, but the text move and 8. Qf3 have their adherents} Nc6 {The text is the most popular, although GM Timman has shown that 8...Nfd7 is quite playable.} 9. g5 hxg5 10. Bxg5 {Usual here is 10. hxg5, but the text is fine and transposes to a version of the 7. g5 line. I have a pretty good library of books about the Scheveningen and only one (an old monograph by FM D. Eckert on the Keres Attack) considers this position.} a6 {If I had known White intended Nxc6, then the immediate Qb6 would be better than the text. However, the problem with 10...Qb6 is that White plays 11. Nb3 and follows it up with 12. Qe2! and this is known to be good for White. This is why black waits for Qd2 before playing Qb6 in this line.} 11. Qd2 Qb6 {Eckert suggests that White now has nothing better than transposing to the main line with 12. Nb3} 12. Nxc6!? {As far as I can tell, this is a TN and it's probably good too. Interestingly, SPIKE also prefers Nxc6 (+0.71) over Nb3 (+0.44). During the game, I assumed that my strong center ensured me a good game, but analysing the game with the aid of SPIKE has convinced me otherwise.} bxc6 13. O-O-O Rb8 14. b3 d5 {I think this is the most natural move in the position, though I did give serious consideration to e5. SPIKE gives d5, e5, and Qc5 all roughly equivalent evaluations, though in all cases, it evaluates the position as being in White's favor.} 15. exd5 {Natural enough, though SPIKE suggests maintaining the tension with Rhe1 may be somewhat better.} cxd5 16. f4?! {A serious inaccuracy, allowing Black to equalize. 16. h5 and 16. Rhe1 were both much better.} g6? {Unfortunately, I proceed to immediately return the favor. Correct was 16...Qb4 with equality. I was overly concerned with preventing f5 and h5 ideas, whereas I should have been more concerned with increasing my piece activity. SPIKE now evaluates White's advantage as being worth almost a whole pawn.} 17. Rhe1 Bb7 18. Bxf6? {Perhaps not an outright blunder, but still pretty bad, since it tips the evaluation from significant White advantage to a small Black advantage. 18. Na4 was much better and would have maintained the pressure.} Bxf6 19. Bxd5?! {19. Nxd5 was better, though Black has the advantage in either case.} Qd6? {Right piece, wrong square! Best was 19...Qb4 and Black wins material. Also 19...Rd8, though not as good as 19...Qb4, would have led to a moderate Black advantage. After the text, the game is roughly even.} 20. Kb1! {A very fine move, giving White the option of moving the Knight.} Kf8! {It is important to move the King to a less vulnerable square (relative to the Knight).} 21. Ne4 {After this, the following sequence is practically forced.} Qxd5 22. Qxd5 Bxd5 23. Nxf6 Bc6 24. Rd6 {24. h5 may have been a shade better, but the text should be ok.} Rc8 25. Ng4 Ke7 26. Rd4 f6! {It's critical to prevent Ne5.} 27. Rc4 Bb7!? {Simpler is 27...Rxh4, but the h-pawn isn't going anywhere.} 28. Rb4 Bf3 {Bc6 is equal and implicitly offers a draw. The text attempts to keep the game alive.} 29. Ne3 {Nf2 intending Nd3 is better. Now Black has a small advantage.} Rxh4 {At this point, White resigned, which came as a bit of a shock to me. Though I felt Black was somewhat better (and SPIKE agrees), I certainly didn't feel like it was an easy win. I'm guessing that my opponent was so disillusioned that the White attack had not proceeded as planned, that he just decided, with no guarantee of success (and realistically, a draw is the best he could hope for) it wasn't worth the effort to fight on. So, the game ends rather prematurely. Such is life.} 0-1

Game #2
Game 4889508
{It is worth noting that I have both books that IM Craig Pritchett has written on the Sicilian Scheveningen. My main source of information for this game is the more recent of the two (published in 2006).} 1. e4 c5 2. Nf3 d6 3. d4 Nf6 4. Nc3 cxd4 5. Nxd4 e6 6. g4 h6 7. h4 Be7 8. Bg2 {Up until my 7th move, I thought it was merely coincidence that this game was a repeat of the previous one. However, after White's 8th move, I began to wonder if my opponent had analyzed my previous game and prepared an improvement over my previous opponent's play. This worry turned out to be groundless as my opponent had a different variation in mind.} Nc6 {Should I ever reach this position again, I may opt for Timman's interesting 8...Nfd7} 9. g5 hxg5 10. hxg5 {This is the usual move here and the game has now diverged from the previous one.} Rxh1 11. Bxh1 Nh7 12. f4 {Pritchett considers 12. Nxc6 a reasonable alternative.} Nxd4 {Also possible are 12...g6 and 12...Nf8.} 13. Qxd4 e5 {The point of 12...Nxd4.} 14. fxe5 Nxg5 {Pritchett gives two games (both draws) that GM Rytshagov (spelling may vary in some databases), as Black, has played from this position. Hector-Rytshagov (1999) now continued 15. exd6 Ne6 16. Qd3 Bxd6 and was drawn on move 29.} 15. Be3 {This move was played in Van den Doel-Rytshagov (1995). My opponent and I proceed to follow that game for the next several moves.} Ne6 16. Qd5 Bg5 17. Bf2 dxe5 18. Qxe5 Qd2+ 19. Kf1 {Van den Doel-Rytshagov now continued 19...Qxc2 20. Nd5 Bd7 and Pritchett comments that this position is unclear.} Bd7 {Though I was not the least bit sure, it seemed to me that the text was an improvement over Rytshagov's materialistic 19...Qxc2. Unfortunately for me, analysis with SPIKE indicates the exact opposite. Nevertheless, despite the fact that Bd7 is inferior to Qxc2 (and a6 as well), SPIKE evaluates this position as being clearly in Black's favor. But consider that for that to be true, then this implies White must have made a serious error on a previous move! Backtracking, analysis with SPIKE indicates that Van den Doel's 15. Be3 deserves a question mark and leads to a very bad game for White, whereas Hector's 15. exd6 gives Black only a small edge. Interestingly, SPIKE states that White's best 15th move is 15. Bxg5! and evaluates the position as slightly in White's favor. We return to the game.} 20. Rd1 {A very natural looking move, but during the game, I suspected it was a mistake and analysis with SPIKE confirms this. SPIKE suggests 20. Qd5 as best though Black still has a large advantage after 20...Qxc2. Had White actually played 20. Qd5, it had been my intention to respond with the solid (but, as it turns out, inferior) 20...Bc6. After 21. Qxd2 Bxd2, SPIKE evaluates that line as moderately in Black's favor, though the win is far from assured.} Qxc2 21. Bf3 {After this move, the game is lost, though after the preferable Qd5, White would remain in great difficulties.} Bf6 {Not surprisingly, SPIKE prefers the materialistic 21...Qxb2, but the text move should be sufficient to preserve the win.} 22. Qd6 {Better here would have been Qd5 followed by Qb3, though White's prospects remain dismal.} Rd8 {Practically forced, but quite good.} 23. Qh2? {Looks tricky, but ultimately a blunder. Qa3 puts up more resistance.} Bxc3 {Winning material. 24. bxc3 would be met by 24...Bb5+.} 24. Qh8+ {Rxd7 would have prolonged the game, though the outcome is no longer in doubt.} Ke7 {I thought Nf8 might be a big mistake here because of Bc5, but as it turns out, Black has a forced mate in that line. Anyways, SPIKE confirms that Ke7 is best.} 25. Qh5 Bxb2 {25...g6 is more accurate.} 26. Bxa7 g6 {And here, Bc6 is more accurate.} 27. Qa5 {Hastening the end. 27. Qh4+ would have allowed White to play on a bit longer.} Qc3 {One of the big differences between humans and programs is that even when the game is already won, programs continue to be materialistic, whereas humans usually prefer to simplify into winning endgames where the opponent has no tactical tricks. SPIKE prefers Qc4 to win more material, but I prefer the text as it forces the exchange of Queens, making the win rather trivial.} 28. Qxc3 Bxc3 {And here, White had seen enough and resigned.} 0-1

Thank you wittywonka.

Good annotations.

Hi Korch, greetings from Athens, Greece;

Sharp game! I like the way Black handles it, however I think 7.h4 is a premature; the White should try 7. Rg1 and 8.Be3 in order to drive more carefully and prepare to castle queenside waiting a bit for his kingside development (i.e. Matulovic-Buckic 1-0, Yugoslav Ch. 1967/ Belyavsky-Ghinda 1-0, Bucharest 1980/ Liberzon-Formanek 1-0, Beersheva 1972). But anyway your game is a Scheveningen original and your evaluations look to me pretty correct (I play 100% Scheveningen when Black on 1.e4).

Best regards!

Originally posted by black beetle
Hi Korch, greetings from Athens, Greece;

Sharp game! I like the way Black handles it, however I think 7.h4 is a premature; the White should try 7. Rg1 and 8.Be3 in order to drive more carefully and prepare to castle queenside waiting a bit for his kingside development (i.e. Matulovic-Buckic 1-0, Yugoslav Ch. 1967/ Belyavsky-Ghinda 1-0, Bucharest 1980 ...[text shortened]... ations look to me pretty correct (I play 100% Scheveningen when Black on 1.e4).

Best regards!

Korch: Thank you for the compliment. I put a fair amount of time and effort into the annotations, so it's satisfying to know that it did not go unnoticed.

black beetle: A few decades ago, the most popular line was 7. g5 hxg5 8. Bxg5. This was superseded by the 7. h4 and 8. Rg1 (or 7. Rg1 8. h4) line, which in turn has been superseded by 7. h4 and 8. Qf3 (or 8. Bg2) which is the recommended line by Gavrikov in the recent book "Experts vs. the Sicilian". The line you prefer (7. Rg1 and 8. Be3) is not without venom, but it appears that more recently GMs have preferred other lines, perhaps because of games like Ghizdavu-Spassky, Nice 1974, Browne-Andersson, Milan 1975, and Karpov-Timman, Plovdiv 1983. A line becoming less popular among GMs is hardly an absolute condemnation of that line, but it may be an indicator that other lines are more promising. Nice to meet another practioner of the Scheveningen and good luck with your games on the site.

Hmmmm... Very interesting games 🙂