Unfortunately our series of annotated correspondence games is about to end (there is some AlphaZero stuff in the making, though). I enjoyed the journey, but despite the fun and the growing ambition while competing with the best guys this kind of chess turned out to be too time consuming (this game should give you an idea of the work required). Also I found back to over the board chess eventually.
So, for now this is the final game in our correspondence series. One of my favourites, also one of the most complex I ever played. I tried hard to make most of the ideas and concepts involved accessible via annotations and sample lines.
White manages to come out of the opening with a slight pull, and just by relentlessly fueling the fire (and not counting material) the pull gets stronger and stronger until it is too much to handle for Black. I couldn’t even pin down the mistakes (i.e. microblunders) he made, all I can say is that at some point suddenly White is winning. Of course no human on his or her own could play like this, still this game illustrates well what a powerful tool the initiative is even on a 3.600 Elo level.
For you to replay and analyze more conveniently I have put all the moves and lines below into a Lichess study. The fully annotated game represents the latest addition to our database with annotated games, game excerpts and opening surveys, available via the PayPal link on the right.
Not only is this a cool piece of chess, the game was the decisive one in order to win a tournament in the highest league. Such a win is required to secure a direct place in the quarterfinal of the LSS World Championship 2020 after missing the preliminaries. Another reason to be happy about it.
Hero (2.208) – Villain (2.199)
Lechenicher Schachserver 2017, Grünfeld Indian Defense
1. d4 Nf6 2. c4 g6
Grünfeld or Kingsindian, whatever Black is planning, I was happy to see him announce he wants a fight instead of calmly trying to equalize.
Discourages Black from going Grünfeld immediately. Without a white knight on c3 …d7-d5 makes less sense since after cxd5 Nxd5 e4 the typical …Nxc3 bxc3 c5 with pressure on the white centre isn’t available. However, while f3 is a perfectly fine place for the knight the move is somewhat inflexible. Anything with f2-f3 and/or Ng1-e2 is out of the picture now, the Sämisch Kingsindian for example.
(3. h4 Since early h2-h4 against the Grünfeld is the topic of this game Topalov’s misguided attempt 3.h4 against Giri in the 2016 candidates should be noted. I remember how excited I initially was when watching this live, it seemed like Toppi was burning bridges and going for Giri’s throat from the get-go. But after Giri’s cool reaction I soon calmed down. The Dutchman seized the opportunity to get a pleasant Benkö Gambit (in which the move h2-h4 makes no sense). Already at move 5 it became obvious that White is just a tempo down compared to regular Benkös. 3… c5 4. d5 b5 5. cxb5 a6 6. bxa6 Bg7 Topalov,V (2780)-Giri, A (2793), Moscow 2016, draw (68).)
(3… d5 Few grandmasters keep insisting that 3…d5 is playable nonetheless, but the vast majority won’t let White build a perfect e4/d4 pawn centre for free. With …Nxc3 bxc3 c5 not available Black will have to attack it later via … f5, and that looks a bit artificial and lose. 4. cxd5 Nxd5 5. e4 Nb6 6. h3 Bg7 7. Nc3 O-O 8. Be2 Nc6 9. Be3 f5 could qualify as a tabiya of 3.Nf3 d5. Mamedyarov and Korchnoi among others played this with the black pieces. I wouldn’t.)
4. Nc3 d5
Now we have a Grünfeld proper. White can choose between a ton of reasonable moves, 5.Bg5, 5.cxd5, 5.e3, 5.Bf4 or 5.Qb3 for instance.
Over the span of a century of Grünfeld games this is the 8th most popular 5th move. But if you only consider games played since 2012 5.h4 is pretty hot stuff. It had been tried by some unknowns, then in 2012 the great Alexander Morozevich took it into battle during a blitz and rapid tournament in China. Inspired by his countryman Alexander Grischuk used it in the 2013 candidates to surprise Magnus Carlsen. Since then 5.h4 is played regularly.
Other than Topalov’s 3.h4 here the advance of the h pawn makes sense. Black’s centre is somewhat shaky, so White immediately attacks on the wing. To stop him via …h5 would only open the g5 square for a White bishop that would feel perfectly at home at g5. So Black must react with central play, either counterstrike (5…dxc4, 6…c5) or stabilize (5…c6).
(5. g3 is the 6th most popular, but probably inferior. Recently at the Grenke tournament after only ten moves German grandmaster Georg Meier (a theory wizard usually) managed to be worse against Magnus Carlsen. 5… dxc4 6. Qa4+ Nfd7 “I already like Black,” said Peter Leko when doing live commentary. 7. Qxc4 Nb6 8. Qd3 O-O 9. Bf4 Nc6 10. Rd1 Bf5 Meier,G (2621)-Carlsen, M (2838), Karlsruhe 2017, 0-1 (41).)
The fighting move, heading towards a huge mess.
(5… c6 Leads to calmer, more positional play than 5…dxc4. 6. Bg5 h6 (6… O-O “Show me what you got.” (M. Carlsen) 7. Bxf6 (7. h5 Nxh5 and it is not at all clear how White wants to initiate a meaningful attack.) (White would like to play 7. e3?!, but after 7… Nh5! Black threatens …h6 and the Bf4 is in trouble.) 7… Bxf6 8. h5 Bf5 This guy makes sure that the black kingside will be safe. 9. e3 e6 10. hxg6 Bxg6! (10… hxg6 11. Ne5 and g2-g4 is in the air.) 11. Bd3 Nd7 and Black was solid in Grischuk,A (2764)-Carlsen,M (2872), London 2013, draw (23).) 7. Bxf6 Bxf6 8. e3 e6 9. Qc2 Nd7 10. O-O-O O-O is the most recent top level game in the 5.h4 line, played during the Grand Prix in Moscow, Tomashevsky,E (2696)-Svidler,P (2755), Moscow 2016, draw (25).)
6. e4 c5 7. d5 b5 8. h5
(8… Nxh5 9. Nxb5 O-O is hardly explored, but seems playable and rather unclear. To me this looks like one of the cases where both sides have reason to complain about their poor position. (Edit: In fact both Peter Svidler and MVL, probably the two greatest Gründfeld experts in professional chess, tried 8…Nxh5 shortly after this correspondence game was published for the first time on r/chess.))
(8… b4? ends badly after 9. h6)
9. hxg6 fxg6!
Usually we automatically capture towards the centre, but in this case the counterintuitive …fxg6 is necessary to keep the h file closed and open the f file.
(After 9… hxg6? 10. Bh6 White threatens Qd2 or Qc1 (10…b4 11.Qc1!).)
10. Rh4!? (N)
This would be a novelty, had I not played it before in another correspondence game. The machine keeps showing the odd looking 10.Rh4 among its top choices in this position. In the earlier game in which I had played 10.Rh4 for the first time I had ignored a move as ugly as this one initially, but when I wasn’t able to find anything convincing in the main lines after 10.e5 I decided to at least briefly check the machine move. Well, in the end I checked it quite extensively and liked what I found. 10.Rh4 doesn’t win or anything, but it leads to a completely unexplored mess, and that’s as good as you can hope for in correspondence chess against good opponents. Here 10.Rh4 comes with the advantage that the villain won’t have any theory to rely on while I already have a file on 10.Rh4.
The move may look stupid, but there is some sense to it. First of all it covers e4 so that 11.Nxb5 (followed by Bxc4) is a threat right now that Black has to take care of. Somewhat deeper down the road the rook may discourage Black from installing pieces on g4 since there will be pleasant exchange sacrifices. And in some lines (see below) the rook may even become an attacker.
(10. d6!? N would be a novelty as well. It will likely transpose into the main line after 10…e6, but it’s also an invitation for Black to try 10… exd6 or 10… Qxd6 (feel free to ask your engine about these two, they`re quite interesting). 10… e6 11. e5 Ng4 12. Rxh7 would be a transposition (see 10.e5).)
(10. e5 Ng4 11. d6 e6 12. Rxh7 is the tabyia of the 5…dxc4 line. White has advanced his central pawns deep into the black territory, but they won’t survive there for long. White needs to throw everything against Black’s king immediately as long as the e5/d6 combo hinders the black coordination. Taking the rook will lead to trouble as 2.700 GM Markus Ragger found out in a game against fellow 2.700 GM Nikita Vitiugov in 2013.12…Rf5 (probably best), 12…Nd7 and 12…Nc6 seem to be the reasonable tries with unclear consequences. 12… Nc6 (12… Kxh7 $2 13. Ng5+ Kg8 14. Qxg4 $16 Rf5 15. Be3 Nc6 16. Qh4 Nxe5 17. O-O-O Bd7 18. Nce4 and White was more than happy in Vitiugov,N (2719)-Ragger,M (2680), Tromsoe 2013, 1-0 (31).) 13. Rxg7+ Kxg7 14. Bg5 was a World Cup game Morozevich-Ponomariov in 2013.)
(Now 10. Bh6? isn’t anything: 10… Bxh6 11. Rxh6 Bg4 and Black is better already.)
Stabilizes his queenside pawn mass for now. Later in the game Black will want to get his 4 vs 2 majority rolling, it will be his major (if not only) source for counterplay against the
white advances towards his king.
(All 10… b4?! 11. Na4 accomplishes is that now the pawns c4/c5 are hanging. The game remains to be complicated after 11… c3, but Black has to fight hard to keep the balance.)
Since the time to exchange blows hasn’t come yet, White develops a piece and indirectly covers g4 (where we don’t want Black to anchor anything). White badly wants to advance in the centre, but he needs to wait until Black can’t answer e4-e5 with …Nfd7. For Black it’s not so easy to find a meaningful developing move exept the one he executed.
(11. e5?! Too early, now the advanced White centre becomes shaky before it can have any effect. 11… Nfd7 12. e6 (12. Bf4 Rxf4! 13. Rxf4 Nxe5 =+) (12. d6 e6 =+) 12… Bxc3+ 13. bxc3 Nf6 and the d5 pawn is a goner. White can hang in there with 14.Ng5, but Black has entered the driver’s seat.)
That’s the one White was waiting for. With … Nfd7 no longer possible the time has come to burn bridges.
(12… Nh5 13. d6 The immediate idea of 13.d6 is Qd5+ with an attack on the Ra8. 13… Nxe5 (13… e6?! 14. Bg5 Qe8 15. Qd2 an White has the better game. He plans 0-0-0 and Ne4 with strong kingside play. 15… Nxe5? 16. Nxe5 Bxe5 17. Bxh5 and White is close to winning already.) 14. Rxh5! Now Qd5+ is a threat. (The immediate 14. Qd5?? backfires: 14… e6 15. Qxa8 Nxf3+ 16. Bxf3 Qxh4 -+) 14… Qxd6 15. Qxd6 exd6 16. Rxe5! dxe5 17. Ng5 += (diagram) With a rook and three pawn against two knights the material balance looks exellent for Black. However, his majority on the queenside is immobile, his pair of bishops without sting, c5 may become weak and e4 a beautiful post for a White knight. It’s a fight, but White should be better.)
Threatening Qd5+. Also Nd5 combined with Bg5 is a strong idea as long as the e5/d6 combo harasses Black’s dark squares.
(14. Nd5!? The machine loves this one since it basically wins a piece due to the threat 15. Bg5. 14… exd6 (14… Bxd5? 15. Qxd5+ Kh8 16. Ng5 game over.) (14… Nxe5? 15. Nxe7+ Kh8 16. Ng5 game over.) 15. Bg5 Ndf6 16. exf6 Nxf6 Looks unclear to me. White has pleasant piece play and some attacking prospects for now, but once Black stabilizes and his pawn mass starts rolling it will be quite fearsome.)
(15. Nd5!? Qxd6 16. Nxe7+ Kf7 (16… Kh8? 17. Nxe5 Bxe5 18. Qxd6 Bxd6 19. Nxg6+ Kg8 20. Nxf8 Bxf8 21. O-O-O +-) 17. Rf4+ It’s not so stupid to have a rook on the 4th rank after all 😉 17… Ke6 18. Nd4+! (diagram) The only move, and a spectacular one. The idea is to involve the Be2 into the king’s hunt. 18… Kd7 (or 18… cxd4 19. Bg4+ Rf5! 20. Nxf5 gxf5 21. Bxf5+ Kf7 and White has a bunch of possible checks to continue the king’s hunt, but according to the machine it won’t lead him anywhere. The position is unclear.) 19. Bg4+ Kc7 The king escapes to the queenside, and White doesn’t have anything better than grabbing some material at least. 20. Ne6+ Kb6 21. Nxf8 With an unclear position that eludes human evaluation.)
15… Qd7 16. Bxf8
Black doesn’t feel bad about giving up an exchange. The absence of White’s dark squared bishop will help him stabilize things around his king. The black pair of bishops looks quite forceful, while none of the white pieces makes much of an impression.
16… Kxf8 17. Rf4+ Kg8 18. Ne4
However, White has one powerful trump left, the pawn on d6. With his attack fizzling out somewhat White must keep this sting in Black’s flesh alive in order to keep his prospects intact. And so 18.Ne4 indicates a strategy switch. Instead of sacrificing his central pawns in order to crush through with brute force, White holds on to his d6 passer and tries to combine threats connected to the passer with play against the black kingside.
In order to get rid of d6 Black wants to go …Nf7 and …Rd8, but he needs to cover c5 first.
(18… Nxf3+ 19. Bxf3 Bxb2 Black can’t afford this immediate try to take over yet. 20. Bg4 Qg7 21. Be6+ Kh8 22. Rf7 is good for White.)
(18… Bxe4 Also the try to get rid of the d6 intruder immediately turns out favourable for White: 19. Rxe4 Nf7 (19… Qxd6 20. Nxe5 Bxe5 21. Qxd6 Bxd6 22. Rd1 +-) 20. Re7 Qxd6 21. Qxd6 Nexd6 22. O-O-O +=)
The only move for White to hold on to the initiative, generated by machine power and a helping human hand that enabled the machine to look deeper than it could on its own. It’s hard to get for a human, but the main purpose of covering f4 via 19.g3 is to prevent Black from going 19…Rd8.
Quite logical. Black puts more pressure on d6 and opens the long diagonal, eyeing the b2 pawn and the queenside where all his hopes lay. The game is very much on the edge now. If White doesn’t act energetically, Black will take over. Fortunately I had an exchange sacrifice prepared to keep the initiative.
(After 19… Rd8?! we see what 19.g3 was good for (24.gxf4): 20. Nxe5 Bxe5 21. Bf3 Bxf4? 22. Nf6+ Nxf6 23. Bxc6 Bxc6 24. gxf4 +-)
20. Rxf7! Kxf7 21. Nfg5+ Kg8
When playing 19.g3 I already had a bunch of analysis on this position, but still wasn’t sure what to do. One thing is clear: White cannot afford anything timid and slow (i.e. protect b2), he needs to continue pushing. While the tempting 22.d7 leads to disaster, both 22.Bf3 and 22.a4 keep White in control. Only which one is the better move?
(22. a4 Usually it’s better to hold still and not open anything up on the side where the opponent is stronger. In this case, though, 22.a4 has a concrete idea that makes it a good move. It threatens 22. axb5 axb5 23.Rxa8 followed by d6-d7, and this is not easy to parry for Black and keep the queenside structure intact at the same time.)
(The tempting 22. d7? runs into the devilish 22… Bd4!! 23. dxe8=Q+ Rxe8 and Black wins.)
Once more White has a difficult choice how to push for more. While the obvious 23.Nf6+ leads to nothing, 23.Rb1 and the less obvious 23.Kf1 are reasonable tries. White has two main concerns: connect his unemployed queen to his army of light pieces and get the king out of its exposed spot on e1. The line after 22.d7 Bd4!! illustrated how dangerous it is to lurk around in the centre with the king while the rest of the board is on fire.
The second exchange sacrifice within four moves, and a logical one. While the Ra1 isn’t doing anything, Black’s dark squared bishop does a life saving job as the defender of the dark squares around Black’s king. Also the sequence …Bxa1 Qxa1 immediately includes the white queen into the play against the black monarch.
(23. Nf6+?! Nxf6 24. Bxc6 Bxc6 and Black is fine.)
(23. Rb1 A close second to 23.Kf1. 23… Bd4 24. Qe2 makes Nf6+ a serious threat since now the option Qe6+ is in the mix. (24. Nf6+?! too early: 24… Nxf6 25. Bxc6 Bxc6 and Black has plenty of compensation.) 24… Qd7 Gets out of the long diagonal and covers e6, but not for long. Sooner rather than later White will go Bg4, and the weakness of the e6 square will be a major issue for Black. White is clearly better.)
(23… Rd8 24. Nf6+ Bxf6 25. Bxc6 Bxc6 26. Qg4 Ng7 27. Re1 with a huge initiative. This line shows another point of 23.Kf1. Not only is the king safe, the e file is free to include the rook into the attack (in case Black doesn’t take it, of course).)
Black needs to deal with the Nf6+ threat.
He decides to give back some material to get rid of the nasty d6 pawn at least.
(24… Qc8 The try to stick to the material leads to a fascinating, yet lost endgame. 25. Nf6+ (25. Qd1!?) 25… Nxf6 26. Qxf6 Qd7 27. Bg4 Qg7 28. Be6+ Kh8 29. Nf7+ Kg8 30. Nd8+ Kh8 31. Qxg7+ Kxg7 32. Nxb7+- (diagram). Soon Black will have to give up his rook for the d passer, but in addition to his queenside pawn mass he will also create a passer on the h file. This is eventually won for White, but despite being two pieces up temporarily the win is far from trivial. It required a lot of analysis to make sure that the black pawns can be stopped.)
25. Nxc5 Qxd6 26. Nxb7 Qf6
After getting rid of the d6 intruder Black looks for further relief by exchanging queens, something that White cannot allow. He needs to continue his play against the black monarch, keep the initiative. With queens off the board Black would have reached a more than comfortable ending.
27. Bd5+ Kg7 28. Ne6+ Kh8 29. Qd1
Since White needs to avoid the exchange of queens, Black finally gets a short break and an opportunity to consolidate and coordinate. But instead of one move he would need at least two (…Rc8 and …Ng7) in order to repair the most urgent damage.
(29… Rc8 Getting the rook out of the vis-a-vis with the Bd5 and behind the c passer looks tempting, but Black can’t afford it. A sample line from the ocean of possibilities: 30. Nbc5 Ng7 31. Ne4 Qe5 32. N6g5 Rf8 33. Nf7+! Rxf7 34. Bxf7 Qxe4 35. Qd8+ Ne8 36. Qxe8+ Qxe8 37. Bxe8 +- was one of dozens of favourable, but critical endgames I needed to check in order to make sure there aren’t any hidden saves for Black.)
30. Nd6 Rb8 31. Ng5
31…Qxd6?? fails to 32.Nf7+ of course, but Black can now at least exchange one of the white knights that are dancing around his king. However, he can’t do so without making a crucial structural concession.
31… Nf5 32. Nxf5 gxf5
(32… Qxf5?? 33. Qd4+ and checkmate. The black queen must stay on the long diagonal under all circumstances, otherwise it’s game over.)
33. Ne6 Rc8 34. Nf4
The perfect place for the knight and the position I was aiming for after I had figured out how I wanted to continue after 27…Qf6. It’s not obvious, took me a lot of time to convince myself, but it very much seems that White is winning at this point. His piece play is more powerful than the combined black passers ever will be. In fact even at absurd depths the best defense attempt for Black the machine comes up with is sacrificing some pawns in order to open lines for the rook and generate counterplay. When making sure this is won for White I tried do do better and hold the black position against the machine (with help from a machine), but couldn’t do it. Still, this is far from being resignable, there’s still a lot of work to do and some drawn endgams to avoid for White.
34… c3 35. Bb3 c2 36. Bxc2 Qb2
At the expense of the c pawn Black has created some activity himself.
Answers Black’s threats on the c file with counterthreats against the Black king. If Black does nothing now, White goes 38.Qd7 winnning.
The only move to avoid 38.Qd7 (which would now be answered by 38…Rxc2, and White has nothing better than to give a perpetual).
A move extremely hard to play for a human. Discoordinating my guys and presenting them on the c file felt uncomfortable even with the machine telling me that all is fine. But in fact 38.Nc7! is the only move that wins. The ideas behind it are manifold and purely tactical. Most of all it renews the threat 39.Qd7 which would be deadly if Black plays a null move like 38…b4?. in some lines the knight wants to jump via d5 to e3, cover c2 and hunt down the f5 pawn.
Prevents 39.Qd7 since now Black could go 39…Rxc7.
And that, finally, is the third idea behind 38.Nc7, a profane one: Take the pawn a6. But the pawn hunting knight is out of the game for now, and Black uses its absence to create play against the white king.
Besides, Black likes to get rid of his isolani, and he likes to trade pawns. In an ideal world he reaches an ending rook vs bishop, knight and a pawn, which is drawn as soon as the defending king can get in front of the passer. (Yes, it may surprise you, but this kind of ending is drawn. Check with a tablebase if you like.)
40. Qd3 Qg7 41. Nb4 fxg3 42. fxg3
One pawn exchange on the kingside is fine, but White very much wants to keep the g pawn alive in order to avoid the aforementioned theoretical draw with R vs K, B and a passer.
42… Rf8+ 43. Kg2 Qf7 44. Qxb5 Qf3+
Black will have a lot of disturbing checks, but they won’t lead to anything serious. His main idea to save the game remains exchanging the last pair of pawns on the kingside.
45. Kh2 Qe3 46. Qd5 Qf2+ 47. Kh3 Qf1+ 48. Qg2 Qf7 49. Qe2 Re8 50. Qd1 Qe6+ 51. Kg2 Qe3 52. Bb3 Qe4+ 53. Qf3 h5 54. Nc6 h4 55. gxh4 Qxh4
Black has reached his goal but no way to reach a theoretical draw. With the queens still on the board White can combine attacking ideas with the march of the a pawn. Black still can give a lot of checks, but eventually he will be lost. Even if Black manages to exchange queens, White will be able to only allow this with the black king too far offside so that the position will remain won.
56. Ne5! Qg5+ 57. Ng4
Thankfully Black decided to resign instead of dragging this out for another few weeks or months.