Annotated Game #361: The Stonewall does its job, but I let it down

Often you will see players complaining about how their openings let them down, but I tend to have the reverse problem. The opening gives me the best result it can, but I let it down instead by not fulfilling its promise. The following last-round tournament game is a perfect illustration of this, as I could not have asked for anything more out of the Stonewall as Black, but was not able to press an overwhelming attack to its conclusion. However, it is precisely these kinds of specific failures that merit close analysis and study. For example, here a key idea was for me to play ...Ne4, the same blind spot recently highlighted with colors reversed in Annotated Game #360.

[Event "?"] [Site "?"] [Date "????.??.??"] [Round "?"] [White "Class B"] [Black "ChessAdmin"] [Result "1-0"] [Annotator "ChessAdmin/Dragon 3.2"] [ECO "A85"] [PlyCount "69"] [GameId "2339533383637239"] 1. Nf3 d5 2. d4 c6 3. c4 e6 4. Nc3 f5 {we now have a Slav Stonewall.} 5. e3 Nf6 6. Bd3 Bd6 7. Bd2 O-O 8. Qc2 Ne4 {correct Stonewall play. The knight hops into the e4 strong point and bottles up White's pieces, while projecting its influence into the enemy camp.} 9. cxd5 {I got the impression my opponent was expecting/hoping for a recapture with the c-pawn, which is in fact OK for Black. However, the superior recapture on d5 in the Stonewall formation is usually with the e-pawn, as it gives the Bc8 more scope while keeping the c-file from opening.} exd5 $15 10. Ne2 {White looks to transfer his knight to a more useful square, while keeping castling options open.} Nd7 {controlling e5, with the option to go to f6.} 11. Rf1 $2 {with the intent of protecting f2 prior to castling queenside. However, this is too slow.} (11. O-O $15 {the engine prefers kingside castling, although Black would have a comfortable, classic Stonewall kingside attack plan afterwards.}) 11... Ndf6 $17 {while this is a good move and increases the activity of both the knight and the Bc8, I should have better remembered the principle of forcing open the position when the opponent's king is in the center.} (11... c5 $5 {dissolving the Stonewall formation with the ...c5 break is an art, which I need more experience with. Here my pieces fully control c5 and White is in no position to exploit the c-file. I also immediately threaten ...c4 and pose my opponent more problems.} 12. dxc5 Ndxc5 $17 {with the simple but effective threat to follow of ...Bd7 followed by ...Rc8.}) 12. O-O-O Qe7 (12... Nxd2 {it is also an art knowing when to "cash in" the Ne4 with a piece exchange. Here the engine assesses eliminating White's dark-square bishop is worth it.} 13. Nxd2 Qe7 $17) 13. Be1 {this looks strange but the engine validates it as the best available move. White overprotects f2, thereby freeing up the rook; the bishop was doing nothing on the c1-h6 diagonal anyway.} Bd7 14. Kb1 a5 $6 {it would be better to "invite all the pieces to the party" before moving ahead with the pawn storm attack.} (14... Rfc8 $17) 15. Ng3 $2 {it is difficult for White to find something useful to do, but this was simply detrimental to my opponent's defense. Now the pawn storm gains significantly in strength, with one less piece available to defend the king.} b5 $1 $19 16. Nd2 a4 {the most straightforward continuation of the attack.} (16... Nxg3 17. hxg3 a4 $1 {is perhaps a better version of the idea, found by the engine. At the time I wanted to keep the f-pawn, but snatching it would be too materialistic, eliminating the defending light-square bishop in exchange and displacing White's queen for a tempo.} 18. Bxf5 Bxf5 19. Qxf5 a3 $1 20. b3 (20. Qc2 axb2 $19) 20... Ne4 $19) 17. Ndxe4 fxe4 {by this point I was (correctly) confident in the overwhelming strategic advantage I had, with White's pieces in disarray and his king in the crosshairs.} (17... dxe4 {might be even better, to open the g8-a2 diagonal for the attack.}) 18. Be2 b4 19. Qc1 Rfc8 {the strongest placement for the rook, creating threats opposite the queen.} 20. f3 c5 {correctly ignoring White's attempt at counterplay and driving forward with the pawn lever.} 21. fxe4 {I spend a good deal of time on the next few moves, knowing they were critical.} cxd4 {a "good enough" move.} (21... Bxg3 {is evaluated as even stronger by the engine. The dark-square bishop is not required to support the attack, and eliminating the defender of e4 would allow my knight into a strong central post.} 22. e5 Ne4 23. hxg3 cxd4 $19) 22. Qd2 b3 {still playing strong moves, this was the best follow-up.} 23. exd4 {I now had the longest think of the game and was beginning to tire of the calculations.} Rc2 {still "good enough" for an attacking continuation.} (23... a3 $1 {I rejected this during the game because I did not see the key follow-up after} 24. axb3 {as} Bxg3 $1 {again the idea of enabling ...Ne4} 25. hxg3 Nxe4 26. Qb4 Nf2 $1 (26... Nc3+ {also works.})) 24. Qe3 Ba3 $6 {not the best.} (24... Bxg3) (24... Ng4) 25. Bc3 {this surprised me and the resulting complicated calculations wore me out.} Rc8 {the correct reaction, with the Ba3 tactically protected.} 26. Rd2 Rxd2 $4 {I thought this was the "bail-out" line, but it has a fatal flaw. I should have gone for it with} (26... bxa2+ {which I decided against because} 27. Kxc2 {I thought ended all my attacking chances, but Black has} Qb4 $1 $19 {with unstoppable threats, due to the pin on the Bc3 and the lack of escape squares for White's king.}) 27. Qxd2 $18 {the problem I did not see until too late is that the Ba3 has no retreat square, due to a pawn fork on e5.} dxe4 {I decided on this as the best practical attempt at obtaining counterplay, but my opponent rises to the occasion and closes out the game with solid play.} 28. bxa3 $18 Qxa3 29. Bb4 $1 bxa2+ 30. Qxa2+ Qxa2+ 31. Kxa2 Nd5 32. Bc5 Rb8 33. Bc4 Be6 34. Ka3 e3 35. Nf5 1-0

Annotated Game #360: Don't let that Black knight into e4

A systematic review of your own games inevitably turns up recurring themes in your losses. Ideally, you can then better understand and avoid them in the future, in the process advancing your practical chess knowledge and becoming stronger for it. This next tournament game produced several useful insights for me, including to keep looking for useful ways to play an e-pawn advance in the Colle, but the decisive one was witnessing how my opponent, as Black, obtained his final decisive edge by placing a strong knight on e4. I have lost too many times when this has happened - not just in the Colle - so it is obviously a threat I need to have on my radar. How Black's knights combine in the end to dominate my pieces, along with the queen, makes a strong impression.

[Event "?"] [Site "?"] [Date "????.??.??"] [Round "?"] [White "ChessAdmin"] [Black "Class A"] [Result "0-1"] [Annotator "ChessAdmin / Dragon 3.2"] [ECO "D04"] [PlyCount "72"] [GameId "2340929602847893"] 1. d4 Nf6 2. Nf3 d5 3. e3 c5 {a reverse Queen's Gambit.} 4. dxc5 e6 5. a3 Bxc5 6. b4 Be7 7. Bb2 a5 {Black chooses to immediately challenge the b-pawn. I had not seen this before.} 8. b5 {the correct decision, validated by the engine. Otherwise opening lines on the queenside, or alternatively supporting the pawn again with c2-c3 but shutting in the Bb2, would be better for Black..} Nbd7 9. Be2 $11 {deliberately holding back on c2-c4 for the moment, although now would be the first (and perhaps best?) opportunity to play the pawn break.} Qc7 {I thought that the intent of this move was to restrict c2-c4 - and it probably was - but the engine shows that it can be played anyway.} 10. Nbd2 (10. c4 $5 dxc4 11. O-O {White can afford to keep on with development and then round up the undefended Black c-pawn, with a small advantage. Pursuing materialism would be unhelpful for Black, for example} Nb6 $6 12. Be5 Qc5 13. Bd4 Qxb5 $6 14. Nc3 Qa6 15. Qb1 Nbd5 16. Nb5 Bd7 17. Bxc4 $14) 10... O-O 11. O-O Nb6 12. Be5 $6 {good enough for equality, but White has better.} (12. c4 {and again this is only a temporary pawn sacrifice.} dxc4 13. Rc1 Rd8 14. Qe1 $16 {White will have good play down the c-file and superior minor pieces.}) (12. a4) 12... Bd6 13. Bxd6 Qxd6 14. c4 a4 {controlling b3 and restricting the Nd2.} 15. Qc2 {developing the queen and connecting the rooks.} Bd7 {Black connects his in response.} 16. Qb2 {again, good enough to maintain the balance. The queen moves off the soon-to-be-opened c-file and occupies the long diagonal, once again controlling e5.} (16. e4 $5 $14 {this thematic Colle e-pawn break would give White's position more life.}) 16... Rfc8 17. Rfd1 {activating the rook and placing it on the d-file opposite the queen.} Qf8 18. Ne5 Be8 {this looks a little funny, but correctly preserves the bishop.} 19. Rac1 Rc5 {apparently intending to double rooks on the c-file, so I proceed with the central pawn exchange before that can happen.} 20. cxd5 Rxd5 $6 {an unnecessarily awkward placement of the rook, but I did not take advantage of it.} 21. Ndf3 $6 (21. e4 {is the most direct way, gaining the initiative.} Rc5 22. Rxc5 Qxc5 23. Rc1 Qe7 24. Nec4 Nxc4 25. Rxc4 $14) 21... Rad8 22. Rxd5 Rxd5 23. h3 {here I was cautious and wanted to eliminate potential back-rank mate threats.} (23. Rc7 $5 {does not promise much more.} Qd6 24. Rxb7 Rd1+ 25. Bf1 $11 {White is a pawn up but Black has full compensation with his greater piece activity; White has no remaining threats.}) 23... Qd6 {preventing the rook from getting to the 7th rank.} 24. Ng4 $2 {unfortunately I now run out of ideas and allow Black to establish a strong knight in the center as a result. I recognized it was best to move the knight, but this is the wrong square.} (24. Nc4 {this would at least exchange down further towards a draw.} Nxc4 25. Bxc4 Rc5 26. Bf1 Ne4 {now this is not as threatening.} 27. Rc2 $11) 24... Ne4 $1 $17 25. Nd4 $6 {the centralized knight looks good, but can also be a target.} (25. Nge5 $5 Rc5 $17 {with similar issues, but not as severe.}) 25... Rc5 {now Black does not have an immediate knockout blow, but I cannot cover all of my weaknesses in the medium term, including the back rank, a- and b-pawns, and squares in reach of the Ne4.} 26. Bf3 $2 {the wrong way of kicking the knight. The main defect is that Bf1 is no longer possible to defend on the 1st rank.} (26. Rxc5 {trading down does not work either.} Qxc5 27. Bf3 Nd6 28. Be2 e5 $19) 26... Rxc1+ $1 27. Qxc1 Nc5 $6 {this gives me some breathing room.} (27... Qc5 $1 28. Qxc5 Nxc5 29. Be2 Ne4 $19 {and the engine assesses the endgame is won for Black, due to the aforementioned queenside weaknesses.}) 28. Qc3 h5 {giving the Black king some luft, with tempo.} 29. Nh2 g6 30. g4 {a reasonably good try at activity, but Black has a good counter in the center.} (30. Qb4 $5 e5 31. Nc2 $15 {looks ugly, but at least my kingside is intact.}) 30... e5 $1 31. Nc2 h4 {my opponent played this confidently. At first I thought he had over-reached.} 32. Qb4 {the best practical way to pose problems for Black, as in the above variation. Indeed, my opponent immediately went astray.} Nd5 $6 {my opponent wants to break the pin by counterattacking, and succeeds at that, but my queen now goes to} 33. Qd2 $1 $11 {with an effective new pin on the d-file.} e4 34. Be2 $4 {this "obvious move" retreating the attacked bishop only loses.} (34. Nb4 $1 $11 {the idea here, which I did not seek out, is the tactical trade of pieces} exf3 35. Nxf3 Bxb5 36. Qxd5 $11) 34... Qe5 $17 (34... Nb3 $1 $19 {is even stronger.}) 35. Nf1 $4 {after this useless move, my opponent now finds the devastating} (35. Bf1 Nb3 36. Qd1 g5 $17 {and my position is miserable, but not yet dead.}) 35... Nb3 $1 $19 {as the two knights and queen combine in a highly effective manner.} 36. Qd1 Nc3 {the b-pawn will fall after the Be2 is forcibly exchanged, then Black will simply dominate.} 0-1

Annotated Game #359: Rook activity ftw

In this second-round game, my opponent as White plays an unusual knight retreat in the opening, a Caro-Kann Exchange, which gives me easy equality. I correctly do not try to punish the move - which is solid enough, if not theoretically valid - and instead play similarly solid waiting moves to see how the situation evolves. A certain lack of patience (or prudence) ends up costing my d-pawn, but that galvanized me to seek additional piece activity as compensation. It is instructive to see how much difference the level of rook activity makes in a short time span, as my superior heavy pieces quickly were able to achieve full compensation. 

After rejecting a draw offer in an objectively equal position, I leveraged the momentum and initiative for additional pressure, under which my opponent soon crumbled. Although this was not a clean game, I was pleased to see it result in a (flawed) win after correctly activating my pieces, a general area for improvement. (Naturally, it would be better to maximize piece activity before committing inaccuracies that give my opponent a strategically won game...)

[Event "?"] [Site "?"] [Date "????.??.??"] [Round "?"] [White "Class B"] [Black "ChessAdmin"] [Result "0-1"] [Annotator "ChessAdmin/Dragon 3.2"] [ECO "B13"] [PlyCount "68"] [GameId "2338400286845303"] {[%evp 0,68,10,35,34,32,31,12,14,8,23,18,21,19,18,22,34,-13,-41,-15,-6,4,8,5,12,-10,11,15,8,1,15,31,17,14,17,-33,11,19,39,0,146,32,107,98,110,96,80,85,116,172,136,79,82,60,55,-13,-107,-36,-3,-355,-409,-365,-390,-345,-404,-427,-421,-29987,-29988,-29995,-29996]} 1. e4 c6 2. d4 d5 3. exd5 cxd5 4. Bd3 Nc6 5. c3 Nf6 6. Bf4 g6 7. Nd2 {the second most popular choice in the database, after Nf3. A key difference is that with the text move, the Bd3 is now undefended.} Bf5 {this is the point behind playing ...g6, to support the bishop development on f5 and pose the question to White whether to exchange there.} 8. Nf1 $11 {now out of the database. My opponent decides not to exchange on f5 and protects the bishop in a very unusual way.} Bxd3 {the rule is normally to take the bishop on d3 if White does not exchange on f5.} 9. Qxd3 Rc8 {played after some thought. The primary idea is to tactically restrain White from castling queenside, for example with an annoying ...Nb4 follow-up.} (9... Bg7 {is perhaps more natural development.}) 10. Nf3 Bg7 {intentionally non-committal, playing something useful while waiting to see how White develops.} 11. Ng3 O-O (11... h5 $5 {I considered as a possibility, threatening to chase the Ng3, but rejected it as unnecessary.}) 12. O-O Re8 {another non-committal but useful move, placing the rook on a more active file and clearing f8 if it is needed in the future.} 13. Rfe1 {the first real think by my opponent, who ended up playing the obvious mirroring move with his rook.} Na5 {played after some thought. Black has few ways to make real progress, but it is still good to stay active on the queenside. The knight would like to go to c4, perhaps provoking a weakening advance of White's queenside pawns.} (13... e6 $5 {is a solid choice as well.}) 14. Nd2 Nc4 {this is not very patient, but I did not see other useful moves. The engine is more patient.} (14... e6 {locking down the e-file and protecting d5 with a pawn.}) (14... Qb6 {targeting the b2 pawn.}) (14... a6 {controlling b5 and supporting a potential b-pawn advance.}) 15. Nxc4 {this is White's best choice, exchanging off the comparatively better knight.} Rxc4 {the engine prefers capturing with the pawn, but I did not like the idea of giving up the e4 square.} 16. Nf1 {this is a useful redevelopment of the knight, but it does let me chase the bishop now.} Nh5 {the engine prefers to withdraw the rook to c6 or c8, getting it out of potential danger.} 17. Be5 f6 18. Bg3 Nxg3 19. Qxg3 Qc7 $2 {I thought this was clever and good to finally develop my queen on a useful digaonal, but I ignored the fact that the d5 pawn is now hanging.} (19... e5 $1 {is the best follow-up, since it is now possible. "No pawn break, no plan."} 20. dxe5 fxe5 21. Rad1 {during the game I did not like the prospect of having to defend the central pawn duo with pieces, but the engine considers it equal (and better than the game continuation). Black nicely controls the center and White has no real play.}) 20. Qh3 $6 (20. Qf3 $1 Qd7 21. Ne3 $18 {wins the d-pawn.}) 20... e5 $2 {if the d-pawn were not hanging, this would be pretty good, per the previous note. However...} (20... Rc6 $11) 21. Ne3 {this knight fork was an unfortunate surprise, but after the obvious rook retreat there is at least a bit of compensation for the pawn.} Rc6 22. Nxd5 $16 Qf7 23. Nb4 Rce6 (23... Rd6 $5 24. dxe5 fxe5 $14 {followed by doubling rooks on the d-file gives better compensation.}) 24. dxe5 $1 {my opponent correctly continues to exploit his advantage.} fxe5 {the engine prefers to recapture with the rook, but that would lead to a trade, diminishing my practical chances.} 25. Re4 $6 {my opponent's decision to split his rooks in the face of the enemy now gives me full compensation for the material, due to my superior major piece activity.} Rd8 {it wasn't clear to me which rook belonged first on the d-file.} (25... Rd6 $5) 26. Qe3 Red6 $11 27. h3 {evidently played to avoid back-rank issues.} a5 {settled on after some thought. Unfortunately I cannot quite trap the knight.} (27... Rd2 28. Qxa7 Rxb2 $11) 28. Nc2 {here my opponent offered a draw; the engine actually agrees with this, in the sense that the position is evaluated as complete equal. However, at the time I felt that I was better (incorrectly) and had the momentum (which was true).} Rd2 {again played after careful thought, although the idea of regaining the pawn is obvious.} 29. Qb6 $2 {going for a cheap trap, which however does not solve White's multiple problems. I got the feeling my opponent did not want to even temporarily go down material, which would have occurred with more solid play.} (29. Ne1 Rxb2 30. Nf3 Rxa2 $11) 29... R8d6 $1 $19 {this attack on the queen gains a tempo, so now White has two undefended pieces under simultaneous threat and cannot recover.} (29... Rxc2 $4 30. Qxd8+ $18) 30. Qc5 (30. Qb3 Qxb3 31. axb3 Rxc2 $19) 30... Rxc2 {I made sure to stay careful with my calculations. The Rd6 is tactically protected, due to my threat to take on f2 with the queen.} 31. Rf1 Rdd2 {by this point White is quite lost, although my opponent hurries the process.} 32. Qc8+ Bf8 33. Rxe5 Rxf2 34. Ree1 Rxg2+ 0-1

Bogus chess imagery: House of Staunton blog

Being on the House of Staunton chess store mailing list, I recently received a newsletter with links to two blog posts: "Master the Scotch Game" and "Chess Trolling: The Art of Playing Bad Openings to Tilt Your Opponent". They are curiously related, as I'll shortly demonstrate, by the included bogus chess imagery.

Here's the excerpted beginning of the Scotch Game post, with the accompanying image:

If I had to choose one chess opening for beginners, I might choose the Scotch Game. This confrontational opening not only embodies many classic opening principles, but also tends toward exciting tactics and strong middlegame plans that are fun and easy to understand.

But that isn’t to say the Scotch Game isn’t ideal for advanced chess players as well. Several lines boast some of the highest win rates of any opening on online platforms like lichess.org, and it has remained a strong offbeat choice among grandmasters for decades, too!

In this guide, you’ll learn the best way to play the main line, as well as some of the deadly gambits that lurk inside this ancient opening. 

Now, one might expect the (obviously AI-generated) accompanying pic to illustrate a Scotch Game position. But no! It's actually a Bongcloud opening, as can be seen by the White king located on e2. This seems intentional, given the companion "Chess Trolling" post, which includes a reference to the Bongcloud. The image creator also gets everything else visible about the board setup correct, which AI-generated images are notoriously bad about. Black in response has chosen an interesting f6-e5-d6 pawn wedge defense, so maybe both sides are trolling here (as well as the creator). In any case, it has nothing to do with the Scotch Game.

Terming the above a "bogus" chess image is perhaps a bit harsh, as the chessboard itself is legally set up. However, I still feel justified in using that term, since the AI-generated reflections on the glass table show both sides' pieces as black.

Annotated Game #358: A reverse Queen's Gambit

In this first-round tournament game, as White in a Colle System I face a common approach by Black, which is to go for a kind of reverse Queen's Gambit. A tempo down, it is not necessary challenging for White, but can be a little tricky. Here I opt for a quiet approach which is enough to get me equality with active pieces and a comfortable middlegame plan. While theoretically the Colle is not as challenging as the Queen's Gambit to face as Black, I have found that Black players can often go astray in seemingly harmless positions, which happens in this game after I am able to engineer the dominance of my dark-square bishop. Unfortunately I make some lesser follow-up choices and give away the advantage, although rise to the occasion in what turned out to be a more exciting major-piece endgame than I had expected.

[Event "?"] [Site "?"] [Date "????.??.??"] [Round "?"] [White "ChessAdmin"] [Black "Expert"] [Result "1/2-1/2"] [Annotator "ChessAdmin/Dragon 3.2"] [ECO "D04"] [PlyCount "95"] [GameId "2338422791188714"] 1. d4 d5 2. Nf3 Nf6 3. e3 c5 4. dxc5 Nc6 {with this move, Black indicates he is less concerned with recovering the pawn quickly, instead focusing on development in a reverse Queen's Gambit type position.} 5. a3 {controlling b4} (5. c3 $5 {is another way to do this.}) 5... e6 {I had expected ...e5, given my opponent's previous move, but he decided to play more solidly than dynamically in the center.} 6. Be2 {a quiet option, but White can press here.} (6. b4 a5 7. b5 Nb8 $14 {and White has a pleasant choice between the c4 break and Bb2.}) (6. c4 $5 dxc4 7. Qxd8+ Nxd8 8. Bxc4 Bxc5 9. b4 Be7 10. Bb2 $11 {with a comfortably equal position for White.}) 6... Bxc5 7. b4 Bd6 8. Bb2 O-O 9. O-O Bd7 $6 {this is a bit slow and clogs the d-file.} 10. Nbd2 (10. c4 {the text move is fine, but now would also be a good time for the c-pawn break, as ...dxc4 is impossible due to the hanging Bd6.} Qe7 11. cxd5 exd5 12. Nc3 $14 {with play against Black's isolated queen pawn.}) 10... Rc8 11. Rc1 Bb8 {this suprised me, but it does remove the bishop from the weak square. However, the c4 break is now clearly White's best option for making progress and challenging Black's center, so I play it.} 12. c4 Re8 13. Nb3 $1 $16 {I spent a lot of time here thinking about the middlegame plan. The knight's influence is maximized on b3.} Ne7 {opening the c-file for the rook and reinforcing the d5 strong point.} 14. Nc5 {a good deal of thought was spent here as well, validated by the engine.} (14. cxd5 $5 {is also possible.}) 14... b6 15. Na6 {another well-calculated decision. I could exchange off either bishop; here the point is to eliminate Black's dark-square bishop, so its White counterpart can dominate.} Bd6 16. c5 {the key follow up, continuing to chase the dark-square bishop.} bxc5 17. bxc5 Bb8 {my opponent spent a fair amount of time here. It is difficult to choose which retreat square to go to, but the engine considers c7 a better option.} 18. Nxb8 Rxb8 19. Be5 $1 $18 {completing the calculated sequence. The bishop is now a monster, its activity more than compensating for the c-pawn's isolation.} Ra8 20. Bd3 $6 {my first miscalculation. White's edge is considerably lessened now, as the initiative recedes. The light-square bishop needs to be activated, but there is a much better square.} (20. Ba6 {I looked at first, but it did not seem to work. The point would be to corral the black rook stuck in the corner.} Bc8 {I thought this essentially shut down the idea, but White does well after} 21. Bb5 {the two White bishops still dominate, in tandem with the advanced c-pawn.} Bd7 (21... Rf8 22. Nd4 $18) 22. c6 $1 {was the idea I missed. The tactical point being} Bc8 23. c7 $18 {trapping the queen.}) 20... Nc6 21. Bd6 $16 {some more significant thought here, to make sure this worked. White has to protect against the Bd6 being traded off.} e5 {now my opponent, although still worse off, generates some initiative with the central pawn duo. I need to move the Bd3 to avoid the pawn fork on e4, something I had missed initially.} 22. Bb1 {this is OK, but better options existed.} (22. Ba6) (22. Bb5) 22... Bg4 {activating the bishop. Now I thought that} 23. h3 {was simplest and best as a response - correct according to the engine. However I was not expecting the retreat} Bh5 $6 {and I did not calculate the critical variation correctly.} 24. Qa4 {second best, according to the engine. I thought that playing g2-g4 would lead to losing the Nf3 due to a pin, so searched for a tempo-gaining move with the queen. Hitting Black's hanging Nc6 fit the bill.} (24. g4 Nxg4 25. hxg4 Bxg4 {with the threat of ...e4, but now} 26. Qxd5 $1 $18 {and the queen breaks the pin and simultaneously protects the knight, backwards along the diagonal.}) 24... Qd7 {now I knew a piece should go to h4, but picked the wrong one.} 25. Qh4 $6 (25. Nh4 $1 {with the threat of Bf5, kicking the Black queen away from the defense of the Nc6. After} Bg6 26. Nxg6 hxg6 {I thought Black was fine, but the engine still shows a large plus for White after} 27. Bd3 $18 {renewing ideas of Ba6 or Bb5.}) 25... Bg6 26. Nd2 {here I played defensively, being most concerned about preventing ...Ne4 and an exchange on d6.} (26. Ba2 $1 {instead actively pressures the now-weak d5 pawn. If} Ne4 $2 (26... Rac8 27. Rfd1 $16) 27. Bxd5 Nxd6 28. cxd6 {and the pawn is tactically protected.} Rac8 (28... Qxd6 29. Bxc6 $18) 29. Rfd1 $18) 26... Bxb1 27. Rxb1 Ne7 $11 {somewhat ironically, now that my advantage has disappeared, I decide to "go for it" with} 28. g4 {clearly surprising my opponent. This is a bit too loosening of the kingside pawns, but the best alternative would be to exchange on e7 and reach sterile equality, which I was not interested in. The text move also prevents the threatened ...Nf5 fork.} Nc8 $6 (28... h6 $11) 29. g5 $1 {the other point of the g-pawn advance, as the Nf6 does not have any good squares left.} Ne4 30. Nxe4 dxe4 31. Rfd1 $1 {I thought for some time before choosing this more dynamic rook move (correctly) over the immediate capture on e4.} Nxd6 {unfortunately now I choose the wrong recapture on d6, with time pressure starting to be a factor.} 32. cxd6 $6 (32. Rxd6 {I erroneously thought that Black now would have time to concentrate his forces against the isolated c-pawn, but this is not the case.} Qc7 33. Rd5 Rac8 34. Qxe4 $18 {the Qc7 is awkwardly placed, preventing the c-pawn's capture.}) 32... Re6 $2 (32... Rad8 33. Rbc1 Qe6 34. Rc6 Qg6 $14) 33. Qg4 {this relied too much on a cheap tactic which I did not really think my opponent would fall into.} (33. Qxe4 $1 {now is the time to capture the pawn.} Rd8 (33... Rxd6 $4 34. Qxa8+) 34. Rb7 $1 $18) 33... Rd8 $11 {now the position is equal again.} (33... Rxd6 34. Qxd7 $18) 34. Qxe4 Rxd6 35. Rxd6 {forced} Qxd6 36. h4 {consolidating g5} h6 37. Kg2 {I felt I had to be careful to maintain the balance. This allows the rook to transfer to the g- or h-files in the future, to shore up my kingside or potentially threaten Black's king.} hxg5 38. hxg5 Qxa3 {snatching the pawn, which however lets me exploit the h-file with} 39. Rh1 Qd6 40. Qh7+ Kf8 41. Qh8+ Ke7 42. Qxg7 Qd5+ {now I had to make sure I kept safe from checks, so carefully calculated the next sequence.} 43. Kh2 Qf3 44. Qxe5+ {my opponent said he missed this, although other moves are equal as well. Now I can force a repetition in this drawn position.} Kd7 45. Qd4+ Ke7 46. Qe5+ Kf8 47. Qh8+ Ke7 48. Qe5+ 1/2-1/2