Annotated Game #326: Going against principles, and a lesson at least partially learned

In this tournament game I made a couple of key moves on general "principles" which in fact went against the actual principles inherent in this Advance Caro-Kann setup, most notably not exchanging off White's key knight on f3 when prompted. I also miss several saving/winning resources from my opponent, including the final march of the kingside pawns in the endgame. I will nonetheless give myself at least partial credit for the pawn sac on move 22 after castling (!) - should have done so earlier, naturally, but it was good to see the engine validate my choice to prioritize piece activity over material.

[Event "?"] [Site "?"] [Date "????.??.??"] [Round "?"] [White "Class B"] [Black "ChessAdmin"] [Result "1-0"] [Annotator "ChessAdmin/Dragon 3.2"] [ECO "B12"] [WhiteElo "?"] [BlackElo "?"] [WhiteFideId "-1"] [WhiteFideId "-1"] [PlyCount "97"] [GameId "2248621732107422"] 1. e4 c6 2. d4 d5 3. e5 c5 4. c3 Nc6 5. Nf3 Bg4 6. Be2 e6 7. h3 $6 {while not completely losing, this has a relatively simple refutation, which I did not bother to calculate, just moving on general principles.} Bh5 $11 {Black is equal here, but no more.} (7... Bxf3 $1 {this is also correct based on the positional structure, in which Black pressures d4.} 8. Bxf3 $15 {and Black has a pleasant choice between ...Qb6 and ...cxd4 for a lasting edge.}) 8. Nbd2 Qb6 (8... cxd4 $5 {is a simpler approach.}) 9. dxc5 Bxc5 10. O-O Qc7 {long think here. It is fine after the text move, but White missed a more pointed option.} (10... a5 {would restrain b2-b4.}) 11. g4 (11. b4 {now the minor pieces are pushed backwards or towards the rim.} Be7 12. b5 Na5 13. Qa4 b6 14. Bb2 $16) 11... Bg6 $11 12. Nb3 {this gets in the way of the b-pawn advance, so is helpful for me.} Bb6 13. Bf4 Nge7 (13... h5 {immediately is more impactful, also deferring the knight placement in the event another square is useful. For example after} 14. Nfd4 (14. g5 Nge7 $17) 14... hxg4 15. hxg4 Nxe5 {the knight may be able to go to f6.}) 14. Bg3 h5 $15 (14... O-O-O $5 $15 {is an interesting idea I also contemplated, giving more freedom to a kingside attack.}) 15. g5 $6 {avoiding the exchange on g4, which would open the h-file, but now my knight gets an excellent outpost.} Nf5 $17 16. Bh2 {unfortunately after the bishop retreat I ran out of good ideas.} h4 $6 {there is no target for the h-pawn, so this simply wastes time and makes it potentially more vulnerable. The general principle of advancing pawns towards your opponent's king as a threat does not apply here.} (16... O-O $19 {gets the king out of the center and activates the h8 rook. The engine evaluates Black as positionally winning, at this point.}) 17. a4 a6 {a prophylactic move, guarding against a potential future a-pawn advance.} (17... Rd8 {increasing piece activity is better, since the a-pawn cannot advance further at the moment.}) (17... a5 $5 {with a similar idea, but preventing further advance of the a-pawn, which becomes important shortly.}) 18. Nbd4 {White lacks any good moves to make progress, but this has a tactical flaw.} Qe7 $6 {choosing to pressure the g5 pawn, instead of exchanging off the knight.} (18... Bxd4 $1 19. Nxd4 (19. cxd4 Qb6 {is similar}) 19... Nfxd4 20. cxd4 Qb6 21. Bf4 Nxd4 22. Be3 Nxe2+ 23. Qxe2 d4 {I did not see this last move in calculating the sequence, which the engine evaluates as much better for Black.} 24. a5 Qc5 $19) 19. Nxc6 bxc6 20. Bf4 Qa7 $6 {this supposedly clever move, threatening ...Ng3 while establishing a Q+B battery, is neutralized by White's next (which I did not see.)} (20... a5 $17 {would secure the Bb6.}) (20... Ng3 {is better than the move played, but immediately also runs into} 21. a5 $1 Ba7 22. Bxa6 Nxf1 23. Bxf1 {with complicated play and partial compensation for the exchange.}) 21. a5 $1 $11 Bc7 {I debated between this and the d8 square, which would have been a little better.} 22. Qa4 O-O {a pawn sac approved by the engine. This was based on getting the rook in the fight, which was the right thing to do, albeit a bit late.} 23. Qxc6 Ng3 {I finally get this in.} 24. Bxg3 $6 {I had looked at this, but thought simply moving the rook away would be better.} (24. Rfe1 $11) 24... hxg3 25. Bxa6 {a somewhat tricky move, although Black has several good responses.} Qxa6 {the best one, although after} 26. Qxc7 Qe2 $6 {I place the queen on the wrong square. I debated between this and the correct one (d3), with the clock counting down contributing to the failure of visualization/calculation. Both moves target the hanging Nf3.} (26... Qd3 $1 27. fxg3 Be4 {with a double attack on the pinned knight, and now if for example} 28. Rad1 (28. Kg2 Qe2+ 29. Rf2 Bxf3+ $1 $19) 28... Qe3+ $1 {and the knight cannot be protected successfully.}) 27. fxg3 Qxb2 $11 {the correct bail-out decision in this line.} (27... Be4 $2 28. Rf2 $18 {and Black's attack is gone.}) 28. Qc5 Bd3 (28... Rfc8 $5 {immediately gets the rook in the game, with tempo.} 29. Qb4 Qe2 $11) 29. Rfe1 Rfc8 $6 (29... Be4 $5) (29... Rfb8 $11) 30. Qb4 {I saw this was the likely response, but I mis-evaluated the resulting endgame.} Qxc3 31. Qxc3 Rxc3 32. Re3 $1 $16 {I missed this idea, otherwise Black has it easy.} Rac8 33. Nd4 $6 (33. g6 {is the engine's idea, forcing a weakening of Black's pawn structure.} Bxg6 $4 {loses because the a-pawn advances after} 34. Rxc3 Rxc3 35. a6 Rc8 36. a7 Ra8 37. Nd4 {followed by Nb5-c7}) 33... Rc1+ {a correct decision made under more serious time pressure.} 34. Re1 Rxe1+ 35. Rxe1 Rc5 $2 {attacking the wrong piece, and the losing move.} (35... Rc4 {I saw this earlier in my calculations but instead focused more blindly on directly attacking the pawn.}) 36. Ra1 $18 Ba6 37. Kf2 $6 (37. g6 fxg6 38. Nxe6 $18) 37... Kf8 $2 {the wrong idea. The king is needed to cover white's pawn majority on the kingside, which is dangerous even if fractured.} (37... Rc4 $1 {again is the answer, forcing a concession from White, either moving away the knight and letting the active rook onto the 2nd rank, or} 38. Rd1 Kh7 $11 {with ...Ra4 to follow.}) 38. Ra3 (38. g6 {this undermining idea is still very effective.}) 38... Ke7 $2 $18 {now the king is too far from the future pawn action on the g/h files.} (38... Rc4 39. Ke3 g6 $14) 39. Ke3 Kd7 40. Kd2 Rc4 {too late with this idea, if my opponent plays correctly.} 41. Nf3 $2 (41. Nc2 $18) (41. Ke3 $18) 41... Kc6 $2 {after a long think, I still ignore potential kingside danger.} (41... g6 $11 {and the king can come back and defend as needed.}) 42. Rc3 $1 {now there is no return, with the rooks forced off.} Kb5 43. Rxc4 Kxc4 44. h4 {only now did I realize the real extent of the problem.} g6 45. g4 d4 46. h5 gxh5 47. gxh5 Bb7 48. h6 {the knight can be left to its fate.} Be4 49. a6 1-0