Caro-Kann Opposite Sides Castling

Today I played a 16+10 game with Caro Kann Classical Variation (I am black pieces). lichess.org/jZDKftPu/black

I'll do my best to explain my thought process on various moves, along with my interpretations of Stockfish's suggestions.

12.0-0-0: This position is similar to a game I played with my friend, Alexander (Chapter 1 of this study: lichess.org/study/FbebwbJV), however in this position, the light squared bishops have been traded. I believe that makes the position worse for black because 1. I don't have my attacking bishop on h7, and also white's bishop on c4 is gone, which won't give me the several tempi for my queenside attack headstart.
12...b5 Since the game with Alexander played out so well, I decide to go right ahead and start my queenside expedition. I also have the moves Nb6 and Bd6 in mind.

14...Nd5: I'm thinking I can win the bishop and be stronger.
15...Nxf4? I immediately decided to take the bishop, but I should have continued with my plan of Nb6! Now his queen gets more active.

19...I was trying to get more pieces lining up to his king, but afterwards I realized that I have no pawn breaks for several more moves! :[

20.g4 In contrast to the game I had with Alexander, white's kingside attack is very speedy.
20...Nf6? I realize my queenside attack is too slow, so I try to defend my kingside first. Stockfish disagrees with this move.

24...Rfd8?? I thought I had something with this move, taking advantage of the back rank. However I'm too late.

26...Be7?? Initially I thought this was a good move... But I completely forgot about white's other rook.

28...Qc7?? Finally I try to bring my queen back for defense, but it's too little, too late.

I believe that once I realized my queenside plan was too slow, I started reacting to my opponent and that caused me to tumble, as Jeremy Silman says! :/ I'm wondering if anyone could list a few pointers from this game. Thanks and Happy Checkmating!

This is a theoretical highway in the Caro-Kann. After 12.0-0-0, you have three main moves : 12...Nd5 (Dreev,...) and 12...Be7 (Anand, Bareev,...), all the rage in the early 2000's, and then 12...Qa5 (Jobava,...) which is more modern. 12...b5 is not a novelty, but it is an interesting move that has not been explored thoroughly.

Since you play 13...Be7 anyway after 13.Rhe1, I looked for a transposition 12...Be7 13.Rhe1 b5, but I couldn't find it. The moves themselves look reasonable and don't let my Komodo scream in horror, but they seem to be avoided by specialists.

Komodo's try for advantage after 12...b5 is 13.Kb1 Be7 14.Ne5 attacking the c6 pawn (left undefended by the b pawn as a consequence of b5). But Black can either dive into the melee with 14...Nd5 15.Nxc6 Nxf4 (you attack my queen, I attack yours) with a mess or play calmly with 14...Nxe5 15.Bxe5 Qd5, followed by Rd8. The calm line is a pity because Rd8 is a concession. When you play 12...Qa5, you have to play Rd8 after White's Ne5, to defend your Nd7. But if you play 12...b5, you hope to keep the big guns behind the pawn storm on the queenside.

My conclusion is that 12...b5 is an improvement on 12...Be7 only if the tactical melee 12...b5 13.Kb1 Be7 14.Ne5 Nd5 15.Nxc6 Nxf4 leads somewhere ; it is unclear for now. Perhaps you can consider 12...Be7 13.c4 b5, which is an approved theoretical line. If Black develops steadily (Be7, 0-0,...) then the queenside pawn storm will be much quicker than White's attempts on the kingside. Actually the threat of the pawn storm will prevent White from distracting pieces away from the center, and equality is reached.

In the game, White errs immediately with 14.Qe3?! which triggers 14...Nd5 15.Qd2 N7b6 16.Ne2 Nc4 and Black attacks ! If White doesn't play 16.Ne2, then 16...Nc4 will actually win the exposed bishop on f4, so it makes absolutely no sense to exchange the bishop early with 15...Nxf4. It deprives Black of his main threat. The sequel of the game also proves that 15...Nxf4 16.Qxf4 gives White a momentum for his own attack, which is why Black should seek an exchange of queens with 20...Qb8. Komodo even recommends a slight improvement with 19...b4 20.g4 Qb8.

After 20.g4 Nf6??, Black does not make the most of the opportunities provided by White's mistakes (22.Rxe4?? and 25.Nf3??) because his logic grasp on the game has been perturbed. Mistakes and lost opportunities do not appear at random, in general.

It is a very interesting game with a nice critical phase (between 12...b5 and 20...Nf6??). Whomever makes sense of 12...b5 13.Kb1 Be7 14.Ne5 Nd5 15.Nxc6 Nxf4 will take the real credit for the 12...b5 move, which has yet to show its full potential.

If you plan on opposite side castling, then the maneuver 11...Qa5+ 12. Bd2 Bb4 13. c3 Be7 14. c4 Qc7 is necessary to get enough counter play from the move b5. The difference between this variation and when you played 12...b5 in your game was that there was no pawn on c4, thus white could just ignore 12...b5.

Good advice above. As white has no weakness your attack was doomed to fail.

@A-Cielbleu @ace1886 @tpr So you're saying, if white has a pawn on c3 or c4/bishop on c4, then that is an easy target and the attack should be fine. But if white doesn't have such a target, either try to create a target, or castle long as well and play in the center?

You should always play for the centre, regardless of whether you are attacking or defending. As to castling long or short, both are feasible in the Caro-Kann. Point is you tried to attack with ...b5, and later you realised your attack was leading nowhere and you started re-routing pieces for defence. Before attacking you should provoke weaknesses by play in the centre. See #3 above: ...Qa5, ...Bb4 provoke c3 and c4. Also the profound lines in #2 revolve about play in the centre.

That's right as White has a development lead and a better control of the center. Gaining tempo's during the attack is a way to catch up. This position was debated among top GM's, so don't expect to find a set of rules that lead to good decisions. But it is instructive ; if you keep playing it, you will accumulate valuable experience.

As a whole the Caro-Kann is a good opening for players with a good sense of danger (Karpov). You see threats coming from afar.

It is an erroneous thought that opposite castling necessitates opposite attacks on the kings. Here is an expert example, where white castles long and black castles short, but both grandmasters play in the centre. White does not win in an attack, but in the endgame.

@tpr #6 @A-Cielbleu #7 The Master Database plays the line 1 e4 c6 2 d4 d5 3 Nc3 dxe 4 Nxe Bf5 5 Ng3 Bg6 6 Nf3 Nd7 7 h4 h6 8 h5 Bh7 9 Bd3 BxB 10 QxB e6 11 Bf4 (Indication white is probably going to castle long) Qa5+ 12 Bd2 Bb4 13 c3 Be7 14 c4 Qc7 15 0-0-0 Ngf6 16 Kb1. The most popular move is 16...0-0 but is it the best way for black to play? Does black have a possible idea of castling long? What are the factors that should guide black on how to play? Black's attack hasn't started yet but if it's chosen, will it be effective? When should black trade queens based on the castling and attack speeds? Sorry if I'm asking too many questions...

@tpr #8 The queens were exchanged really quickly. Doesn't that make the position calmer and less threatening for the kings, therefore the other pieces need to be improved generally playing in the center?

@FunnyAnimatorJimTV Black shouldn't castle queenside if they have provoked the c pawn to c4 with the maneuver Qa5+ and Bb4. If you want to castle queenside, play 11...Qa5+, 12. Bd2, Qc7. When black castles queenside, the pawn on c4 is a pro for white, as b5 is taken out of consideration, and c5 comes with the weakening of black's king, whereas when black castles kingside, the c4 pawn is mostly a weakness. On a side note, the lines where black castles queenside have fallen significantly out of fashion, as white gets a comfortable game with better attacking chances.