Bishop sacrifice in weird Ponz thing

Here is a game I just played. I was Black. I won. (but I don't mean to showcase, I'm actually asking questions about this game.) It's Blitz. I've already looked at it with Houdini... so ok I haven't followed advice on this forum before posting, but I think broke the computer analysis, somehow, it's anotating with nonsense or I'm not reading it right, so I am asking for a bit of help if you have the time.

Thanks for your time.

hmmm...

I get that my Queen is trapped after 14. g3 so I guess the 13... Qh4?? annotation makes sense. (despite the computer saying it's better for White to trade Queens after that than to take mine,) but does that mean my Whole Bishop sac was terrible? or was there a better move? Did I have to play 13... Qf5? (It seems like the only move.)

OK, back from dinner and ready to rock. I'm guessing your main queries are about the B sac, and the computer lines.

After 7...Bb4 (all the moves and analysis before that look good), we have a dynamic situation that's good to take stock. White has a 1:0 central majority, but it's an IQP. His knights are pinned by your bishops, but that's temporary. Black's plan speaks for itself: make use of the current situation to develop fast and exert some real central control/piece play. Exchange your bishops for knights to gain time and central presence.

Thus, the computer's 9...Bxf3 can be more readily understood. Black should always be on the lookout for chances to trade B for N in exchange for the centre. The Nc3 can be chopped to maintain the queen on d5 and the Nf3 killed to hit d4; the computer's using the latter, combined with a threat on c3 to seize equality. Point being black needs fast piece play, or white's d4 pawn will assert itself eventually.

10...Qa5 with Scandinavian play is also my choice. The bishops on g4 and b4 are more suited to central play than a kingside attack, especially the one on b4. The lines after 11. h3 also intuitively look good for white... although computer analysis shows odd-looking things lurking in the shadows.

12...Bxh3!? doesn't look like it goes anywhere for black... but white's defence is hard to see. One thing about computers carrying out tactical attacks on either side is that they'll make quiet/nonsensical-looking moves to make the whole thing watertight. My guess is 13. gxh3 Qxh3 14. Bd3 prevents 14...Ng4 via 15. Bf5+. White's defence is practically hard if he takes the bishop, but it may be objectively defensible.

14. g3 Qxd4 15. Bxd4 Bxe5 16. Re1 (16. Bxe5 Rxd1 17. Rfxd1 [17. Bxf6 and after some simplifications black's up a pawn or two] Nxe5 and black has some comp) Bxd4 and black's N+B+P for Q, but with strong activity and threatens ...Bxc3. There is also some kingside pressure and ideas of maybe ...Ne5-g4 against f2. White will be forced on the defensive for quite some time, and as a biased personal belief, I think 2 minors + strong activity/bind = Q usually.

Thus, 14. g3 Qxd4 15. Nxc6 removes the threat of ...Bxe5 and really threatens Bxd4 with no compensation. So black is forced into 15...Qxd1 16. Nxa7+ zwischenzug, where white emerges up a clear piece while black has little for it. To sum it up, 15. Bxd4 wins a Q for B+N, but allows black lots of compensating activity, while 15. Nxc6 nets white a clear piece without permitting counterplay. Which choice is better? I'd rather the latter too, although it's hard to judge in the midst of battle.

A shame you didn't take 14...Bxe5 with clear advantage.

18...g6 19. Bc2 g5 gains a tempo by chasing away a defender. Or 18...g6 19. Bc2 Ng4!? intending to send a minor to h2.

The computer rates 21. Rxf3 as a blunder, but after the line it gives, black still retains a clear edge after 21. Qxf3 Ng4 22. Ne2 Rhg8 with many many threats. One thing with computers is that they'll take the best line as the position's evaluation, even if it's the only safe tightrope over the pits of hell. Silicon cares not for such things as "practical chances".

I'm not all that strong and don't use engines even to analyse, so my opinion on the computer analysis of 14. g3 could be wrong. Still, it stands to reason that piece activity can compensate for material loss. Try giving this entertaining opening a spin on Houdini: 1. e4 c5 2. f4 d5 3. Nf3 dxe4 4. Ng5 Nf6 5. Bc4 Bg4 6. Qxg4!? I assure you, even a queen up wouldn't be reassurance to human players as black, and the computer should agree.

Thanks for your comment Remellion. I was mostly just annoyed at Houdini for it's weird choices of when and when not to suggest sidelines. It doesn't say my Bishop sacrifice is bad, but after my Bishop sacrifice in the line of the game it gives one of my moves ?? and the only other line it suggests instantly loses my Queen. I think it's showing that move to show the tactic of the pin but I think it's just a mistake it doesn't show an alternative there which is good. If I had it look at the position and not analyze the whole game it'd probably be better. I didn't, at first, see my Queen was pinned a move later, and once I did all the rest of Houdini's analysis with "desparado" Queen moves made much more sense.

Your line is interesting. FYI, the computer doesn't think Black is winning but does think Black is better after:

-1.02/21 in the Kc6 line. Only -.30, more or less equality after the other (less accurate?) King move. However in a position with that material imbalance I don't have much faith in the computer lines. Black is obviously not out of the woods yet at the end of Houdini's lines and it's quite possible there's something better that's been figured out.

I admit I am stumped by the computer's 13...Nxe5 14. Bxh5. Probably it thinks black's losing at that point and picks something strange, since even 13...Qf5 14. gxh3 Qxh3 15. Bg4+ Nxg4 16. Qxg4+ Qxg4 17. Nxg4 leaves white a piece up. Strange stuff.

As for the move 6 queen sac, I believe 9...Qe8 10. Ne6 is fairly strong. Google "Bryntse Gambit" (after Arne Bryntse, and played later by Dana Mackenzie) for more on this ridiculous-looking but playable (!?) opening. It's also an excellent example of a line where computers can't really grasp the truth of the position and their evaluations cannot quite be trusted; I've a feeling white is slightly better out of the opening.

Yeah, I don't know... the computer does 2 things when it adds sidelines, though. It either shows improvements or it shows bad moves and why they're bad. I think it's line 13... Nxe5 is showing a bad option though annotating with "annalates a defender" is hardly the best "reason" why Black shouldn't play that way. If Black is already losing, though, I would have liked to know where I went wrong. If it was on my Bishop sacrifice then Houdini should have given that a question earlier. Usually it is more insightful. I think my odd game did kind of "break" the analysis function but only in that small way. The rest of it makes much more sense if you think about it.

Hm... that is a very intersting line that could eat up a lot of my time. For now I think I'll let it be. I don't play the Sicilian hardly at all anymore and I always thought 2. f4 as White was an innacuracy... it seems like Black has enough other options before the Queen sac that even if I studied it with the intent to play as White I'd end up in a lot of other odd f4 Sicilians.

I created an interesting Bishop sac on the other side of the 2. f4 sicilian several years back, btw, and I've played it three or four times in Blitz. White has to play very badly to achieve it but it happens.

It's obviously not the best move but I think it is sound and I have a 100% score with it... more because the people who play into it are not strong than because it's actually a good idea. Black will get either three pawns for the piece or two pawns and enough compensation. Maybe I'll punch it into Houdini and get an updated computer opinion... I think it was Fritz I used to check it way back when I was figuring it out and that engine is not as strong.

Black's only main deviations are from 2...d5 (only people who know this supposedly equalises will play it; or 1. f4 d5 2. Nf3 c5!? is rare too) 4...Nf6 (4...Bf5 5. g4!? is my line here) and 5...Bg4 (5...e6 6. Nc3 Nc6 7. Ncxe4 and I think white is doing OK.) The resulting positions are quite comfortable for either side. In practice though, nobody allows the queen sac, mostly choosing 5...e6.

And that bishop sac is very interesting. First impression of accepting is 5. gxf5 Qh4+ 6. Ke2 Qxf4 7. Nf3 Qxf5 8. d4 and it's fairly OK, white's king is odd but only black's queen is out. Dynamically equal perhaps. Declining by the foolhardy 5. h4 Be4 6. Nf3 d4 where white's probably not good at all, so white's pretty much forced to accept the sac, which may or may not be nice for black.

nice