The Best Game I've Ever Played - Fully Annotated!

Edit - If you have any questions, please feel free to ask... I didn't include all possible variations. 

 Anyone who likes attacking whenever possible, even as black should try it.

It tends to be best suited to those who are comfortable using their queen extensively however the best dutch players understand the intricacies of pawn movement (this game is a living example of that!) 

Good game. A few comments though:

When playing  

14... Bg4, Did you consider 15.Nxe5 dxe5 16.d6         ???

black just doesn't have sufficient compensation there for the material deficit he's about to go into.

The first piece sacrifice is suspicious looking for one simple reason:

After 18.h3 (h4 is really strange), if you retreat the knight you're just a piece down, and if 18...Nf2 19.h4! NOW (see the difference from before?) I'd say white should be winning after this.

Even after 18. h4? Bxh4! Nf1 (actually best in my opinion), the position is unclear.

19...Rxf1?! I think this is just bad if white takes with the bishop. I'd love to hear how you were going to keep the attack going in a way which would justify all the material deficit. The position still looks complicated, but I think white should come out on top here. I'd happily take white in a game after Bxf1.

White obliges with Rxf1, and now black's attack becomes really dangerous. Even if there is any way to save this (I no longer think there should be a way), it looks bad.

 Nice attacking game. Even if it looks a bit overcharged with aggression at some points (I analyzed it for a bit. I believe the initial piece sacrifice should have made black the proud owner of a lost position, had white defended slightly better (h3 instead of h4)).

 Very nice analysis.  I've run the game through fritz at this point and it seems you are almost 100% spot on.  However, I do tend to disagree with fritz in a few issues...

 Let me disect your analysis a little.  The following is a mixture of fritz and my own analysis, it could be wrong but it seems sound.

When playing  

14... Bg4, Did you consider 15.Nxe5 dxe5 16.d6         ???

black just doesn't have sufficient compensation there for the material deficit he's about to go into.

 I did notice this possibility when I made the move but I felt I would be able to produce a good game after 14. ... Bg4 15. Nxe5 dxe5 16. d6 Qf7! which looks only slightly advantageous for white and black is still on the attack (which was my goal anyways).  Your line does, however, seem at least as good if not better.

The first piece sacrifice is suspicious looking for one simple reason:

After 18.h3 (h4 is really strange), if you retreat the knight you're just a piece down, and if 18...Nf2 19.h4! NOW (see the difference from before?) I'd say white should be winning after this.

 This is a phenomenal and rather difficult discovery.  Well done, I believe you are correct.  I missed this. The key to this game was the attack and I think white would have at least taken SOME tempo back which probably would have been enough to win (although, of course, the position couldn't be called for a certainty, especially with a slightly vulnerable white king and my queen on the board). 

I would probably play something like this... (Fritz matches relatively closely with my personal analysis).  18. h3 Nf2 19. h4! g5 (this being the key move) and I feel black can continue it's attack... the key being that a bishop sacrifice may still be possible... 

This was the best (I would argue only) saving continuation for white.  Nice find. 

Even after 18. h4? Bxh4! Nf1 (actually best in my opinion), the position is unclear. Fair.

19...Rxf1?! I think this is just bad if white takes with the bishop. I'd love to hear how you were going to keep the attack going in a way which would justify all the material deficit. The position still looks complicated, but I think white should come out on top here. I'd happily take white in a game after Bxf1. Yes, again... I believe you are right.  Rxf1 may have been wrong.  However both 20. Kxf1 and Rxf1 seem to lose, and are very complicated, thereby forcing a very difficult defense for white.  It should also be noted that Bxf1 STILL gives black an attack after : 20. Bxf1! Be7 [Bg5 also being an option but simplifying too quickly] 21. Re2 Rf8! 22. Bd2! [Again, very difficult to find] 22. ... e4  23. Qc2, Ne5 and at this point fritz is giving white a very small advantage with the past several moves (and a couple more to come) 'forced' moves that are very difficult to see.  This sacrifice was mostly intuitive on my part but I am nearly certain it is sound.

White obliges with Rxf1, and now black's attack becomes really dangerous. Even if there is any way to save this (I no longer think there should be a way), it looks bad. Correct.  White was slightly innacurate but his game was lost anyways.

 Nice attacking game. Even if it looks a bit overcharged with aggression at some points (I analyzed it for a bit. I believe the initial piece sacrifice should have made black the proud owner of a lost position, had white defended slightly better (h3 instead of h4)).

Thankyou very much... I appreciate the analysis.  Would I play this sacrifice again, maybe not against the same person but in general, absolutely.  The second and third sacrifices are not fully clear but I do believe to be sound (fritz as black beats fritz as white :P).

The first sacrifice was dubious but I do get an interesting game! 

Thankyou so much for your analysis.  I hope mine makes as much sense :P. 

From my experience with the top Chess engines, Fritz is probably one of the worst of these.

Also, fritz is very bad in analyzing many types of sacrifices where the attack/refutation is not immediately visible in his search depth. Any of Toga, Rybka,Hiarcs are better (actually Shredder and Junior are slightly better than fritz at these kins of things too).

In those positions, if fritz doesn't find a win for either side, his evaluation can't be trusted.

More concretely:

> I did notice this possibility when I made the move but I felt I would be able to produce a good game after 14. ... Bg4 15. Nxe5 dxe5 16. d6 Qf7! which looks only slightly advantageous for white and black is still on the attack (which was my goal anyways).  What white did looks better to me.

Be careful of making concrete decisions on superficial grounds, which is exactly what is happening to you here. You seem to have a mentality which says: if I'm attacking, things are ok. You can get yourself into many lost positions (for attack) like that. This will really block your Chess improvement. Attacking is nice,

but things should always be put into a more complete context of: "what am I getting, and what am I paying for it?", if you pay more than you're getting, you'll lose more often. 

More to the point:  16...Qf7 17.Be3 and only white can have any chances to win here.

Black's best may be 17...Bf3 18.Bxf3 Qxf3 19.dxe7 Re8 20.c5+ Kh8 21.Red1. This is horrible for black. Also, I can't find anything other than 17...Bf3 for black.

After what white played, the advantage is either with black, or it's equal. So what white played is definitely inferior to Nxe5.

I'll post some more analysis on the other lines, you seem to be missing a few important things in your general approach in my opinion.

In any case, letting fritz play against itself is pointless here. From my experience with computer chess, Fritz's evalation, in my opinion, is close to garbage in these complicated positions. 

As to your suggested 17...Bxd6 18.Bxa8 c6 (forced, else the bishop gets back to the defense and black has done nothing) 19. c5 (I want to exchange queens or drive the bishop back for one move) 19...Be6 20.Qc2 Bxc5

(Black has no time for 20...Bh3 because after 21.Qe4!, and not only will black never break through, but also white is already thinking about saving the a8 bishop, which would be a catastrophe, so black has to respond. white has a near decisive (possibly already decisive) advantage)

21.Bxc5 bxc5 22.Rad1 Bxa2 (Black's prtoblem is that white never needs to respond to 22...Bh3 because Qf3 can always be answered by Qe4, his pawn structure is terrible, and again I think I should already be able to win this easily as white).

 After this (Bxa2) 23.Rd2 Bd5, white should be at least better. I'm already thinking about Rxe5 and finding the right moment to give the exchange on that bishop. Black's pawn structure is so bad that he may be lost if he loses all attacking chances once and for all.  After all, that's all black ever had.

Nxe5 just leads to trouble for black 

Interesting.

I do outrightly disagree with you regarding one issue...

Fritz.

Despite being an older engine, Fritz 5.32 is STILL considered one of the better tactical engines and I've seen it beat Shredder (what version, I have no clue).

Nxe5, yes, you've won me over... although white has to play rather well, it DOES lead to advantage for white.  Nice find, I was too hasty in my analysis.

The trio of sacrifices, however, still seems like a good idea... with the first one being dubious but not refuted, and the last two being sharp if not sound.

I think that's a fair conclusion.

Thanks again for all the help!

-Matt 

Yeah this game had a lot of 'levels'.

I played and won this game and I'm still figuring my moves out :P.

By the way, send me another game over some time eh? 

> It should also be noted that Bxf1 STILL gives black an attack after : 20. Bxf1! Be7 [Bg5 also being an option but simplifying too quickly] 21. Re2 Rf8! 22. Bd2! [Again, very difficult to find] 22. ... e4  23. Qc2, Ne5 and at this point fritz is giving white a very small advantage with the past several moves (and a couple more to come) 'forced' moves that are very difficult to see.  This sacrifice was mostly intuitive on my part but I am nearly certain it is sound.

First of all, nice going with this line. Didn't work that one out before. However, even after this rather impressive display of power, 24.Bf4 ends the attac: 24...Nf3+ (forced, no?) 25. Kf2 g5 (otherwise there is no attack) 26.Rxe4 (and if 26...gxf4 27.Rxf4 is cute and totally unclear...(IF black finds he has to play Bf6)).

And if not 26...gxf4, then what else? 26...Ne5 27.Qe2, still, I'd say it's still unclear

As for being certain of the soundness of the sacrifice, even GMs are not certain of their intuitive sacrifices until truly extensive analysis is performed. That's really the only way to even be remotely certain of it. You have a tendency to stop too often and say: "with attack...". That is a good approach if you're writing an opening theory book, not when you're actually asking about the soundness of a sacrifice.

Now rewind a bit to 18.h4? 

You say that 18.h3 Nf2 19.h4 may be white's  "only saving move". The problem is,

 this move is just winning for white. The move you gave, 19...g5, gives black some attack. Unfortunately, he can never gain as much as he has given here.

Black is attacking, but the position is not even unclear anymore. It's just an easy win for white. There are at least three winning moves here. I've analyzed 20.Rf1, 20.Bf3, and 20.Nf3. There's nothing there. A few threats for a few turns... Not really much of an attack.

(Nf3 is my personal favourite). 

You say you'd play this sacrifice again. Ask yourself this question: Do you think this sacrifice would work on stronger players? if no, then by playing these things, you are making sure you're holding back your own development and improvement.

If you think it would work on stronger players, you need to check and double check that it is sound at all. Otherwise you'll find yourself in positions where you are attacking... And completely and utterly lost. Once again blocking your own progress.

 Yeah it has, it's pretty tricky.  I think grolich may have a few points but he's neglecting to consider that the 'right' moves for white were SUPER hard to find and didn't really give white much better a game anyways!

Thanks for pointing that out about the bishop... I knew the white queen was pretty inactive but I hadn't thought of that rook sitting over there...

And finally, so as to avoid argument, the Dutch is a great tool but it's best used by attackers, sit back and wait and you'll be choked.  In fact... I'll post a game where I played a Correspondence International Master and was crushed... 

edit - nevermind, can't find it.. it was more than a month ago :( 

Just read your last post grolich, we must have been cross posting...

very nice, very interesting.  and yes, I'd play this again .. Tal beat GM's with unsound sacrifices and I think this one is borderline sound :P.

Nxe5 is the ONLY thing that bothers me and I can't find any way out of saying advantage white.

the other sacrifices are too interesting and too complex (unclear if you like that word) to say for sure. 

edit- however I will say one thing, my original annotation is correct.. the trio has one dubious, one sound, and one sharp (unclear).

edit #2 - I've sent you a challenge, I think we could learn a lot from eachother.. we have very different approaches to this game :).  hope 3 days is alright?

>Fritz.

>Despite being an older engine, Fritz 5.32 is STILL considered one of the better >tactical engines and I've seen it beat Shredder (what version, I have no clue).

As someone who is deeply interested (and involved in what is current) in the field of Computer Chess (I'm a computer scientist), I feel like I should dispell a few illusions you seem to have:

1)

Diffrentiate between "I've SEEN it beat (it CAN beat them)", and "it usually beats" (which isn't true). 

 It's not that it CAN'T. That's not even an issue. It's that it will lose the vast majority of the games.

That's a fact proven by thousands of games run by many people throughout the world:

fritz 5.32 will lose ~70% of the games to modern versions of shredder, ~80% to modern versions of hiarcs, and ~85%-90% to the latest version of Rybka (I've run many of those tests. Also, there are numerous people running countless of these tests and publishing them on the net. Their results and mine tend to be roughly level).

The fact that it is not THAT bad that it can never win at all does not add much to it.

2) Another thing is. its tactical ability has been compared with modern engines and it is just worse EVEN in that field. 

   Modern searches are more efficient, in many ways. There are many reasons for that, but one of the main reasons is that Modern engines know a lot better than fritz 5.32 which lines should be selected for deeper analysis, resulting in much more accurate results in most compicated positions.

In the days of 5.32, there wasn't much selection used at all (other than "quiescence search", which is usually not considered together with "selection", although it is a kind of selction, in that it does searches certain lines more deeply.)

The end result of this and other advances is that in the vast majority of positions, modern leading engines (Btw, even the latest version of Fritz is not really considered among them anymore) will tactically see a LOT further than Fritz 5.32 where it counts. There are positions where this will not be true, but you should prefer the one that is better, not the one that can do better in a small percentage of cases.

 3)

The most important thing is this: fritz 5.32 is horrible as an analysis helper because its evaluation is horrible. It does not understand the effects of nothing that is out of its tactical range, its handling of complex long term sacrifices is its weakest point. It does not know how to defend those positions at all. even I managed to beat him from similar positions where I was lost (As was proven later by other players and other engines. And I'm a relatively weak player... compared with any engine).

Modern engines have had their evaluation function greatly improved. That affects the value the position gets at the end of every search line. In fritz 5.32, this value is not very meaningful in complex positions where concrete tactical results are not within fritz 5.32's search depth.  

(Not only is it slightly weaker tactically than most modern engines, you don't even NEED the tactical capability alone. You need an evaluation that you can count on. I've reversed that engine's eval many times before, despite the fact that it is stronger than me by far)

If that is what you were using for analysis to this day, rest assured, you've been missing a lot of important lines it just didn't understand, and you received a lot of incorrect evaluations.