29 May 2017

How to think for yourself in opening preparation



One of the "secrets" of advancing towards mastery in chess, as in many other disciplines, is that the more advanced you get, the more you should be thinking for yourself in order to make real progress, not just uncritically following other people's recommendations.  In chess, I would say that this process of independent thinking should start relatively early on, once you get past the initial technical subjects to be learned such as mates, identifying and constructing tactics, fundamental opening principles, and basic endgames.  All phases of the game require independent thought, evaluation and judgment.  At the most basic level, chess players need to answer for themselves the question why they are making each move.

The opening in particular is subject to a near-overwhelming amount of advice and provision of expert information, in the form of instructional material (books/DVDs/etc.), computer engine evaluations, and database statistics.  It's certainly a good idea to take advantage of others' expert preparation and work, although it's debatable as to how much effort an improving player should put proportionally on opening preparation, versus middlegame and endgame skills.

Regardless of the total amount of effort spent on opening selection, evolving your repertoire, practice and understanding holistic concepts, I believe it's important to underline the benefits of doing serious evaluations of your opening lines (and finding new ones when necessary).  This level of mental engagement will not only serve to strengthen your overall repertoire, it will - perhaps even more importantly - boost your recall and effectiveness when playing the openings in question, as you are regularly and actively evaluating different lines and their resulting positions.

This type of active management of your openings is easily implemented using a simple database structure, which can be updated whenever you run across related material.  A recent personal example of doing this on a systematic level was comparing the recommendations in Play the Caro-Kann (made easier by its e-book format) with my repertoire database and evaluating the author's recommendations.  By no means did I accept all of her ideas, but studying the differences and determining why I preferred one line over another (or perhaps an entire variation) was valuable in itself.  This process is also quite useful when going over individual master-level annotated games that you come across, as in the Gormally example below.

Computer tools can be quite valuable for your preparation, but also misunderstood or misused.

Database programs easily display for you the most popular and highest-scoring lines, but their statistics can be misleading in various ways.  On the positive side, databases can identify shifts in popularity of particular lines and you can relatively easily pick out the important games that cause them.  On the other hand, sometimes there is a quick shift away from using a particular line that means the "old" (and maybe busted) version still has a relatively high percentage result, one of the reasons why you have to evaluate lines for yourself.
  • Popularity may also depend on the predicted result of the line - for example, many people may avoid a frequent drawing line as White, but perhaps you in fact want to have that as a solid weapon against higher-rated players.  Other lines may be unbalancing or relatively risky, but again that may be exactly what you need, as long as you understand the trade-offs in the positions you reach.
  • It matters which databases you use and why.  Correspondence games can be far more accurate than OTB collections, for example, so are very valuable to theory.  For practical use in OTB or online tournaments, though, it can be a bad decision to pick the theoretically "best" line if it runs 20+ moves of memorization, with multiple branching variations, and any deviation from it will likely benefit your opponent.
Engine recommendations also cut both ways.  They can be helpful, mainly for checking tactics and ideas to see what responses are likely and/or best.  They can also be potentially harmful to your game away from the computer.  Anyone who has worked extensively with engines knows that they may come up with certain moves in the opening that may look all right in the short term, but go against the main (human) ideas for the opening and so will cause problems 10-15 moves later.  One example I ran across early in my studies was in the main line of the Caro-Kann Exchange Variation, where most engines (even up until recently) evaluated 7...Na5 as best; you can even still see this on the ChessBase LiveBook with Fritz evaluations showing this as recently as 2016.  However, as Fischer-Petrosian (Belgrade 1970) and others have shown, it really does not work very well in practice.
  • Sometimes engine preparation can also give you a false sense of security, as illustrated by this entertaining example from GM Danny Gormally on a failed experiment in the Slav's Geller Gambit (as White).  On a practical level, I found his annotations useful and adjusted my own related Slav repertoire line as a result - but only after checking other database lines and engine possibilities and looking for myself at the positions.
  • Some computer products will give you a hard-coded numeric engine evaluation for literally every opening move, or you can replicate that by just running an engine.  I don't believe that these are very helpful in general and to evaluate lines you will have to follow them to the end and understand the why rather than letting something like a computer's "+0.25" fully define your view of them.
On a broader level, I think it's also important to understand and acknowledge the amount of work that will be needed in order to understand your chosen opening's positions and typical middlegame plans, then execute them in practice.  Some level of memorization is needed, if not of entire variations then things like key ideas, squares and principles for a particular opening.  Examples include the central important of the d5 square in the English Opening, the theme of the f5-f4 pawn advance in the Dutch, and the critical importance of the light-square bishop for Black in the King's Indian.  This type of opening lore I think is among the most valuable information for improving players and is why finding insightful explanatory material or having a coach who understands and can impart these principles is more important than following the latest professional theory, which often times is only expressed in long variations.

In the end, perhaps it's best to recall Kortchnoi's advice (mentioned in Annotated Game #175) to just go ahead and start playing a new opening, as - if you analyze your own games - that is how you will learn best what works and what doesn't in the opening.

28 May 2017

Annotated Game #175: Epic Stonewall exhaustion

This final round tournament game followed Annotated Game #174 and was the first time that I had essayed playing the Stonewall Dutch, outside of a simul game with GM Sam Shankland.  It taught me a lot about the opening, above all the need for patience (which I did not have enough of) when constructing a kingside attack.  There are many ups and downs in the course of the game - the critical phase starts at move 28 and goes all the way to the end of the game - and we were one of the last ones to finish in the round.  The toll of fighting a complicated battle for 30 moves straight along with the psychological downward trend in the end did me in, as I was exhausted from what felt like an epic fight, with my opponent on the ropes but eventually coming back.  However, there will be other opportunities.  It's also another data point telling me that energy management is something critical to watch (and improve) for my overall performance.

On that note, it's worth recalling something GM Viktor Kortchnoi said when asked about when someone should start playing a new opening they are in the process of learning.  Basically he asserted that you should just go ahead and start playing it in serious games, why not?  Losses will be inevitable, but there's really no other way to get better at it.  I like this outlook, which shouldn't be taken too literally by Class players - some preparation and study is essential, beyond just knowing the first few moves of a chosen opening - but it helps avoid the perfectionist trap of always thinking that your preparation is never "good enough" to play.  At some point, you just need to fire away.

Class C - ChessAdmin

Result: 1-0

A85: Dutch Defence: 2 c4 Nf6 3 Nc3
[...] 1.d4 d5 2.c4 c6 3.¤f3 e6 (3...¤f6 is the Slav Defense.) 4.e3 f5 with this move-order we have what is called a "Slav Stonewall". 5.¤c3 ¤f6 6.¥d3 ¥d6 the Modern Stonewall, instead of ...Be7. 7.O-O ¤bd7 ...O-O immediately is much more common here. No reason to wait. 8.b3 a standard plan for activating the dark-square bishop. 8...O-O 9.a4 done to allow the bishop to get to a3 and exchange off its counterpart on d6. 9...¤e4 a standard and often necessary move for Black in the Stonewall. In this position it is forcing, as the Nc3 is unprotected. 10.¤e2 £e7 keeping my options open and also deterring Ba3. 11.¤e1 I welcomed this, as I felt it was a waste of time for White. The intent is obvious, to push f3, but moving the knight back to the first rank does not seem worth it. 11...g5 I was in an aggressive mood from the start of the game and this move shows it. Not a very sophisticated approach.
11...a5!? would be good prophylaxis against White's queenside play. 12.f3 ¤g5 13.£c2
12.f3²12...¤ef6 13.¤c2 g4 a logical follow-up, as Komodo agrees.
13...¢h8 however might have been best to play immediately, as the king needs to vacate the g-file for a rook and I only do this much later in the game.
14.¥a3 c5?! not a good decision, although my opponent does not take advantage of it.
14...¥xa3 is what the engine considers best. During the game, I wanted to preserve the bishop for use in the kingside attack. 15.¤xa3 ¢h8²
15.dxc5
15.cxd5!? dissolves the center to White's advantage. 15...¤xd5 16.e4 gxf3 17.exd5 fxe2 18.£xe2±
15...¤xc5 16.b4? now my opponent is too aggressive. 16...¤xd3³17.£xd3 b6?! it seems that I am not really looking hard at the position and its requirements. Developing the Bc8 is a nice idea, but there are other things that are more urgent, given the pawn tensions at f3 and c4 and a potential weakness at h2.
17...gxf3 would be the direct approach. 18.¦xf3 dxc4 19.£xc4 b5 20.£xb5 £c7³
17...£c7 gives White no good options. 18.cxd5 ¥xh2+ 19.¢h1 ¥e5³
18.cxd5 this would have been strong earlier (move 15), but now I'm OK. 18...¤xd5
18...gxf3!? is better, as once the Nf6 moves away it no longer can recapture on g4 and get a good outpost. 19.gxf3 (19.¦xf3 ¥b7³) 19...¤xd5³
19.b5 my opponent now looks to simplify.
19.fxg4 would break up Black's kingside to good effect. 19...£c7 20.g3 fxg4 21.¤cd4²
19...¥xa3 20.£xa3 ¥b7 21.£xe7 ¤xe7 we now have a very equal-looking middle/endgame position. 22.f4 ¦ac8 23.¤cd4 threatening e6. 23...¢f7 24.¦ac1?! this "obvious move" gives me the initiative as my Ne7 now springs to life. (24.¢f2) 24...¤d5³ returning the favor by threatening e3. 25.¢f2 ¤b4 threatening the fork on d3. 26.¦b1 (26.¦fd1 ¦xc1 27.¤xc1 ¥e4³) 26...¤d3+ this is still a strong move. 27.¢g3µ White's king safety is now something of a problem, which along with my nicely centralized Nd3 gives me an advantage. (27.¢g1 h5µ) 27...h5 here I correctly find the logical follow-up, which raises mate threats. 28.¦fd1? this should lose, but the winning continuation is not obvious. (28.h4 gxh3 29.¢xh3 ¦g8µ) 28...¥e4 a good follow-up move, but not nearly as good as the best move.
28...h4+ secures the point, comments the engine via the Fritz interface. 29.¢xh4 ¤f2! now the White king has no way back. 30.¢g3? (30.¤xe6 ¦h8+ 31.¢g5 ¢xe6−⁠+) (30.¤xf5 ¦h8+ 31.¢g3 ¤e4+ 32.¢xg4 ¦cg8+ 33.¢f3 ¤d2+ 34.¢f2 ¦xg2+−⁠+) 30...¤e4+ 31.¢h4 ¦h8#
29.¤c6µ eyeing the jump to e5 and threatening a7, something I gave too much weight to. 29...¦h8? now I'm not thinking aggressively enough.
29...¢f6 removes the check on e5. 30.¤c3 h4+ 31.¢xh4 ¤c5µ
30.¢h4 this is enough to restore equality.
30.¦xd3! is a simple forking tactic that gets two pieces for a rook. 30...¥xd3 31.¤e5+ ¢f6 32.¤xd3±
30...¤c5 31.¦b4 this solves the dual threat to the Rb1 and a4, but not in the best way.
31.¤e5+!?31...¢f6 32.¦bc1 and now Black cannot go pawn snatching: 32...¥xg2 (32...¤xa4??33.¦d7! with mate coming.) 33.¦xc5 ¦xc5 34.¦d7 and now 34...¦xe5 is forced. 35.fxe5+ ¢xe5 36.¤f4² snagging the bishop, as a fork on g6 is threatened.
31...¥xg2³
31...¥xc6 was the other option. 32.bxc6 ¦xc6³ this had the advantage of getting rid of preventing the knight from reaching e5.
32.¤e5+ I was quite aware of the fact that I had potential mating threats, but now so does White, given the location of his knight and potential rook action on the 7th rank. 32...¢f6 33.¤g3 naturally the h5 pawn is poisoned and can't be taken, due to the subsequent pin against the king. 33...¥d5 34.¦bd4?! (34.¦c1!?) 34...¤b3 the best move, but at this point I was tired and had relatively little time on the clock, so I didn't have a coherent follow-up plan. 35.¦4d3? looks obvious, but should lose.
35.¦xd5 is necessary and only leaves White slightly worse. 35...exd5 36.¦xd5³
35...¦c2−⁠+ again another best move and obvious follow-up, but without clear vision of a winning continuation. However, the next series of moves are simple enough. 36.¤f1 ¦hc8 37.¤d7+ ¢e7 38.¤e5 ¤c5 good but perhaps not best. I felt I should at least keep making threats, feeling somewhat frustrated that I could not find a breakthrough. 39.¦d4 ¢f6 played to take away the g5 square from White's king. 40.h3 now I felt I should be able to break through. 40...gxh3 41.¢xh3 ¥g2+ unfortunately here I could not find a winning idea, under pressure.
41...¤e4!? would bring another necessary piece into the attack, since d7 does not in fact need to be guarded. 42.¤d7+ (42.¦xe4 fxe4 43.¤g3 ¦c1−⁠+) 42...¢e7 43.¤e5 ¦g8 44.¦xe4 ¦g1−⁠+ and mate threats mean White loses material.
42.¢h4 ¥d5 43.¢h3 ¦g8 (43...¤e4 again is the key. 44.¦xe4 fxe4 45.¤g3 ¦c1−⁠+) 44.¤g3 h4
44...¦g2 Black missed this excellent chance, comments the engine. 45.¤xh5+ ¢e7 46.¤c6+ ¢d6−⁠+
(44...¦h8 is also good, preparing to push the h-pawn.) 45.¤h5+µ45...¢e7 46.¤c6+ ¢f8 now we're back to equality... (46...¥xc6 47.bxc6 ¦gg2µ) 47.¤f6? except that this (again) should lose for my opponent.
47.¦xd5 leads to a perpetual. 47...exd5 48.¦xd5 for example 48...¦g1 49.¦xf5+ ¢e8 50.¤f6+ ¢f8 51.¤h5+ ¢e8 etc.
47...¦gg2 48.¦h1 ¤e4?! unfortunately this was a good idea several moves ago, not now.
48...¦g3+ and Black wins 49.¢xh4 forced 49...¦g6 with a double attack on the Nf6 and the h6 square (threatening the Rh1 via a skewer check). 50.¦xd5 exd5−⁠+
(48...¦g6?! immediately doesn't work, as White simply replies Nxd5.) 49.¤xe4³49...¥xe4 50.¦d8+ ¢g7 51.¦d7+? again my opponent offers up an opportunity. (51.¤e5) 51...¢h6−⁠+52.¤e5 ¦g3+ a great idea...on move 48. Here it blows the discovered attack by the Be4 on the Ra1, since the Rg3 will be hanging.
52...¦gf2 moving to e2 works fine as well. 53.¢xh4 forced 53...¥xh1 54.¤f7+ ¢g7 55.¤e5+ ¢f8−⁠+
53.¢xh4 by this point I'm totally exhausted and out of ideas. 53...¦g7 simplification is actually a good route to go and should result in a draw. 54.¦h3 ¦xd7 55.¤xd7 ¦g2 keeping hopes of a mate threat alive. 56.¤e5 The knight dominates, comments the engine (correctly). 56...¥c2 57.¦h1 ¥xa4 58.¦a1 ¥xb5 59.¦xa7 ¥e8? this really made no sense, but my brain was too tired from all the calculating and I missed the simple follow-up. The original idea was to dominate the Ne5. (59...¦g7 was simplest.) 60.¦e7 at this point I just gave up, seeing that I would lose the two pawns and was exhausted. The game is far from over, though. (60.¦e7 ¥a4 61.¦xe6+ ¢g7 62.¦xb6²)
Powered by Aquarium

05 May 2017

E-book completed: Play the Caro-Kann



I've had the e-book version of Play the Caro-Kann (2007, Everyman Chess) by IM Jovanka Houska for a while now, but just recently took the time to go through it in a systematic way, after a previous false start.  There's a lot of material and it's more advanced than Starting Out: The Caro-Kann, although Houska's latest (2015) Opening Repertoire: The Caro-Kann is about double its predecessor's size.  (I have that as well and plan to follow up this year by looking at it seriously, which should provide some interesting comparisons.)

This is a valuable opening book in general, based on selected major repertoire lines for the Caro-Kann.  It does some things that many other opening books (whether a repertoire or "comprehensive" opening manual) do not.  In this case, Houska does a particularly fine job of highlighting both strategic and tactical issues associated with deviations from the central line, something clearly beneficial to improving players.  One of my general complaints as a Class player is that many opening books only give the "best play" and most contemporary lines.  Below master grade (but really even then), it's unlikely that your opponent will go straight down the "best" path that the book is so enthusiastic about.  (I put "best" in quotes because the "best" lines from 5, 10 or 20 years ago are unlikely to be the ones given today.  Sometimes entire major variations, like the Berlin Defense, are largely ignored but then resurrected and become "best", so it's best to think for yourself when building your own repertoire.)

Knowing on a deeper level the likely opening deviations and how to punish them (or at least get a better game) is key to getting a practical advantage out of your selected openings.  This is closely allied to the importance of understanding your openings' concepts, not just the latest variations; in many books play may be taken simply for granted until far into the game and the GM writing the book may not bother to even mention normal-looking alternatives.

Below are the chapter headings for the book with some commentary on content.
  • Chapter 1 - Main Line: Introduction and 11. Bf4.  The author chooses to start off with the 3. Nc3 main line, which is probably the most deeply analyzed historically, and selects the Classical (aka Capablanca) variation to meet it, 4...Bf5 (which I play).  Houska does a good job of discussing some of the introductory concepts about the variation while fast-forwarding ahead to move 11, which in this case is legitimate; I actually have played a number of games in which White can rattle off things automatically to this point (and beyond).  This particular chapter may not be 100% relevant to your repertoire if as a Caro-Kann player you don't play an early ...e6 instead of developing first with ...Nf6, but many of the lines transpose - it's hard to avoid playing ...Nf6 - and the various ideas are good to see in action regardless.  In particular, the ideas of hitting a white pawn on c4 with ...b5 and how to handle White's attacking idea of g2-g4 are well treated.
  • Chapter 2 - Main Line: 11 Bd2.  This is really the meat of the analysis for my repertoire and the foundation of the Caro-Kann main line, so all the ideas are relevant.  Knowing how to handle thematic White attacks like the sacrifices on h6, g6, f5 and e6 are very important, as is setting up Black's counterplay on the queenside and along the c-file in particular.  Caro-Kann players will get a sense of the cut-and-thrust of these positions and also have their morale boosted by seeing how dangerous-looking White attacks can be foiled.  In this chapter Houska also treats the early deviation 8. Bd3, in which White basically plays the same way but foregoes h4-h5, and suggests that Black go for queenside castling.
  • Chapter 3 - Main Line: 6. Bc4 and Early Deviations.  Here we find a White transposition to the Blackmar-Diemer Gambit (via 4. f3) and Houska advocates taking the pawn, which is principled but can be dangerous if you don't really know the lines. Her preference of following up with 5...Bf5 and subsequent play looks pretty straightforward for Black.  Separately, the main line with 5. Nc5 (used by Fischer a few times) is treated with 5...b6 as the standard response.  The various 6th move alternatives, especially 6. Bc4 and 6. N1e2, need to be studied carefully by Black players, since the ideas are different from the regular main line.  6. f4 is also a rather violent line and should be looked at, since it has good surprise value for White and Black needs to know recommended piece development (Bf8-d6, Ng8-e7).
  • Chapters 4-5 treat the Panov-Botvinnik Attack, using the theoretically recommended 5...Nc6 line, which requires Black (especially) to really know the sequences, since the variations are sharper and White has the initiative and some lasting pressure; however, Black is fine (or sometimes better) in the end, due to better structural and positional factors.  Instead I play the 5...e6 line, which is more solid and typically transposes into a position classified as a Semi-Tarrasch defense (from queen's pawn openings).
  • Chapter 6 - Exchange Variation.  This is especially popular with Bobby Fischer fans and relatively straightforward to play.  Houska uses 5...Qc7 for her repertoire, which is a strong move (but not the main line 5...Nf6) that pre-empts the usual Bf4 development for White. (For an entertaining post on the variation by a fellow improvement blogger, see "The Grinch during off season") 
  • Chapters 7-8 go over the Advance Variation, which at top levels has long replaced the "main line" (3. Nc3) as the most popular approach for White.  Interestingly, this hasn't really happened at the Class level (at least in my experience), which means that players of both sides may be put off to some extent by the large amount of theory after 3. e5 Bf5, which is the most logical choice for Black, who places the light-square bishop outside of the pawn chain before playing ...e6.  Houska instead advocates the sideline 3...c5 and does a good job of covering it in depth, which is rare in Caro-Kann literature.  It is the only real gambit variation in the Caro-Kann but it is still a quite solid approach in most lines, as White usually either has to give back the pawn or is left with no real prospects for making progress.  Black (as Houska advises) should concentrate on the positional compensation rather than desperately trying to recover the pawn, so it is a genuine gambit line.  It's worth noting that Houska after the main theoretical continuation 4. dxc5 only presents 4...e6 for Black, which is the second most popular choice (after 4...Nc6).  Both lines score about the same in the database (43% for Black).  I think it's mostly a matter of taste, as the position naturally becomes more French-like after the early ...e6. 
  • Chapter 9 - Fantasy Variation.  This line (with 3. f3!?) is a favorite of White players impatient with normal Caro-Kann lines who just want to attack on the kingside (similar in some respects to the approach of the Blackmar-Diemer Gambit per above).  The Kenilworthian blog summarized interest in the variation back in 2010.  If Black players just make mindless exchanges that leave the center wide open, then White has a nasty quick-developing attack.  That said, the theoretically most critical line has Black play 3...dxe4 followed by 4...e5 (and then the e5 pawn becomes the linchpin of the position).  Houska, who tried the line herself briefly, has an excellent summary analytic overview and emphasizes White's weakness on the dark squares (including f2), arguing that should be the basis of Black's strategy.  She notes that if Black players want to just have a solid game rather than challenging White, then 3...e6 and 3...g6 are good alternatives.
  • Chapter 10 - Panov's Little Brother: 2. c4.  It's good that Houska devotes an entire chapter to this line, which can be reached from multiple transpositional possibilities and can itself transpose into a full-fledged Panov-Botvinnik Attack.  Black can't simply ignore the unusual move and should be prepared for it.  Houska treats the main line approach, which after the pawn exchanges on d5 is to play 4...Nf6 rather than immediately recapturing with 4...Qxd5, which is reminiscent of the Center Counter defense.
  • Chapter 11 - Two Knights Variation.  This has always been at least somewhat popular at the Class level, I think largely because it's an obvious piece development.  Also probably because White gets some cheap points after unwary Black players try to treat it like the Classical Caro-Kann and exchange on e4 followed by 4...Bf5?!, which gets punished every time by White (per the above link).  The standard antidote (covered by Houska) is 3...Bg4, which is good for Black if the positional ideas behind it are understood (maintain the light-square bind with the pawns, don't open the position further).  It's again worth noting the quality job Houska does of explaining the key ideas and plans, rather than giving a variation or two with a one-sentence comment.
  • Chapter 12 - King's Indian Attack.  (Yet another Fischer sideline in the Caro-Kann.)  A more common setup against the French or Sicilian (as illustrated in the above link), it can also be used against the Caro-Kann.  Black has nothing to fear, in this case having a couple of advantages (already having the d5/c6 pawn chain and the move ...e5 available early) and so equalizing rather quickly, has been my experience.  Both the KIA and the Caro-Kann are solid openings, so there are few fireworks and the middlegame tends to be one of slow maneuvering rather than breakthroughs.  If you play the KIA as White in other openings and don't want to learn something new against the Caro-Kann, it's certainly a practical option.  Houska devotes a full chapter to it and a number of variations that are unlikely to appear on the board, mainly White deviations from the 5. g3 setup, so it may be more valuable in practical terms for White players looking for alternatives.
  • Chapter 13 - Unusual Lines and the Plain Bizarre.  The most important/relevant line here I think is 2. f4, which Black meets by playing in the style of the Advance Variation (with 3...Bf5) but I've also seen in practice the line 2. Nf3 d5 3. exd5 cxd5 4. Ne5, which has been used by some top Grandmasters for surprise value.