Sunday, March 15, 2009

Repertoire Renovations

Somebody once said that each time you change your opening repertoire you "grow" as a chessplayer. Well, I should be a grandmaster by now for all the growing I've done over the years. As I put the finishing touches on my latest opening system, which may be my most coherent to date, I begin to wonder if I'm really growing so much as settling into a new approach to the game.

I think I've had dozens of repertoires over the years. The last one I wrote about in these pages was my Knightmare Repertoire (built around mirror systems with e4, Nc3, and Nf3 as White and e5, Nc6, and Nf6 as Black), which bears some connection to my current approach. I also documented my Caveman or Left Hook Repertoire for White (which included the Urusov Gambit, Two Knights with d4, Steinitz-Sveshnikov Attack, Evans Gambit, Modern Horowitz Max Lange, Anti-Petroff with d4, Left Hook Grand Prix, Left Hook Austrian, Wing Gambit French, Caveman Caro-Kann, and Saemisch Attack vs. the Alekhine) and various unusual systems as Black (including a universal 1...Nc6 system and an early d6 system that included the Philidor, Panther, and Janowski Indian or Janowski's Brother). Along the way I've also flirted with various completely different ideas, including the d-pawn repertoire built around the Barry Attack and Colle-Zukertort as laid out by Aaron Summerscale in A Killer Chess Opening Repertoire and recently updated by Richard Palliser in Starting Out: D-pawn Attacks.

I really don't know how other players go about constructing their repertoires, and it might make for a good interview question -- or a question for readers. How do you go about it? Steve Giddins once devoted an entire book to the subject (How to Build Your Chess Opening Repertoire). Perhaps most players simply pick up one of the many books on the market offering a ready-made repertoire and read it carefully from cover to cover. While I love repertoire books, I rarely find myself adopting more than one or two lines from any of them. Most book repertoires are simply not coherent enough for my taste. Perhaps some players work with coaches to create a coherent system, or perhaps they just let things fall into place over time, with new acquisitions coming on organically along the way.

It's my desire for coherence that drives my changes, and I am rather dogmatic in my approach. I want to have a coherent system, and so a change in any opening line will inevitably trigger a cascade of adjustments throughout my repertoire. The Knightmare Repertoire came about because I found myself playing the Tango (1.d4 Nf6 2.c4 Nc6 3.Nc3 e5) and the Berlin Variation of the Ruy Lopez (1.e4 e5 2.Nf3 Nc6 3.Bb5 Nf6), and soon it just seemed natural to adopt the Four Knights, the Two Knights French, and so on down the line until I was playing practically every system where you move your two Knights and your e-pawn in the first three or four moves. A similar process created the Caveman repertoire for White, which arose out of my desire to develop a dark square attacking formation (generally centered around an e5 advance).

My current system began much the same way. I had been building up an Open Game system with 1.e4 e5 as Black when I started playing lines with a kingside fianchetto, especially the Three Knights (1.e4 e5 2.Nf3 Nc6 3.Nc3 g6). Meanwhile, I was struggling to find a good line against the Italian Game that did not involve too much study or any gambits, and just around that time I happened to annotate the game Weeramantry-Bisguier, USATE 2008 which suggested that a kingside fianchetto might work more universally than I had imagined possible. I decided to see how far I could take the idea and started researching other lines where Black plays an early g6 advance in the Open Games, including the Smyslov Variation of the Ruy Lopez (1.e4 e5 2.Nf3 Nc6 3.Bb5 g6). I started playing it in blitz and really enjoyed the types of positions I was getting in those lines, which reminded me of my most positive experiences with the King's Indian Defense. I had given up the KID years ago when it just seemed too difficult to keep up with the theory. But maybe a modified King's Indian system was possible, one that did not involve too much study? As I studied the Open Games with g6, the KID seemed inevitable -- and a natural addition since some KID lines could arise via the Open Games (such as by 1.e4 e5 2.Nf3 Nc6 3.d4 exd4 4.Nxd4 g6 5.c4, which heads toward an interesting line that could arise via the Classical KID).

With the KID and the Open Games with g6 on board, I soon found myself looking at other lines with a kingside fianchetto, such as the Glek System of the Four Knights (1.e4 e5 2.Nf3 Nc6 3.Nc3 Nf6 4.g3), the Vienna with g3 (1.e4 e5 2.Nc3 Nf6/Nc6 3.g3), the Closed Sicilian or an Open system with g3 (depending on Black's set-up), and an anti-Pirc system with g3. It all seemed so natural.

Of course, there were some complications. For one thing, in order to avoid taking on the whole King's Indian, I decided it made sense to keep the Tango against any line where White did not commit to Nf3. That way I could avoid the Four Pawns Attack, the Saemisch, and a whole host of other White systems. Keeping the Tango also allowed me to focus generally on lines where Black plays an ...e5 advance, as in the Classical and the Fianchetto (with Nbd7 and e5). And the Tango made sense because I was also drawn to the Two Knights French and Two Knights Caro-Kann, which were practically like reversed Tango systems in some lines (especially where Black plays a d4 advance).

There were many other adjustments, of course, too numerous to mention -- and I am still trying to work out all the transposition tricks and marginal lines. Here is the broad outline of what I have so far and what books I've been looking at to help me organize my study (suggestions for additional lines and useful books or articles are most welcome):

Black

Smyslov Variation of the Ruy Lopez (1.e4 e5 2.Nf3 Nc6 3.Bb5 g6)
Italian Game (1.e4 e5 2.Nf3 Nc6 3.Bc4 g6 or 3.d4 exd4 4.Bc4 g6!?)
Three Knights and Scotch (1.e4 e5 2.Nf3 Nc6 3.Nc3 g6 or 3.d4 exd4 4.Nxd4 g6)
Center Game (1.e4 e5 2.d4 exd4 3.Qxd4 Nc6 4.Qe3 g6!?)
Tango (1.d4 Nf6 2.c4 Nc6)
King's Indian Defense (1.d4 Nf6 2.Nf3 g6)
English (1.c4 e5 2.Nc3 Nc6 3.Nf3 Nf6)

White

Paulsen-Mieses Vienna or Glek Four Knights (1.e4 e5 2.Nc3 Nf6/Nc6 3.g3 or 1.e4 e5 2.Nf3 Nc6 3.Nc3 Nf6 4.g3)
Closed Sicilian or Open with g3 (1.e4 c5 2.Nc3)
Two Knights French (1.e4 e6 2.Nc3 d5 3.Nf3)
Two Knights Caro-Kann (1.e4 c6 2.Nc3 d5 3.Nf3)
Alekhine (1.e4 Nf6 2.Nc3)
Pirc (1.e4 d6 2.d4 Nf6 3.Nc3 g6 4.g3)

16 comments:

Anonymous said...

I favor convenience over coherence. I don't play much now, but I answer both 1.e4 and 1.d4 with 1...c5; As white, I play 1.c3 and usually follow 2.c4. Essentially I need to only learn the Benoni and the Sicilian. While I'm giving up the advantage of White to my opponent in every game, I'm getting a 2 fold advantage in preparation. Plus, I can use other options, like 1.c3 e5 2.d4 for a reversed Caro, if I think my opponent is a Sicilian killer.

-Devin

katar said...

Very coherent and well-thought out. With your professional writing background and obvious passion for chess, i would love to see a GM/Goeller collaboration on a book published by Gambit or QualityChess. Hey, if NIC published "The Lion, Predator's Weapon", anything is possible!

Michael Goeller said...

I picked up "The Lion" when it came out and think it is quite good. I should really do a post on "good and bad opening books by amateurs," since it is something I seem to collect.

The idea has definitely occurred to me before to do a book, and I have several projects in mind (including simply an edited book of stuff from the blog -- or from chess blogs in general). But I am so busy with book, web, and film projects at work that it is tough to imagine at the moment. Perhaps with a collaborator, as you suggest, it might be more doable. More food for thought.

Anonymous said...

Wow, so deep, I like this

Ian said...

Hey Mike,
I faced the same problem when I took up the game again a couple years ago - I couldn't remember any openings so I had to start from scratch. I built my repertoire randomly at first, but settled on the general theme of asymmetry and solidity. That gave me the nimzo/queen's indian against d4, positional sicilian lines against e4, and solid 1.e4 lines for white. Granted, there's a lot of lines to learn to fill in the gaps, but that adds to the variety and challenge. And if I had to play the London system or something like that every game I would hang myself.

Anonymous said...

Hi

Thanks for a very interesting and useful article. Your repertoire is VERY similar to my current repertoire (ELO 1900). The few differences are:

- With black I like the Nimzowitsch defence 1. e4 Nc6 2. d4 e5, which I notice you play in some of the games you publish. I obviously hope for 3. d5 and Tangoish play. After a lot of trial and error, I have ended up playing e5 in response to 2. Nf3 – followed by many of the g6 open lines you give, both to avoid theory and because I enjoy the King's indian feel to them.

- Otherwise more or less the same, although I also play the two knights with ...Bd7 after 4. Ng5/d5/exd5/Na5/Bb5+ or h6/g6/Bg7 against the more quiet 4. d3-variation. I also play the g6 Scotch, but find it somewhat more vulnerable to an attack than some of the other g6 open lines, may be due to the loss of the e5 stronghold.

- Playing white against the Sicilian, I like 2. Nc3 Nc6 3. Bb5, and Bxc6 given the chance. Then d3, f4, Nf3, 0-0 etc, as described in Gawain Jones' book. Very easy to play, and some of the plans are quite similar to the Nimzo-defence and Tango with Bb4. If 2. d6 then either the closed or the GP.

- I'm not completely sold on the french two knights as I find black often ends up with a very strond centre. So I have been playing the 2. Qe2 with g3/Bg2 kind of KIA.

- Against the Pirc, I either play the g3 variation you mention or a GP set-up with the white-squared bishop on either e2, c4 or b5. It depends how early black plays c5/Nc6.

- Otherwise the four knights Glek, two knights caro-cann, 2. Nc3 against the Alekhine AND the Scandinavian (!) etc

The common thread in many variation here is the pawn formation d3/e4/f4 or with black d6/e5/f5. Nice active play with a space advantage on the kingside. As well as opportunities to attack, my own king position is very safe. As a probable “Activist” in Lars Bo Hansen’s terms, I am really crap at defending, so active play with a safe king is key. The fact that the attack takes a while to build up is of no concern. In some lines I go for Bb5 instead of Bg2, which quite often damages my opponent’s pawn structure. I thus have that to fall back on, if the strategy of opening the f-file and attack on the king side fails.

Anyway, that was a LOT of words. Just got inspired by your excellent article.

Michael Goeller said...

Thanks for the last comment. It's fascinating how two people can arrive at such a similar repertoire. I was playing the Nimzovich for a while, as you know, along with 1.d4 Nc6 (which is in some ways easier than the Tango since you avoid the Trompowsky and such). But I lost interest in the Chigorin (after 1.d4 Nc6 2.Nf3 d5) and alternatives (such as 2...d6) led to other stuff I wanted to avoid (1.d4 Nc6 2.Nf3 d6 3.e4!) Meanwhile, like you I was always transposing to the Open Game anyway, so I decided to just skip 1...Nc6 and go right into 1.e4 e5.

I agree about the Two Knights French and have had a love-hate relationship with it for many years. I rediscovered it after reading The Flexible French....

I'll have to go through your post more carefully. thanks for the helpful and thoughtful response.

Anonymous said...

I think opening repertoire building is a hit-and-miss process for almost everyone.
"...most players don't have an elegant opening repertoire. They play a motley collection of openings which they have accumulated, more or less by chance, over the years." -Nunn, Secrets of Practical Chess.

I've tried "coherent" repertoires, like the Nc6-based one from Keene and Joacobs' book. One day I made some statistics of my results, and found I did brilliantly with 1.e4 Nc6 but extremely poorly with both 1.d4 Nc6 and 1.d4 d5 2.c4 Nc6. So I moved on to other defences to 1.d4. Now I face a similar problem with my beloved "1...d6 against everything" repertoire; the Pirc is kind of the cornerstone but I find it hard to believe in theoretically.

I've ended up with a very empirical approach: Browse some databases and books, try out lines i like the look of, and keep only those I do well with. I also try to keep an eye on what GMs are doing; if lots of strong players use a move order that seems to "kill" one my openings (i.e. equalize easily with black or get an edge with white) I must worry that my opponents will take it up. F.ex if I were to take up the Two Knights French again, I would have to find something good against 3...d4 4.Nce2 c5 5.c3 Nf6! [5.Ng3!?] and 7...Nxc5!?, both played with success by some very respectable "Frenchies".

As White my empirical approach has led me away from GPAs, Bishop's Games and d-pawn Attacks to more and more main lines. Many of them are fun to play, and the great thing with both the Open Sicilian and 1.d4/2c4 is that you still have several options to fight for an edge against most black defences. Besides both the "classics" and current top chess becomes more interesting if the openings played are directly relevant to my repertoire.

A bonus of switching to main lines is that the occasional return to a GPA or a Trompowsky retains its surprise effect! Main lines require some well-developed instincts though; you need to sense in which lines and against which opponents you are likely to get out-prepared and avoid them, unless you are supremely well-prepared of course. I'm apalled by the number of amateurs who enter the Najdorf (as White or Black) with almost no theoretical knowledge, just because they saw Fischer and Kasparov play it, so it must be good...

I must say though that after 1.e4 e5 I think of the Italian as a main line these days, and your analysis on this blog deserves quite a bit of credit for that!

S. Martinsen, Norway

Michael Goeller said...

In his latest "Ask GM Joel" column, in response to a reader who is thinking of picking up the "Polar Bear" (or reversed Leningrad Dutch) as a way of keeping play within his known territory, Joel Benjamin writes: "I’m not a big fan of systems in general because the richness and diversity of positions you receive from an opening stimulates your learning and improvement. Systems are easier to play, but unless you have a tendency to get blown out in the opening or get into time pressure habitually, I don’t recommend them." He goes on, though, to suggest that the Polar Bear sounds like a pretty good choice -- so long as he does not mind putting up with From's Gambit (1.f4 e5).

I agree with Benjamin: it is probably a good idea as a developing player to try out a wide variety of openings. But I also think that mastering systems and then moving on to other systems makes sense also and gets you rather deeper into the ideas of positions than the sometimes superficial study of a wide variety of lines. It's basically the difference between going deep and studying broadly. Ultimately you need both. I am not taking on my latest system to avoid learning new things; I am taking on a system that I expect to teach me some very deep understandings of related positions.

You have to make your own decisions about these things. But I think you have to spend time with opening ideas before you really know them well. I can think of specific lines it has taken me years to understand beyond the first dozen moves, for instance, while others I could learn in a single day.

katar said...

FYI, Wisnewski's Play 1...Nc6 covers your line against the English from a repertoire angle.

David Rudel said...

Michael,
If you decide you want to play the Colle-Zukertort again, I could probably find a review copy of my book Zuke 'Em: The Colle-Zukertort Revolutionized, prefaced by GM Summerscale. It's particularly useful (for higher-level players) at dealing with Anti-Colle lines.

-David Rudel

Anonymous said...

For some years I have built my Black repertoire around the move 1.-e6, which is flexible and good when French Defence is included. For example 1.d4 e6 2.c4 f5 is good way to transpose to Dutch and avoid some problematic variations like 1.d4 f5 2.Nc3 or 2.Bg5. Also 1.d4 e6 2.c4 Nf6 transposes easily to Nimzo/Bogo/Benoni while avoiding Trompowsky. With White I have used recently only 1.d4, although some overanalyzed variations of the solid Queen´s Gambit Declined and sharp Benoni are not to my taste. Maybe Colle approach with 1.d4, 2.Nf3, 3.e3 is a good alternative to theoretical mainlines. My guess is that 1.d4 Nf6 2.Nf3 c5 3.e3!? makes many Benoni and especially Volga players quite unhappy.

Michael Goeller said...

It's funny, but I was just recommending this repertoire the other day to a chess player with whom I've had a correspondence -- suggesting especially the Stonewall Dutch. I think it makes a great fit with the Colle-Zukertort and the "Killer Chess Opening Repertoire" as White also. Still another choice is something like the Queen's Gambit Declined, Lasker's Defense or Tartakower. I have myself been experimenting in blitz with a 1...e6 repertoire featuring the Rubinstein French and Lasker's Defense.

Anonymous said...

I,m trying to make a Bg5, Bg4 system work but it is no garantee that you end up in a systemlike position.
The problem is that white doesnt have to give up his centre and that black does!
White: Torre d4 Nf3 Bg5
Black : Scandinavian e4 d5 ed5 Nf6 Nf3 Qd5 and later Bg4
Slav Steiner: d4 d5 c4 c6 Nf3 Nf6 Nc3 dc4 Bg4
With white play is slow and steady but with black play is dynamic so at the club I have to change my state of mind every week!
A repertoire should be coherent at that point.

Anonymous said...

I once formed a repertoire for White based on the KIA starting with 1.e4. I agree with Joel Benjamine's commentsabout systems - too restrictive and ultimately boring. But the KIA approach allowed for growth - I felt the KIA was insufficient against the Caro-Kan, so I added in the Panov. Later I thought the 'Dragon' setup against the KIA was too tame, so I added in the Hungarian Attack - 1. e4 c5 2. Nf3 d6 3. d4 cd4 4. Qd4, using the KIA against 2...e6 and the French. Similarly, against Pirc I dropped g3 and went to the Karpov approach. Against 1. e4 e5 I played2. Bc4 - no petroff, no long Ruy lines. I went from the Bishop's Opening into the Vienna. The exact choices are not important - a matter of taste - but after a while I had not only a rather fierce White repertoire but also I could fall back to the pure KIA approach as a safety net. There is no boredome, great elimination of opponent's special preperation, room to grow.

Michael Goeller said...

That sounds fairly parallel to my own development, which also included the King's Indian Attack -- and which may explain why I gravitate toward these fianchetto systems. I keep bouncing around to different repertoires, but now I feel like I also have a lot of possible ways to play, depending on opponent, situation, or even mood.