Monday, June 4, 2007

Interesting Opening Statistics

What follows is a statistical analysis of Kasparov's 1.e4 openings using Chessbase. This is just a quick and dirty dissection of the information I found in the Big Database 2006. You can quibble exact numbers for my statistics, but shouldn't change the take-home message.

I was playing around with Chessbase's Player Dossier feature by pulling up Kasparov's information, and was surprised how few main-line Spanish games there were. I then did some searches in the database for Kasparov vs. people rated 2100 and up (to weed out the simuls and such). I did all of these as position searches so that I didn't lose any games because of transpositions. Here's what I found, and what the implications are for the average club player.

  • Games played by Kasparov with 1.e4: 340
  • Games where Kasparov reached the Spanish with 3...Bb5: 69 (so 20% of his e-pawn openings brought him here).
  • Games where the opponent played 3...a6: 58
  • I then searched for the main position that arises after 1. e4 e5 2. Nf3 Nc6 3. Bb5 a6 4. Ba4 Nf6 5. O-O Be7 6. Re1 b5 7. Bb3 d6 8. c3 O-O 9. h3: 22 games (so less than a third of his Spanish games end up here, or 6% of his 1.e4 games).
  • At this point, only 4 games continued with 9...Na5! Instead, 14 continued 9...Bb7 (11 of them from games with Karpov). In other words, only 6% of Kasparov's Spanishes as white (and only 1% of his 1.e4 games) resulted in Black's primary system in the Closed Spanish.

You can make arguments such as "well, that's because the Berlin and the Marshall Gambit have really taken off", or "but Kasparov has played the Scotch", or "people avoid the main line against Kasparov because they're afraid of his preparation" and so on. Out of 340 1.e4 games, there were 22 scotches, 3 Evans Gambits, and 6 Italian/Two Knights. In the Spanish, Kasparov pretty much stuck to the main lines, deliberately avoiding the main line on 6 occasions. Still, of the 94% of Spanish games that didn't end up at the main line, only 9% were Kasparov deviating, so the fault lies more with his opponents.

Now, as far as Black deviating to avoid Kasparov's preparation: the Spanish is an opening where the main line is pretty much delineated, as opposed to some branching monstrosity such as the Sicilian. Black has some serious deviations at move 3, at the Open Spanish, and the Marshall Gambit, but apart from that it's pretty much on rails. I have in my possession Glenn Flear's The Ruy Lopez Main Line, which covers only from move 9 on. Kasparov would only have reached positions in Flear's book in a small fraction of his Spanish games. I would have expected a larger number of his games to follow the main lines, but these statistics make it look like I have about as much chance of getting the main line Spanish in my blitz games on ICC as Kasparov did in a GM tournament. Perhaps more so, since some of the ICC players are theory monkeys that belt out the main lines. Lower-level players may play fewer Berlins and Marshalls, but they play far more Steinitz Defenses, Cozios and "junk" like that.

So, what's the take-home message here?

  • The farther down a main line you go, the less of a chance of seeing it in your own games (well, der!)
  • Memorization of main lines won't solve your opening problems.
  • If you buy an opening book with the intent of using it as a map to direct you through the opening, you're wasting your money. Books that deal with common pawn structures, such as Kmoch's Pawn Power in Chess, are a better use of your study time.

Nonetheless, I recommend people play and study main lines. Why?

  • You will find more annotated master games featuring those lines, so you'll develop more understanding about why certain moves or variations are played.
  • You will see examples of good development, effective piece manoeuvres, thematic pawn structures, and typical middlegame plans, which may arise in your own non-mainline games.
So, if you treat an opening book as examples of good opening and middlegame play that you may choose to adopt, and read through the lines critically, they'll have some utility.

In my opinion, you really don't need to worry about any main line past what I consider to be the "real" end of the opening, which is typically where the minor pieces and queen are off the back rank, and the players have castled. I say "typically" because some lines will have a stay-at-home bishop, a delayed queen deployment, no castling with a closed center, or some such exception. If you follow general opening principles of development and central control, you may find out that you've just played the same line as a GM. You're best off treating opening manuals as examples of possible play, rather than "you must commit these lines to memory and execute them unfailingly". Rather, just be sure to analyze your own games (and all phases too...middlegame and endgames as well). If you find that you followed a "book" line, check it over until you find where you or your opponent left theory (often it's as simple as Fritz finding the patzer move), and figure out what you'd do if that were to arise again. If you've actually successfully followed a book line past the end of development, then go ahead and explore some typical lines and master games that also occurred in that line. For this purpose, books such as the "Mastering the..." series that discuss typical pawn structures and associated plans, as well as general material such as Kmoch's book, are of more value to amateurs than more specialized tomes. King and Ponzetto's book Mastering the Spanish is a classic in this regard.

{Postscript: after composing this, I obtained Mihail Marin's Beating the Open Games, which fills the same role as Emms' Play the Open Games as Black (a 1.e4 e5 repertoire for Black, except the Spanish--although Marin mentions the Spanish Exchange as well) but which a whole lot of explanatory text. It may become the only openings book I've ever read cover-to-cover, since it seems to be chock full of instructional material. It's too early for me to give a review, but right now I'll just say that it appears to be an openings book that will make you a better player. When I searched online for a hyperlink to Marin's book, I found that he has a repertoire book for the Spanish coming out soon as well. I'll be looking forward to its release.)}

In a future diary I hope to show an example of what I consider to be the proper way to go about studying an opening system.


Blue Devil Knight said...

There are a lot of good opening books out there, but it's true you really have to look. The best general principles book for beginners IMO is Basman's Chess Openings.

My favorite repertoire book is Greet's on the Ruy. It is unusual in that it doesn't start on move 9 and act like any deviations before that are just easy to handle with 'normal developing moves.' Bullshit. This is the Ruy Lopez, and there are lots of tricky deviations from black. Indeed, Greet's book doesn't do any of the annoying things that most opening books do. Opening book pet peeves from a previous post are:

1) Lazy-ass author squeezes out an annotated game dump and calls it a repertoire.
2) No explanation of the basic plans for each side coming out of the opening.
3) The book advertises itself as an opening about X, but in fact 90% of the book is about a subvariation of X after move ten, and the last chapter includes one or two games that are meant to cover all other "deviations".
4) Lazy-ass author constantly says "And this transposes to another line in the book" without telling you which line (it isn't always obvious, and the author could put in about six seconds of work and improve the book tremendously).
5) Unobjective, too optimistic about the repertoire.
6) No index of variations to be found in the book.
7) Dry, mechanical writing style: the book doesn't seem to have been written by an actual person.

I'm keeping my eye open a book on the Ruy from Black's perspective that doesn't commit most of these sins.

Blue Devil Knight said...

I have high hopes for the Marin book.

BlunderProne said...

I like the Everyman series of opening books if I ma learning a new opening. Lots of theory into the mainline. I tend to shy away from the repertoire books.

For depth, I look for GM games of my opneing on chessbase or some other database. I attempt to annotate them myself. Only after I have gone through completely, i will then follow it up with Fritz or someother resource.

This way, I not only see the piece play but I have a chance to internalize it on my own. The discovery process seems to be more meaningful.