Friday, March 2, 2007

What Can You Learn From Blitz Games?

I play too much blitz and not enough "regular chess". I think it's encouraging superficial thought, and I think I need to reprogram my brain to play what Heisman calls "real chess" on every move. I have a bad habit of going on blitz-crack binges where suddenly it's 3am, my vision is blurring and I'm down 300 ratings points.

However, I think that Blitz, in moderation, can help you improve your game if you stop and try to learn from each game. Here's how:

1. Blitz + Fritz = tactics problems for free. Well, you can use another analysis engine besides Fritz, but that wouldn't rhyme. If you analyze a blitz game with the computer, you are going to find plenty of examples of gross tactical blunders or missed opportunities. You can generate your own books of tactical problems by saving these positions in their own database. Chessbase 9 even allows you to add a "training annotation", where it will hide the next move and prompt you for the correct answer.

There are two advantages that a database of tactics taken from your own games has over store-bought books or software. One is that these are realistic or even mundane tactics. I've been enjoying using the CT-ART tactics program, but it's heavily slanted towards sacrifices and sacrificial mates. In a real game, the tactical opportunities that you miss are often more "boring". For example, maybe your queen could have forked a loose knight and pawn, and your opponent would have lost a pawn with no compensation. Heisman recommends studying lots of simple tactics to improve your pattern recognition, and these are exactly the kinds of errors that you'll be making in blitz games. One common error I see in my games is with "I-take, he-takes, I-take" counting problems, where a slight change in the move order makes a large difference.

I also see many instances where I played a good move, but there was a better move. For example, perhaps I could have won an extra pawn with another move, or been up a whole piece instead of just the exchange. You won't find many problems in books that deal with such mundane differences. The more often you choose the best move instead of the second-best move, the more quickly you can grind down your opponent.

The other advantage of such a book is that, since the positions are taken from your own games, they'll tend to stick in your mind better. Heisman recommends making a "Hall of Shame" database where you make problems out of your blunders. I'm going to start doing this more religiously, and also make some separate bases for 1, 2, and 3-move tactics (where I to move had the tactic) and blunders (where I allowed my opponent the tactic).

2. Opening Preparation. For every game you play, you should find out who left theory first, and what you would play if that position were to arise again. This can be as simple as seeing where someone played a move that's not in your computer program's opening book, or you could go as far as consulting opening manuals and/or searching a large database of games. If you use the program Bookup, you can automatically import your blitz games into opening databases for your white and black repertoires.

When I use Chessbase, I can see not only how often a move was played in my master games database, but what the player's ratings were and how well that move performed. This can be quite informative. For example, you may see that your opponent played a move that wasn't downright awful, but all the people that played it were rated below 2100, and in a couple of simuls a grandmaster opponent played what appears to be the refutation.

If you keep a database of your own games, you may find some interesting results that will save you study time. For example, when I searched for games in the last 7 years where I played the Spanish (Ruy Lopez) as White, I got these statistics:


This tells me a few interesting things:
a) the Steinitz defense (3...d6) occurred in a whopping 23% of my games. This is considered an outdated line by experts, and in some repertoire books gets very little attention.
b) The Cozio defence (3...Nge7) is causing me problems (opponent scoring 56%)
c) I'm doing well against the Berlin defense (3...Nf6; opponent scoring 23%).

If I succumb to the siren call of opening preparation, this tells me that I should concentrate on the Steinitz (which I face the most) and the Cozio (which I face often enough and do poorly against).

A similar search where I play the French as Black showed that I very rarely face the Tarrasch defence (3.Nd2) at my level...which is odd, since at professional level it's the #2 response. This tells me that it should be an excellent choice to play as white, because French players of my level rarely face it.

I'm going to maintain that you only need to know openings as deeply as your typical opponent. If you confine your preparation to finding out where your games deviate from theory, and what the "proper" continuations are, you can spend more time studying the stuff that really matters.